add gym (#1104)

* fix SepDataFrame when we del it to empty

* init_instance_by_config enhancement

* Update test_sepdf.py

.github/PULL_REQUEST_TEMPLATE.md

@@ -8,6 +8,7 @@
 <!--- Why is this change required? What problem does it solve? -->
 
 ## How Has This Been Tested?
+<!---  Put an `x` in all the boxes that apply: --->
 - [ ] Pass the test by running: `pytest qlib/tests/test_all_pipeline.py` under upper directory of `qlib`.
 - [ ] If you are adding a new feature, test on your own test scripts.
 

.github/workflows/python-publish.yml

@@ -12,7 +12,8 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        os: [windows-latest, macos-latest, macos-11]
+        os: [windows-latest, macos-11]
+        # FIXME:  macos-latest will raise error now.
         # not supporting 3.6 due to annotations is not supported https://stackoverflow.com/a/52890129
         python-version: [3.7, 3.8]
 

.github/workflows/test.yml

@@ -33,11 +33,85 @@ jobs:
     - name: Install Qlib with pip
       run: |
         pip install numpy==1.19.5 ruamel.yaml
-        pip install pyqlib --ignore-installed 
+        pip install pyqlib --ignore-installed
+
+    - name: Make html with sphinx
+      run: |
+        pip install -U sphinx
+        pip install sphinx_rtd_theme readthedocs_sphinx_ext
+        pip install --exists-action=w --no-cache-dir -r docs/requirements.txt 
+        cd docs 
+        sphinx-build -b html . build
+        cd ..
+        
+    # Check Qlib with pylint
+    # TODO: These problems we will solve in the future. Important among them are: W0221, W0223, W0237, E1102
+      # C0103: invalid-name
+      # C0209: consider-using-f-string
+      # R0402: consider-using-from-import
+      # R1705: no-else-return
+      # R1710: inconsistent-return-statements
+      # R1725: super-with-arguments
+      # R1735: use-dict-literal
+      # W0102: dangerous-default-value
+      # W0212: protected-access
+      # W0221: arguments-differ
+      # W0223: abstract-method
+      # W0231: super-init-not-called
+      # W0237: arguments-renamed
+      # W0612: unused-variable
+      # W0621: redefined-outer-name
+      # W0622: redefined-builtin
+      # FIXME: specify exception type
+      # W0703: broad-except
+      # W1309: f-string-without-interpolation
+      # E1102: not-callable
+      # E1136: unsubscriptable-object
+    # References for parameters: https://github.com/PyCQA/pylint/issues/4577#issuecomment-1000245962 
+    - name: Check Qlib with pylint
+      run: |
+        pip install --upgrade pip
+        pip install pylint
+        pylint --disable=C0104,C0114,C0115,C0116,C0301,C0302,C0411,C0413,C1802,R0201,R0401,R0801,R0902,R0903,R0911,R0912,R0913,R0914,R0915,R1720,W0105,W0123,W0201,W0511,W0613,W1113,W1514,E0401,E1121,C0103,C0209,R0402,R1705,R1710,R1725,R1735,W0102,W0212,W0221,W0223,W0231,W0237,W0612,W0621,W0622,W0703,W1309,E1102,E1136 --const-rgx='[a-z_][a-z0-9_]{2,30}$' qlib --init-hook "import astroid; astroid.context.InferenceContext.max_inferred = 500"
+
+    # The following flake8 error codes were ignored:
+      # E501 line too long
+        # Description: We have used black to limit the length of each line to 120.
+      # F541 f-string is missing placeholders
+        # Description: The same thing is done when using pylint for detection.
+      # E266 too many leading '#' for block comment
+        # Description: To make the code more readable, a lot of "#" is used.
+        # This error code appears centrally in:
+          # qlib/backtest/executor.py
+          # qlib/data/ops.py
+          # qlib/utils/__init__.py
+      # E402 module level import not at top of file
+        # Description: There are times when module level import is not available at the top of the file.
+      # W503 line break before binary operator
+        # Description: Since black formats the length of each line of code, it has to perform a line break when a line of arithmetic is too long.
+      # E731 do not assign a lambda expression, use a def
+        # Description: Restricts the use of lambda expressions, but at some point lambda expressions are required.
+      # E203 whitespace before ':'
+        # Description: If there is whitespace before ":", it cannot pass the black check.
+    - name: Check Qlib with flake8
+      run: |
+        pip install --upgrade pip
+        pip install flake8
+        flake8 --ignore=E501,F541,E266,E402,W503,E731,E203 --per-file-ignores="__init__.py:F401,F403" qlib
+
+    # https://github.com/python/mypy/issues/10600
+    - name: Check Qlib with mypy
+      run: |
+        pip install mypy
+        mypy qlib --install-types --non-interactive || true
+        mypy qlib
 
     - name: Test data downloads
       run: |
-        python scripts/get_data.py qlib_data --target_dir ~/.qlib/qlib_data/cn_data --interval 1d --region cn
+        python scripts/get_data.py qlib_data --name qlib_data_simple --target_dir ~/.qlib/qlib_data/cn_data_simple --interval 1d --region cn
+        python -c "import os; userpath=os.path.expanduser('~'); os.rename(userpath + '/.qlib/qlib_data/cn_data_simple', userpath + '/.qlib/qlib_data/cn_data')"
+        azcopy copy https://qlibpublic.blob.core.windows.net/data /tmp/qlibpublic --recursive
+        mv /tmp/qlibpublic/data tests/.data
 
     - name: Test workflow by config (install from pip)
       run: |
@@ -48,6 +122,7 @@ jobs:
     - name: Install Qlib from source
       run: |
         pip install --upgrade cython jupyter jupyter_contrib_nbextensions numpy scipy scikit-learn # installing without this line will cause errors on GitHub Actions, while instsalling locally won't
+        pip install gym tianshou torch
         pip install -e .
 
     - name: Install test dependencies
@@ -57,10 +132,10 @@ jobs:
 
     - name: Unit tests with Pytest
       run: |
+        pip install -r scripts/data_collector/pit/requirements.txt
         cd tests
         python -m pytest . --durations=10
 
     - name: Test workflow by config (install from source)
       run: |
         python qlib/workflow/cli.py examples/benchmarks/LightGBM/workflow_config_lightgbm_Alpha158.yaml
-      

.github/workflows/test_macos.yml

@@ -34,10 +34,24 @@ jobs:
         python -m black qlib -l 120 --check --diff
     # Test Qlib installed with pip
 
+    - name: Check Qlib with flake8
+      run: |
+        pip install --upgrade pip
+        pip install flake8
+        flake8 --ignore=E501,F541,E266,E402,W503,E731,E203 --per-file-ignores="__init__.py:F401,F403" qlib
+
     - name: Install Qlib with pip
       run: |
           python -m pip install numpy==1.19.5
           python -m pip install pyqlib --ignore-installed ruamel.yaml numpy
+    - name: Make html with sphnix
+      run: |
+        pip install -U sphinx
+        pip install sphinx_rtd_theme readthedocs_sphinx_ext
+        pip install --exists-action=w --no-cache-dir -r docs/requirements.txt 
+        cd docs 
+        sphinx-build -b html . build
+        cd ..
     - name: Install Lightgbm for MacOS
       run: |
         /bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Microsoft/qlib/main/.github/brew_install.sh)"
@@ -49,7 +63,10 @@ jobs:
         brew install libomp.rb
     - name: Test data downloads
       run: |
-        python scripts/get_data.py qlib_data --target_dir ~/.qlib/qlib_data/cn_data --interval 1d --region cn
+        python scripts/get_data.py qlib_data --name qlib_data_simple --target_dir ~/.qlib/qlib_data/cn_data_simple --interval 1d --region cn
+        python -c "import os; userpath=os.path.expanduser('~'); os.rename(userpath + '/.qlib/qlib_data/cn_data_simple', userpath + '/.qlib/qlib_data/cn_data')"
+        azcopy copy https://qlibpublic.blob.core.windows.net/data /tmp/qlibpublic --recursive
+        mv /tmp/qlibpublic/data tests/.data
     - name: Test workflow by config (install from pip)
       run: |
         python qlib/workflow/cli.py examples/benchmarks/LightGBM/workflow_config_lightgbm_Alpha158.yaml
@@ -60,7 +77,8 @@ jobs:
         python -m pip install --upgrade cython
         python -m pip install numpy jupyter jupyter_contrib_nbextensions
         python -m pip install -U scipy scikit-learn # installing without this line will cause errors on GitHub Actions, while instsalling locally won't
-        python setup.py install
+        python -m pip install gym tianshou torch
+        pip install -e .
     - name: Install test dependencies
       run: |
         python -m pip install --upgrade pip
@@ -68,6 +86,7 @@ jobs:
         python -m pip install black pytest
     - name: Unit tests with Pytest
       run: |
+        pip install -r scripts/data_collector/pit/requirements.txt
         cd tests
         python -m pytest . --durations=0
     - name: Test workflow by config (install from source)

.gitignore

@@ -27,6 +27,10 @@ examples/estimator/estimator_example/
 
 *.egg-info/
 
+# test related
+test-output.xml
+.output
+.data
 
 # special software
 mlruns/
@@ -34,6 +38,7 @@ mlruns/
 tags
 
 .pytest_cache/
+.mypy_cache/
 .vscode/
 
 *.swp

.mypy.ini

@@ -0,0 +1,17 @@
+[mypy]
+exclude = (?x)(
+    ^qlib/backtest
+    | ^qlib/contrib
+    | ^qlib/data
+    | ^qlib/model
+    | ^qlib/strategy
+    | ^qlib/tests
+    | ^qlib/utils
+    | ^qlib/workflow
+    | ^qlib/config\.py$
+    | ^qlib/log\.py$
+    | ^qlib/__init__\.py$
+  )
+ignore_missing_imports = true
+disallow_incomplete_defs = true
+follow_imports = skip

.pre-commit-config.yaml

@@ -0,0 +1,12 @@
+repos:
+-   repo: https://github.com/psf/black
+    rev: 22.1.0
+    hooks:
+    -   id: black
+        args: ["qlib", "-l 120"]
+
+-   repo: https://github.com/PyCQA/flake8
+    rev: 4.0.1
+    hooks:
+        - id: flake8
+          args: ["--ignore=E501,F541,E266,E402,W503,E731,E203"]

.pylintrc

@@ -0,0 +1,5 @@
+[TYPECHECK]
+# https://stackoverflow.com/a/53572939 
+# List of members which are set dynamically and missed by Pylint inference
+# system, and so shouldn't trigger E1101 when accessed.
+generated-members=numpy.*, torch.*

.readthedocs.yml

@@ -17,5 +17,5 @@ python:
   version: 3.7
   install:
     - requirements: docs/requirements.txt
-    - method: setuptools
-      path: .
+    - method: pip
+      path: .

CHANGES.rst

@@ -30,7 +30,7 @@ Version 0.2.1
 --------------------
 - Support registering user-defined ``Provider``.
 - Support use operators in string format, e.g. ``['Ref($close, 1)']`` is valid field format.
-- Support dynamic fields in ``$some_field`` format. And exising fields like ``Close()`` may be deprecated in the future.
+- Support dynamic fields in ``$some_field`` format. And existing fields like ``Close()`` may be deprecated in the future.
 
 Version 0.2.2
 --------------------
@@ -78,7 +78,7 @@ Version 0.3.5
 - Support multi-label training, you can provide multiple label in ``handler``. (But LightGBM doesn't support due to the algorithm itself)
 - Refactor ``handler`` code, dataset.py is no longer used, and you can deploy your own labels and features in ``feature_label_config``
 - Handler only offer DataFrame. Also, ``trainer`` and model.py only receive DataFrame
-- Change ``split_rolling_data``, we roll the data on market calender now, not on normal date
+- Change ``split_rolling_data``, we roll the data on market calendar now, not on normal date
 - Move some date config from ``handler`` to ``trainer``
 
 Version 0.4.0
@@ -167,11 +167,11 @@ Version 0.8.0
     - There are lots of changes for daily trading, it is hard to list all of them. But a few important changes could be noticed
         - The trading limitation is more accurate;
             - In `previous version <https://github.com/microsoft/qlib/blob/v0.7.2/qlib/contrib/backtest/exchange.py#L160>`_, longing and shorting actions share the same action.
-            - In `current verison <https://github.com/microsoft/qlib/blob/7c31012b507a3823117bddcc693fc64899460b2a/qlib/backtest/exchange.py#L304>`_, the trading limitation is different between loging and shorting action.
+            - In `current version <https://github.com/microsoft/qlib/blob/7c31012b507a3823117bddcc693fc64899460b2a/qlib/backtest/exchange.py#L304>`_, the trading limitation is different between logging and shorting action.
         - The constant is different when calculating annualized metrics.
             - `Current version <https://github.com/microsoft/qlib/blob/7c31012b507a3823117bddcc693fc64899460b2a/qlib/contrib/evaluate.py#L42>`_ uses more accurate constant than `previous version <https://github.com/microsoft/qlib/blob/v0.7.2/qlib/contrib/evaluate.py#L22>`_
         - `A new version <https://github.com/microsoft/qlib/blob/7c31012b507a3823117bddcc693fc64899460b2a/qlib/tests/data.py#L17>`_ of data is released. Due to the unstability of Yahoo data source, the data may be different after downloading data again.
-        - Users could chec kout the backtesting results between  `Current version <https://github.com/microsoft/qlib/tree/7c31012b507a3823117bddcc693fc64899460b2a/examples/benchmarks>`_ and `previous version <https://github.com/microsoft/qlib/tree/v0.7.2/examples/benchmarks>`_
+        - Users could check out the backtesting results between  `Current version <https://github.com/microsoft/qlib/tree/7c31012b507a3823117bddcc693fc64899460b2a/examples/benchmarks>`_ and `previous version <https://github.com/microsoft/qlib/tree/v0.7.2/examples/benchmarks>`_
 
 
 Other Versions

README.md

@@ -11,16 +11,28 @@
 Recent released features
 | Feature | Status |
 | --                      | ------    |
-|Temporal Routing Adaptor (TRA) | [Released](https://github.com/microsoft/qlib/pull/531) on July 30, 2021 |
-| Transformer & Localformer | [Released](https://github.com/microsoft/qlib/pull/508) on July 22, 2021 |
-| Release Qlib v0.7.0 | [Released](https://github.com/microsoft/qlib/releases/tag/v0.7.0) on July 12, 2021 |
-| TCTS Model | [Released](https://github.com/microsoft/qlib/pull/491) on July 1, 2021 |
-| Online serving and automatic model rolling | :star: [Released](https://github.com/microsoft/qlib/pull/290) on May 17, 2021 | 
-| DoubleEnsemble Model | [Released](https://github.com/microsoft/qlib/pull/286) on Mar 2, 2021 | 
-| High-frequency data processing example | [Released](https://github.com/microsoft/qlib/pull/257) on Feb 5, 2021  |
-| High-frequency trading example | [Part of code released](https://github.com/microsoft/qlib/pull/227) on Jan 28, 2021  | 
-| High-frequency data(1min) | [Released](https://github.com/microsoft/qlib/pull/221) on Jan 27, 2021 |
-| Tabnet Model | [Released](https://github.com/microsoft/qlib/pull/205) on Jan 22, 2021 | 
+| HIST and IGMTF models | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/1040) on Apr 10, 2022 |
+| Qlib [notebook tutorial](https://github.com/microsoft/qlib/tree/main/examples/tutorial) | 📖 [Released](https://github.com/microsoft/qlib/pull/1037) on Apr 7, 2022 | 
+| Ibovespa index data | :rice: [Released](https://github.com/microsoft/qlib/pull/990) on Apr 6, 2022 |
+| Point-in-Time database | :hammer: [Released](https://github.com/microsoft/qlib/pull/343) on Mar 10, 2022 |
+| Arctic Provider Backend & Orderbook data example | :hammer: [Released](https://github.com/microsoft/qlib/pull/744) on Jan 17, 2022 |
+| Meta-Learning-based framework & DDG-DA  | :chart_with_upwards_trend:  :hammer: [Released](https://github.com/microsoft/qlib/pull/743) on Jan 10, 2022 | 
+| Planning-based portfolio optimization | :hammer: [Released](https://github.com/microsoft/qlib/pull/754) on Dec 28, 2021 | 
+| Release Qlib v0.8.0 | :octocat: [Released](https://github.com/microsoft/qlib/releases/tag/v0.8.0) on Dec 8, 2021 |
+| ADD model | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/704) on Nov 22, 2021 |
+| ADARNN  model | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/689) on Nov 14, 2021 |
+| TCN  model | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/668) on Nov 4, 2021 |
+| Nested Decision Framework | :hammer: [Released](https://github.com/microsoft/qlib/pull/438) on Oct 1, 2021. [Example](https://github.com/microsoft/qlib/blob/main/examples/nested_decision_execution/workflow.py) and [Doc](https://qlib.readthedocs.io/en/latest/component/highfreq.html) |
+| Temporal Routing Adaptor (TRA) | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/531) on July 30, 2021 |
+| Transformer & Localformer | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/508) on July 22, 2021 |
+| Release Qlib v0.7.0 | :octocat: [Released](https://github.com/microsoft/qlib/releases/tag/v0.7.0) on July 12, 2021 |
+| TCTS Model | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/491) on July 1, 2021 |
+| Online serving and automatic model rolling | :hammer:  [Released](https://github.com/microsoft/qlib/pull/290) on May 17, 2021 | 
+| DoubleEnsemble Model | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/286) on Mar 2, 2021 | 
+| High-frequency data processing example | :hammer: [Released](https://github.com/microsoft/qlib/pull/257) on Feb 5, 2021  |
+| High-frequency trading example | :chart_with_upwards_trend: [Part of code released](https://github.com/microsoft/qlib/pull/227) on Jan 28, 2021  | 
+| High-frequency data(1min) | :rice: [Released](https://github.com/microsoft/qlib/pull/221) on Jan 27, 2021 |
+| Tabnet Model | :chart_with_upwards_trend: [Released](https://github.com/microsoft/qlib/pull/205) on Jan 22, 2021 |
 
 Features released before 2021 are not listed here.
 
@@ -37,35 +49,58 @@ With Qlib, users can easily try ideas to create better Quant investment strategi
 
 For more details, please refer to our paper ["Qlib: An AI-oriented Quantitative Investment Platform"](https://arxiv.org/abs/2009.11189).
 
-- [**Plans**](#plans)
-- [Framework of Qlib](#framework-of-qlib)
-- [Quick Start](#quick-start)
-  - [Installation](#installation)
-  - [Data Preparation](#data-preparation)
-  - [Auto Quant Research Workflow](#auto-quant-research-workflow)
-  - [Building Customized Quant Research Workflow by Code](#building-customized-quant-research-workflow-by-code)
-- [**Quant Model(Paper) Zoo**](#quant-model-paper-zoo)
-  - [Run a single model](#run-a-single-model)
-  - [Run multiple models](#run-multiple-models)
-- [**Quant Dataset Zoo**](#quant-dataset-zoo)
-- [More About Qlib](#more-about-qlib)
-- [Offline Mode and Online Mode](#offline-mode-and-online-mode)
-  - [Performance of Qlib Data Server](#performance-of-qlib-data-server)
-- [Related Reports](#related-reports)
-- [Contact Us](#contact-us)
-- [Contributing](#contributing)
 
+<table>
+  <tbody>
+    <tr>
+      <th>Frameworks, Tutorial, Data & DevOps</th>
+      <th>Main Challenges & Solutions in Quant Research</th>
+    </tr>
+    <tr>
+      <td>
+        <li><a href="#plans"><strong>Plans</strong></a></li>
+        <li><a href="#framework-of-qlib">Framework of Qlib</a></li>
+        <li><a href="#quick-start">Quick Start</a></li>
+          <ul dir="auto">
+            <li type="circle"><a href="#installation">Installation</a> </li>
+            <li type="circle"><a href="#data-preparation">Data Preparation</a></li>
+            <li type="circle"><a href="#auto-quant-research-workflow">Auto Quant Research Workflow</a></li>
+            <li type="circle"><a href="#building-customized-quant-research-workflow-by-code">Building Customized Quant Research Workflow by Code</a></li></ul>
+        <li><a href="#quant-dataset-zoo"><strong>Quant Dataset Zoo</strong></a></li>
+        <li><a href="#more-about-qlib">More About Qlib</a></li>
+        <li><a href="#offline-mode-and-online-mode">Offline Mode and Online Mode</a>
+        <ul>
+          <li type="circle"><a href="#performance-of-qlib-data-server">Performance of Qlib Data Server</a></li></ul>
+        <li><a href="#related-reports">Related Reports</a></li>
+        <li><a href="#contact-us">Contact Us</a></li>
+        <li><a href="#contributing">Contributing</a></li>
+      </td>
+      <td valign="baseline">
+        <li><a href="#main-challenges--solutions-in-quant-research">Main Challenges &amp; Solutions in Quant Research</a>
+          <ul>
+            <li type="circle"><a href="#forecasting-finding-valuable-signalspatterns">Forecasting: Finding Valuable Signals/Patterns</a>
+              <ul>
+                <li type="disc"><a href="#quant-model-paper-zoo"><strong>Quant Model (Paper) Zoo</strong></a>
+                  <ul>
+                    <li type="circle"><a href="#run-a-single-model">Run a Single Model</a></li>
+                    <li type="circle"><a href="#run-multiple-models">Run Multiple Models</a></li>
+                  </ul>
+                </li>
+              </ul>
+            </li>
+          <li type="circle"><a href="#adapting-to-market-dynamics">Adapting to Market Dynamics</a></li>
+          </ul>
+        </li>
+      </td>
+    </tr>
+  </tbody>
+</table>
 
 # Plans
 New features under development(order by estimated release time).
 Your feedbacks about the features are very important.
-| Feature                        | Status      |
-| --                      | ------    |
-| Planning-based portfolio optimization | Under review:  https://github.com/microsoft/qlib/pull/280 | 
-| Fund data supporting and analysis  |  Under review: https://github.com/microsoft/qlib/pull/292 |
-| Point-in-Time database | Under review: https://github.com/microsoft/qlib/pull/343 |
-| High-frequency trading | Under review: https://github.com/microsoft/qlib/pull/408 | 
-| Meta-Learning-based data selection | Initial opensource version under development |
+<!-- | Feature                        | Status      | -->
+<!-- | --                      | ------    | -->
 
 # Framework of Qlib
 
@@ -73,18 +108,19 @@ Your feedbacks about the features are very important.
 <img src="docs/_static/img/framework.svg" />
 </div>
 
-
 At the module level, Qlib is a platform that consists of the above components. The components are designed as loose-coupled modules, and each component could be used stand-alone.
 
 | Name                   | Description                                                                                                                                                                                                                                                                                                                                                             |
 | ------                 | -----                                                                                                                                                                                                                                                                                                                                                                   |
 | `Infrastructure` layer | `Infrastructure` layer provides underlying support for Quant research. `DataServer` provides a high-performance infrastructure for users to manage and retrieve raw data. `Trainer` provides a flexible interface to control the training process of models, which enable algorithms to control the training process.                                                       |
-| `Workflow` layer       | `Workflow` layer covers the whole workflow of quantitative investment. `Information Extractor` extracts data for models. `Forecast Model` focuses on producing all kinds of forecast signals (e.g. _alpha_, risk) for other modules. With these signals `Portfolio Generator` will generate the target portfolio and produce orders to be executed by `Order Executor`. |
+| `Workflow` layer       | `Workflow` layer covers the whole workflow of quantitative investment. `Information Extractor` extracts data for models. `Forecast Model` focuses on producing all kinds of forecast signals (e.g. _alpha_, risk) for other modules. With these signals `Decision Generator` will generate the target trading decisions(i.e. portfolio, orders)  to be executed by `Execution Env` (i.e. the trading market).  There may be multiple levels of `Trading Agent` and `Execution Env` (e.g. an _order executor trading agent and intraday order execution environment_ could behave like an interday trading environment and nested in  _daily portfolio management trading agent and interday trading environment_  ) |
 | `Interface` layer      | `Interface` layer tries to present a user-friendly interface for the underlying system. `Analyser` module will provide users detailed analysis reports of forecasting signals, portfolios and execution results                                                                                                                                                                 |
 
 * The modules with hand-drawn style are under development and will be released in the future.
 * The modules with dashed borders are highly user-customizable and extendible.
 
+(p.s. framework image is created with https://draw.io/)
+
 
 # Quick Start
 
@@ -108,6 +144,7 @@ This table demonstrates the supported Python version of `Qlib`:
 1. **Conda** is suggested for managing your Python environment.
 1. Please pay attention that installing cython in Python 3.6 will raise some error when installing ``Qlib`` from source. If users use Python 3.6 on their machines, it is recommended to *upgrade* Python to version 3.7 or use `conda`'s Python to install ``Qlib`` from source.
 1. For Python 3.9, `Qlib` supports running workflows such as training models, doing backtest and plot most of the related figures (those included in [notebook](examples/workflow_by_code.ipynb)). However, plotting for the *model performance* is not supported for now and we will fix this when the dependent packages are upgraded in the future.
+1. `Qlib`Requires `tables` package, `hdf5` in tables does not support python3.9. 
 
 ### Install with pip
 Users can easily install ``Qlib`` by pip according to the following command.
@@ -129,17 +166,11 @@ Also, users can install the latest dev version ``Qlib`` by the source code accor
   ```
 
 * Clone the repository and install ``Qlib`` as follows.
-  * If you haven't installed qlib by the command ``pip install pyqlib`` before:
-    ```bash
-    git clone https://github.com/microsoft/qlib.git && cd qlib
-    python setup.py install
-    ```
-  * If you have already installed the stable version by the command ``pip install pyqlib``:
     ```bash
     git clone https://github.com/microsoft/qlib.git && cd qlib
     pip install .
     ```
-  **Note**: **Only** the command ``pip install .`` **can** overwrite the stable version installed by ``pip install pyqlib``, while the command ``python setup.py install`` **can't**.
+  **Note**:  You can install Qlib with `python setup.py install` as well. But it is not the recommanded approach. It will skip `pip` and cause obscure problems. For example, **only** the command ``pip install .`` **can** overwrite the stable version installed by ``pip install pyqlib``, while the command ``python setup.py install`` **can't**.
 
 **Tips**: If you fail to install `Qlib` or run the examples in your environment,  comparing your steps and the [CI workflow](.github/workflows/test.yml) may help you find the problem.
 
@@ -156,15 +187,17 @@ Load and prepare data by running the following code:
 
 This dataset is created by public data collected by [crawler scripts](scripts/data_collector/), which have been released in
 the same repository.
-Users could create the same dataset with it. 
+Users could create the same dataset with it. [Description of dataset](https://github.com/microsoft/qlib/tree/main/scripts/data_collector#description-of-dataset)
 
 *Please pay **ATTENTION** that the data is collected from [Yahoo Finance](https://finance.yahoo.com/lookup), and the data might not be perfect.
 We recommend users to prepare their own data if they have a high-quality dataset. For more information, users can refer to the [related document](https://qlib.readthedocs.io/en/latest/component/data.html#converting-csv-format-into-qlib-format)*.
 
 ### Automatic update of daily frequency data (from yahoo finance)
+  > This step is *Optional* if users only want to try their models and strategies on history data.
+  > 
   > It is recommended that users update the data manually once (--trading_date 2021-05-25) and then set it to update automatically.
-
-  > For more information refer to: [yahoo collector](https://github.com/microsoft/qlib/tree/main/scripts/data_collector/yahoo#automatic-update-of-daily-frequency-datafrom-yahoo-finance)
+  >
+  > For more information, please refer to: [yahoo collector](https://github.com/microsoft/qlib/tree/main/scripts/data_collector/yahoo#automatic-update-of-daily-frequency-datafrom-yahoo-finance)
 
   * Automatic update of data to the "qlib" directory each trading day(Linux)
       * use *crontab*: `crontab -e`
@@ -189,7 +222,7 @@ We recommend users to prepare their own data if they have a high-quality dataset
   ```python
   import qlib
   from qlib.data import D
-  from qlib.config import REG_CN
+  from qlib.constant import REG_CN
 
   # Initialization
   mount_path = "~/.qlib/qlib_data/cn_data"  # target_dir
@@ -274,32 +307,47 @@ Qlib provides a tool named `qrun` to run the whole workflow automatically (inclu
 ## Building Customized Quant Research Workflow by Code
 The automatic workflow may not suit the research workflow of all Quant researchers. To support a flexible Quant research workflow, Qlib also provides a modularized interface to allow researchers to build their own workflow by code. [Here](examples/workflow_by_code.ipynb) is a demo for customized Quant research workflow by code.
 
+# Main Challenges & Solutions in Quant Research
+Quant investment is an very unique scenario with lots of key challenges to be solved.
+Currently, Qlib provides some solutions for several of them.
 
-# [Quant Model (Paper) Zoo](examples/benchmarks)
+## Forecasting: Finding Valuable Signals/Patterns
+Accurate forecasting of the stock price trend is a very important part to construct profitable portfolios.
+However, huge amount of data with various formats in the financial market which make it challenging to build forecasting models.
+
+An increasing number of SOTA Quant research works/papers, which focus on building forecasting models to mine valuable signals/patterns in complex financial data, are released in `Qlib`
+
+
+### [Quant Model (Paper) Zoo](examples/benchmarks)
 
 Here is a list of models built on `Qlib`.
-- [GBDT based on XGBoost (Tianqi Chen, et al. KDD 2016)](qlib/contrib/model/xgboost.py)
-- [GBDT based on LightGBM (Guolin Ke, et al. NIPS 2017)](qlib/contrib/model/gbdt.py)
-- [GBDT based on Catboost (Liudmila Prokhorenkova, et al. NIPS 2018)](qlib/contrib/model/catboost_model.py)
-- [MLP based on pytorch](qlib/contrib/model/pytorch_nn.py)
-- [LSTM based on pytorch (Sepp Hochreiter, et al. Neural omputation 1997)](qlib/contrib/model/pytorch_lstm.py)
-- [GRU based on pytorch (Kyunghyun Cho, et al. 2014)](qlib/contrib/model/pytorch_gru.py)
-- [ALSTM based on pytorch (Yao Qin, et al. IJCAI 2017)](qlib/contrib/model/pytorch_alstm.py)
-- [GATs based on pytorch (Petar Velickovic, et al. 2017)](qlib/contrib/model/pytorch_gats.py)
-- [SFM based on pytorch (Liheng Zhang, et al. KDD 2017)](qlib/contrib/model/pytorch_sfm.py)
-- [TFT based on tensorflow (Bryan Lim, et al. International Journal of Forecasting 2019)](examples/benchmarks/TFT/tft.py)
-- [TabNet based on pytorch (Sercan O. Arik, et al. AAAI 2019)](qlib/contrib/model/pytorch_tabnet.py)
-- [DoubleEnsemble based on LightGBM (Chuheng Zhang, et al. ICDM 2020)](qlib/contrib/model/double_ensemble.py)
-- [TCTS based on pytorch (Xueqing Wu, et al. ICML 2021)](qlib/contrib/model/pytorch_tcts.py)
-- [Transformer based on pytorch (Ashish Vaswani, et al. NeurIPS 2017)](qlib/contrib/model/pytorch_transformer.py)
-- [Localformer based on pytorch (Juyong Jiang, et al.)](qlib/contrib/model/pytorch_localformer.py)
-- [TRA based on pytorch (Hengxu, Dong, et al. KDD 2021)](qlib/contrib/model/pytorch_tra.py)
+- [GBDT based on XGBoost (Tianqi Chen, et al. KDD 2016)](examples/benchmarks/XGBoost/)
+- [GBDT based on LightGBM (Guolin Ke, et al. NIPS 2017)](examples/benchmarks/LightGBM/)
+- [GBDT based on Catboost (Liudmila Prokhorenkova, et al. NIPS 2018)](examples/benchmarks/CatBoost/)
+- [MLP based on pytorch](examples/benchmarks/MLP/)
+- [LSTM based on pytorch (Sepp Hochreiter, et al. Neural computation 1997)](examples/benchmarks/LSTM/)
+- [GRU based on pytorch (Kyunghyun Cho, et al. 2014)](examples/benchmarks/GRU/)
+- [ALSTM based on pytorch (Yao Qin, et al. IJCAI 2017)](examples/benchmarks/ALSTM)
+- [GATs based on pytorch (Petar Velickovic, et al. 2017)](examples/benchmarks/GATs/)
+- [SFM based on pytorch (Liheng Zhang, et al. KDD 2017)](examples/benchmarks/SFM/)
+- [TFT based on tensorflow (Bryan Lim, et al. International Journal of Forecasting 2019)](examples/benchmarks/TFT/)
+- [TabNet based on pytorch (Sercan O. Arik, et al. AAAI 2019)](examples/benchmarks/TabNet/)
+- [DoubleEnsemble based on LightGBM (Chuheng Zhang, et al. ICDM 2020)](examples/benchmarks/DoubleEnsemble/)
+- [TCTS based on pytorch (Xueqing Wu, et al. ICML 2021)](examples/benchmarks/TCTS/)
+- [Transformer based on pytorch (Ashish Vaswani, et al. NeurIPS 2017)](examples/benchmarks/Transformer/)
+- [Localformer based on pytorch (Juyong Jiang, et al.)](examples/benchmarks/Localformer/)
+- [TRA based on pytorch (Hengxu, Dong, et al. KDD 2021)](examples/benchmarks/TRA/)
+- [TCN based on pytorch (Shaojie Bai, et al. 2018)](examples/benchmarks/TCN/)
+- [ADARNN based on pytorch (YunTao Du, et al. 2021)](examples/benchmarks/ADARNN/)
+- [ADD based on pytorch (Hongshun Tang, et al.2020)](examples/benchmarks/ADD/)
+- [IGMTF based on pytorch (Wentao Xu, et al.2021)](examples/benchmarks/IGMTF/)
+- [HIST based on pytorch (Wentao Xu, et al.2021)](examples/benchmarks/HIST/)
 
 Your PR of new Quant models is highly welcomed.
 
 The performance of each model on the `Alpha158` and `Alpha360` dataset can be found [here](examples/benchmarks/README.md).
 
-## Run a single model
+### Run a single model
 All the models listed above are runnable with ``Qlib``. Users can find the config files we provide and some details about the model through the [benchmarks](examples/benchmarks) folder. More information can be retrieved at the model files listed above.
 
 `Qlib` provides three different ways to run a single model, users can pick the one that fits their cases best:
@@ -309,7 +357,7 @@ All the models listed above are runnable with ``Qlib``. Users can find the confi
 - Users can use the script [`run_all_model.py`](examples/run_all_model.py) listed in the `examples` folder to run a model. Here is an example of the specific shell command to be used: `python run_all_model.py run --models=lightgbm`, where the `--models` arguments can take any number of models listed above(the available models can be found  in [benchmarks](examples/benchmarks/)). For more use cases, please refer to the file's [docstrings](examples/run_all_model.py).
     - **NOTE**: Each baseline has different environment dependencies, please make sure that your python version aligns with the requirements(e.g. TFT only supports Python 3.6~3.7 due to the limitation of `tensorflow==1.15.0`)
 
-## Run multiple models
+### Run multiple models
 `Qlib` also provides a script [`run_all_model.py`](examples/run_all_model.py) which can run multiple models for several iterations. (**Note**: the script only support *Linux* for now. Other OS will be supported in the future. Besides, it doesn't support parallel running the same model for multiple times as well, and this will be fixed in the future development too.)
 
 The script will create a unique virtual environment for each model, and delete the environments after training. Thus, only experiment results such as `IC` and `backtest` results will be generated and stored.
@@ -321,6 +369,14 @@ python run_all_model.py run 10
 
 It also provides the API to run specific models at once. For more use cases, please refer to the file's [docstrings](examples/run_all_model.py). 
 
+## [Adapting to Market Dynamics](examples/benchmarks_dynamic)
+
+Due to the non-stationary nature of the environment of the financial market, the data distribution may change in different periods, which makes the performance of models build on training data decays in the future test data.
+So adapting the forecasting models/strategies to market dynamics is very important to the model/strategies' performance.
+
+Here is a list of solutions built on `Qlib`.
+- [Rolling Retraining](examples/benchmarks_dynamic/baseline/)
+- [DDG-DA on pytorch (Wendi, et al. AAAI 2022)](examples/benchmarks_dynamic/DDG-DA/)
 
 # Quant Dataset Zoo
 Dataset plays a very important role in Quant. Here is a list of the datasets built on `Qlib`:
@@ -334,6 +390,8 @@ Dataset plays a very important role in Quant. Here is a list of the datasets bui
 Your PR to build new Quant dataset is highly welcomed.
 
 # More About Qlib
+If you want to have a quick glance at the most frequently used components of qlib, you can try notebooks [here](examples/tutorial/).
+
 The detailed documents are organized in [docs](docs/).
 [Sphinx](http://www.sphinx-doc.org) and the readthedocs theme is required to build the documentation in html formats. 
 ```bash
@@ -391,17 +449,40 @@ Join IM discussion groups:
 |![image](http://fintech.msra.cn/images_v070/qrcode/gitter_qr.png)|
 
 # Contributing
+We appreciate all contributions and thank all the contributors!
+<a href="https://github.com/microsoft/qlib/graphs/contributors"><img src="https://contrib.rocks/image?repo=microsoft/qlib" /></a>
+
+Before we released Qlib as an open-source project on Github in Sep 2020, Qlib is an internal project in our group. Unfortunately, the internal commit history is not kept. A lot of members in our group have also contributed a lot to Qlib, which includes Ruihua Wang, Yinda Zhang, Haisu Yu, Shuyu Wang, Bochen Pang, and [Dong Zhou](https://github.com/evanzd/evanzd). Especially thanks to [Dong Zhou](https://github.com/evanzd/evanzd) due to his initial version of Qlib.
+
+## Guidance
 
 This project welcomes contributions and suggestions.  
 **Here are some 
-[code standards](docs/developer/code_standard.rst) when you submit a pull request.**
+[code standards](docs/developer/code_standard.rst) for submiting a pull request.**
 
-If you want to contribute to Qlib's document, you can follow the steps in the figure below.
+Making contributions is not a hard thing. Solving an issue(maybe just answering a question raised in [issues list](https://github.com/microsoft/qlib/issues) or [gitter](https://gitter.im/Microsoft/qlib)), fixing/issuing a bug, improving the documents and even fixing a typo are important contributions to Qlib.
+
+For example, if you want to contribute to Qlib's document/code, you can follow the steps in the figure below.
 <p align="center">
   <img src="https://github.com/demon143/qlib/blob/main/docs/_static/img/change%20doc.gif" />
 </p>
 
+If you don't know how to start to contribute, you can refer to the following examples.
+| Type | Examples |
+| -- | -- |
+| Solving issues | [Answer a question](https://github.com/microsoft/qlib/issues/749);  [issuing](https://github.com/microsoft/qlib/issues/765) or [fixing](https://github.com/microsoft/qlib/pull/792) a bug |
+| Docs | [Improve docs quality](https://github.com/microsoft/qlib/pull/797/files) ;  [Fix a typo](https://github.com/microsoft/qlib/pull/774) | 
+| Feature |  Implement a [requested feature](https://github.com/microsoft/qlib/projects) like [this](https://github.com/microsoft/qlib/pull/754); [Refactor interfaces](https://github.com/microsoft/qlib/pull/539/files) |
+| Dataset | [Add a dataset](https://github.com/microsoft/qlib/pull/733) | 
+| Models |  [Implement a new model](https://github.com/microsoft/qlib/pull/689), [some instructions to contribute models](https://github.com/microsoft/qlib/tree/main/examples/benchmarks#contributing) |
 
+[Good first issues](https://github.com/microsoft/qlib/labels/good%20first%20issue) are labelled to indicate that they are easy to start your contributions.
+
+You can find some impefect implementation in Qlib by  `rg 'TODO|FIXME' qlib`
+ 
+If you would like to become one of Qlib's maintainers to contribute more (e.g. help merge PR, triage issues), please contact us by email([qlib@microsoft.com](mailto:qlib@microsoft.com)).  We are glad to help to upgrade your permission.
+
+## Licence
 Most contributions require you to agree to a
 Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us
 the right to use your contribution. For details, visit https://cla.opensource.microsoft.com.

VERSION.txt

@@ -1 +0,0 @@
-0.7.2.99

docs/advanced/PIT.rst

@@ -0,0 +1,136 @@
+.. _pit:
+
+===========================
+(P)oint-(I)n-(T)ime Database
+===========================
+.. currentmodule:: qlib
+
+
+Introduction
+------------
+Point-in-time data is a very important consideration when performing any sort of historical market analysis. 
+
+For example, let’s say we are backtesting a trading strategy and we are using the past five years of historical data as our input.
+Our model is assumed to trade once a day, at the market close, and we’ll say we are calculating the trading signal for 1 January 2020 in our backtest. At that point, we should only have data for 1 January 2020, 31 December 2019, 30 December 2019 etc.
+
+In financial data (especially financial reports), the same piece of data may be amended for multiple times overtime.  If we only use the latest version for historical backtesting, data leakage will happen.
+Point-in-time database is designed for solving this problem to make sure user get the right version of data at any historical timestamp. It will keep the performance of online trading and historical backtesting the same.
+
+
+
+Data Preparation
+----------------
+
+Qlib provides a crawler to help users to download financial data and then a converter to dump the data in Qlib format.
+Please follow `scripts/data_collector/pit/README.md <https://github.com/microsoft/qlib/tree/main/scripts/data_collector/pit/>`_ to download and convert data.
+Besides, you can find some additional usage examples there.
+
+
+File-based design for PIT data
+------------------------------
+
+Qlib provides a file-based storage for PIT data.
+
+For each feature, it contains 4 columns, i.e. date, period, value, _next.
+Each row corresponds to a statement.
+
+The meaning of each feature with filename like `XXX_a.data`:
+
+- `date`: the statement's date of publication.
+- `period`: the period of the statement. (e.g. it will be quarterly frequency in most of the markets)
+    - If it is an annual period, it will be an integer corresponding to the year
+    - If it is an quarterly  periods, it will be an integer like `<year><index of quarter>`.  The last two decimal digits represents the index of quarter. Others represent the year.
+- `value`: the described value
+- `_next`: the byte index of the next occurance of the field.
+
+Besides the feature data, an index `XXX_a.index` is included to speed up the querying performance
+
+The statements are soted by the `date` in ascending order from the beginning of the file.
+
+.. code-block:: python
+
+    # the data format from XXXX.data
+    array([(20070428, 200701, 0.090219  , 4294967295),
+           (20070817, 200702, 0.13933   , 4294967295),
+           (20071023, 200703, 0.24586301, 4294967295),
+           (20080301, 200704, 0.3479    ,         80),
+           (20080313, 200704, 0.395989  , 4294967295),
+           (20080422, 200801, 0.100724  , 4294967295),
+           (20080828, 200802, 0.24996801, 4294967295),
+           (20081027, 200803, 0.33412001, 4294967295),
+           (20090325, 200804, 0.39011699, 4294967295),
+           (20090421, 200901, 0.102675  , 4294967295),
+           (20090807, 200902, 0.230712  , 4294967295),
+           (20091024, 200903, 0.30072999, 4294967295),
+           (20100402, 200904, 0.33546099, 4294967295),
+           (20100426, 201001, 0.083825  , 4294967295),
+           (20100812, 201002, 0.200545  , 4294967295),
+           (20101029, 201003, 0.260986  , 4294967295),
+           (20110321, 201004, 0.30739301, 4294967295),
+           (20110423, 201101, 0.097411  , 4294967295),
+           (20110831, 201102, 0.24825101, 4294967295),
+           (20111018, 201103, 0.318919  , 4294967295),
+           (20120323, 201104, 0.4039    ,        420),
+           (20120411, 201104, 0.403925  , 4294967295),
+           (20120426, 201201, 0.112148  , 4294967295),
+           (20120810, 201202, 0.26484701, 4294967295),
+           (20121026, 201203, 0.370487  , 4294967295),
+           (20130329, 201204, 0.45004699, 4294967295),
+           (20130418, 201301, 0.099958  , 4294967295),
+           (20130831, 201302, 0.21044201, 4294967295),
+           (20131016, 201303, 0.30454299, 4294967295),
+           (20140325, 201304, 0.394328  , 4294967295),
+           (20140425, 201401, 0.083217  , 4294967295),
+           (20140829, 201402, 0.16450299, 4294967295),
+           (20141030, 201403, 0.23408499, 4294967295),
+           (20150421, 201404, 0.319612  , 4294967295),
+           (20150421, 201501, 0.078494  , 4294967295),
+           (20150828, 201502, 0.137504  , 4294967295),
+           (20151023, 201503, 0.201709  , 4294967295),
+           (20160324, 201504, 0.26420501, 4294967295),
+           (20160421, 201601, 0.073664  , 4294967295),
+           (20160827, 201602, 0.136576  , 4294967295),
+           (20161029, 201603, 0.188062  , 4294967295),
+           (20170415, 201604, 0.244385  , 4294967295),
+           (20170425, 201701, 0.080614  , 4294967295),
+           (20170728, 201702, 0.15151   , 4294967295),
+           (20171026, 201703, 0.25416601, 4294967295),
+           (20180328, 201704, 0.32954201, 4294967295),
+           (20180428, 201801, 0.088887  , 4294967295),
+           (20180802, 201802, 0.170563  , 4294967295),
+           (20181029, 201803, 0.25522   , 4294967295),
+           (20190329, 201804, 0.34464401, 4294967295),
+           (20190425, 201901, 0.094737  , 4294967295),
+           (20190713, 201902, 0.        ,       1040),
+           (20190718, 201902, 0.175322  , 4294967295),
+           (20191016, 201903, 0.25581899, 4294967295)],
+          dtype=[('date', '<u4'), ('period', '<u4'), ('value', '<f8'), ('_next', '<u4')])
+    # - each row contains 20 byte
+
+
+    # The data format from XXXX.index.  It consists of two parts
+    # 1) the start index of the data. So the first part of the info will be like
+    2007
+    # 2) the remain index data will be like information below
+    #    - The data indicate the **byte index** of first data update of a period.
+    #    - e.g. Because the info at both byte 80 and 100 corresponds to 200704. The byte index of first occurance (i.e. 100) is recorded in the data.
+    array([         0,         20,         40,         60,        100,
+                  120,        140,        160,        180,        200,
+                  220,        240,        260,        280,        300,
+                  320,        340,        360,        380,        400,
+                  440,        460,        480,        500,        520,
+                  540,        560,        580,        600,        620,
+                  640,        660,        680,        700,        720,
+                  740,        760,        780,        800,        820,
+                  840,        860,        880,        900,        920,
+                  940,        960,        980,       1000,       1020,
+                 1060, 4294967295], dtype=uint32)
+
+
+
+
+Known limitations:
+
+- Currently, the PIT database is designed for quarterly or annually factors, which can handle fundamental data of financial reports in most markets.
+- Qlib leverage the file name to identify the type of the data. File with name like `XXX_q.data` corresponds to quarterly data. File with name like `XXX_a.data` corresponds to annual data.
+- The caclulation of PIT is not performed in the optimal way. There is great potential to boost the performance of PIT data calcuation.

docs/advanced/task_management.rst

@@ -11,7 +11,10 @@ Introduction
 
 The `Workflow <../component/introduction.html>`_ part introduces how to run research workflow in a loosely-coupled way. But it can only execute one ``task`` when you use ``qrun``.
 To automatically generate and execute different tasks, ``Task Management`` provides a whole process including `Task Generating`_, `Task Storing`_, `Task Training`_ and `Task Collecting`_. 
-With this module, users can run their ``task`` automatically at different periods, in different losses, or even by different models.
+With this module, users can run their ``task`` automatically at different periods, in different losses, or even by different models.The processes of task generation, model training and combine and collect data are shown in the following figure.
+
+.. image:: ../_static/img/Task-Gen-Recorder-Collector.svg
+    :align: center
 
 This whole process can be used in `Online Serving <../component/online.html>`_.
 
@@ -74,6 +77,8 @@ If you do not want to use ``Task Manager`` to manage tasks, then use TrainerR to
 
 Task Collecting
 ===============
+Before collecting model training results, you need to use the ``qlib.init`` to specify the path of mlruns.
+
 To collect the results of ``task`` after training, ``Qlib`` provides `Collector <../reference/api.html#Collector>`_, `Group <../reference/api.html#Group>`_ and `Ensemble <../reference/api.html#Ensemble>`_ to collect the results in a readable, expandable and loosely-coupled way.
 
 `Collector <../reference/api.html#Collector>`_ can collect objects from everywhere and process them such as merging, grouping, averaging and so on. It has 2 step action including ``collect`` (collect anything in a dict) and ``process_collect`` (process collected dict).
@@ -82,8 +87,10 @@ To collect the results of ``task`` after training, ``Qlib`` provides `Collector
 For example: {(A,B,C1): object, (A,B,C2): object} ---``group``---> {(A,B): {C1: object, C2: object}} ---``reduce``---> {(A,B): object}
 
 `Ensemble <../reference/api.html#Ensemble>`_ can merge the objects in an ensemble. 
-For example: {C1: object, C2: object} ---``Ensemble``---> object
+For example: {C1: object, C2: object} ---``Ensemble``---> object.
+You can set the ensembles you want in the ``Collector``'s process_list.
+Common ensembles include ``AverageEnsemble`` and ``RollingEnsemble``. Average ensemble is used to ensemble the results of different models in the same time period. Rollingensemble is used to ensemble the results of different models in the same time period
 
 So the hierarchy is ``Collector``'s second step corresponds to ``Group``. And ``Group``'s second step correspond to ``Ensemble``.
 
-For more information, please see `Collector <../reference/api.html#Collector>`_, `Group <../reference/api.html#Group>`_ and `Ensemble <../reference/api.html#Ensemble>`_, or the `example <https://github.com/microsoft/qlib/tree/main/examples/model_rolling/task_manager_rolling.py>`_.
+For more information, please see `Collector <../reference/api.html#Collector>`_, `Group <../reference/api.html#Group>`_ and `Ensemble <../reference/api.html#Ensemble>`_, or the `example <https://github.com/microsoft/qlib/tree/main/examples/model_rolling/task_manager_rolling.py>`_.

docs/component/backtest.rst

@@ -1,114 +0,0 @@
-.. _backtest:
-
-============================================
-Intraday Trading: Model&Strategy Testing
-============================================
-.. currentmodule:: qlib
-
-Introduction
-===================
-
-``Intraday Trading`` is designed to test models and strategies, which help users to check the performance of a custom model/strategy.
-
-
-.. note::
-
-    ``Intraday Trading`` uses ``Order Executor`` to trade and execute orders output by ``Portfolio Strategy``. ``Order Executor`` is a component in `Qlib Framework <../introduction/introduction.html#framework>`_, which can execute orders. ``VWAP Executor`` and ``Close Executor`` is supported by ``Qlib`` now. In the future, ``Qlib`` will support ``HighFreq Executor`` also. 
-
-
-
-Example
-===========================
-
-Users need to generate a `prediction score`(a pandas DataFrame) with MultiIndex<instrument, datetime> and a `score` column. And users need to assign a strategy used in backtest, if strategy is not assigned,
-a `TopkDropoutStrategy` strategy with `(topk=50, n_drop=5, risk_degree=0.95, limit_threshold=0.0095)` will be used.
-If ``Strategy`` module is not users' interested part, `TopkDropoutStrategy` is enough. 
-
-The simple example of the default strategy is as follows.
-
-.. code-block:: python
-
-    from qlib.contrib.evaluate import backtest
-    # pred_score is the prediction score
-    report, positions = backtest(pred_score, topk=50, n_drop=0.5, limit_threshold=0.0095)
-
-To know more about backtesting with a specific ``Strategy``, please refer to `Portfolio Strategy <strategy.html>`_.
-
-To know more about the prediction score `pred_score` output by ``Forecast Model``, please refer to `Forecast Model: Model Training & Prediction <model.html>`_.
-
-Prediction Score
------------------
-
-The `prediction score` is a pandas DataFrame. Its index is <datetime(pd.Timestamp), instrument(str)> and it must
-contains a `score` column.
-
-A prediction sample is shown as follows.
-
-.. code-block:: python
-
-      datetime instrument     score
-    2019-01-04   SH600000 -0.505488
-    2019-01-04   SZ002531 -0.320391
-    2019-01-04   SZ000999  0.583808
-    2019-01-04   SZ300569  0.819628
-    2019-01-04   SZ001696 -0.137140
-                 ...            ...
-    2019-04-30   SZ000996 -1.027618
-    2019-04-30   SH603127  0.225677
-    2019-04-30   SH603126  0.462443
-    2019-04-30   SH603133 -0.302460
-    2019-04-30   SZ300760 -0.126383
-
-``Forecast Model`` module can make predictions, please refer to `Forecast Model: Model Training & Prediction <model.html>`_.
-
-Backtest Result
-------------------
-
-The backtest results are in the following form:
-
-.. code-block:: python
-
-                                                      risk
-    excess_return_without_cost mean               0.000605
-                               std                0.005481
-                               annualized_return  0.152373
-                               information_ratio  1.751319
-                               max_drawdown      -0.059055
-    excess_return_with_cost    mean               0.000410
-                               std                0.005478
-                               annualized_return  0.103265
-                               information_ratio  1.187411
-                               max_drawdown      -0.075024
-
-
-
-- `excess_return_without_cost`
-    - `mean`
-        Mean value of the `CAR` (cumulative abnormal return) without cost
-    - `std`
-        The `Standard Deviation` of `CAR` (cumulative abnormal return) without cost.
-    - `annualized_return`
-        The `Annualized Rate` of `CAR` (cumulative abnormal return) without cost.
-    - `information_ratio`
-        The `Information Ratio` without cost. please refer to `Information Ratio – IR <https://www.investopedia.com/terms/i/informationratio.asp>`_.
-    - `max_drawdown`
-        The `Maximum Drawdown` of `CAR` (cumulative abnormal return) without cost, please refer to `Maximum Drawdown (MDD) <https://www.investopedia.com/terms/m/maximum-drawdown-mdd.asp>`_.
-
-- `excess_return_with_cost`
-    - `mean`
-        Mean value of the `CAR` (cumulative abnormal return) series with cost
-    - `std`
-        The `Standard Deviation` of `CAR` (cumulative abnormal return) series with cost.
-    - `annualized_return`
-        The `Annualized Rate` of `CAR` (cumulative abnormal return) with cost.
-    - `information_ratio`
-        The `Information Ratio` with cost. please refer to `Information Ratio – IR <https://www.investopedia.com/terms/i/informationratio.asp>`_.
-    - `max_drawdown`
-        The `Maximum Drawdown` of `CAR` (cumulative abnormal return) with cost, please refer to `Maximum Drawdown (MDD) <https://www.investopedia.com/terms/m/maximum-drawdown-mdd.asp>`_.
-
-
-
-Reference
-==============
-
-To know more about ``Intraday Trading``, please refer to `Intraday Trading <../reference/api.html#module-qlib.contrib.evaluate>`_.

docs/component/data.rst

@@ -21,6 +21,12 @@ The introduction of ``Data Layer`` includes the following parts.
 - Cache
 - Data and Cache File Structure
 
+Here is a typical example of Qlib data workflow
+
+- Users download data and converting data into Qlib format(with filename suffix `.bin`).  In this step, typically only some basic data are stored on disk(such as OHLCV). 
+- Creating some basic features based on Qlib's expression Engine(e.g. "Ref($close, 60) / $close", the return of last 60 trading days). Supported operators in the expression engine can be found `here <https://github.com/microsoft/qlib/blob/main/qlib/data/ops.py>`_. This step is typically implemented in Qlib's `Data Loader <https://qlib.readthedocs.io/en/latest/component/data.html#data-loader>`_ which is a component of `Data Handler <https://qlib.readthedocs.io/en/latest/component/data.html#data-handler>`_ .
+- If users require more complicated data processing (e.g. data normalization),  `Data Handler <https://qlib.readthedocs.io/en/latest/component/data.html#data-handler>`_ support user-customized processors to process data(some predefined processors can be found `here <https://github.com/microsoft/qlib/blob/main/qlib/data/dataset/processor.py>`_).  The processors are different from operators in expression engine. It is designed for some complicated data processing methods which is hard to supported in operators in expression engine.
+- At last, `Dataset <https://qlib.readthedocs.io/en/latest/component/data.html#dataset>`_ is responsible to prepare model-specific dataset from the processed data of Data Handler
 
 Data Preparation
 ============================
@@ -46,6 +52,8 @@ Also, ``Qlib`` provides a high-frequency dataset. Users can run a high-frequency
 Qlib Format Dataset
 --------------------
 ``Qlib`` has provided an off-the-shelf dataset in `.bin` format, users could use the script ``scripts/get_data.py`` to download the China-Stock dataset as follows.
+The price volume data look different from the actual dealling price because of they are **adjusted** (`adjusted price <https://www.investopedia.com/terms/a/adjusted_closing_price.asp>`_).  And then you may find that the adjusted price may be different from different data sources. This is because different data sources may vary in the way of adjusting prices. Qlib normalize the price on first trading day of each stock to 1 when adjusting them. 
+Users can leverage `$factor` to get the original trading price (e.g. `$close / $factor` to get the original close price).
 
 .. code-block:: bash
 
@@ -213,7 +221,7 @@ The `trade unit` defines the unit number of stocks can be used in a trade, and t
         
         .. code-block:: python
 
-            from qlib.config import REG_CN
+            from qlib.constant import REG_CN
             qlib.init(provider_uri='~/.qlib/qlib_data/cn_data', region=REG_CN)
         
 
@@ -338,7 +346,7 @@ DataHandlerLP
 
 In addition to use ``Data Handler`` in an automatic workflow with ``qrun``, ``Data Handler`` can be used as an independent module, by which users can easily preprocess data (standardization, remove NaN, etc.) and build datasets. 
 
-In order to achieve so, ``Qlib`` provides a base class `qlib.data.dataset.DataHandlerLP <../reference/api.html#qlib.data.dataset.handler.DataHandlerLP>`_. The core idea of this class is that: we will have some leanable ``Processors`` which can learn the parameters of data processing(e.g., parameters for zscore normalization). When new data comes in, these `trained` ``Processors`` can then process the new data and thus processing real-time data in an efficient way becomes possible. More information about ``Processors`` will be listed in the next subsection.
+In order to achieve so, ``Qlib`` provides a base class `qlib.data.dataset.DataHandlerLP <../reference/api.html#qlib.data.dataset.handler.DataHandlerLP>`_. The core idea of this class is that: we will have some learnable ``Processors`` which can learn the parameters of data processing(e.g., parameters for zscore normalization). When new data comes in, these `trained` ``Processors`` can then process the new data and thus processing real-time data in an efficient way becomes possible. More information about ``Processors`` will be listed in the next subsection.
 
 
 Interface
@@ -429,7 +437,7 @@ Dataset
 
 The ``Dataset`` module in ``Qlib`` aims to prepare data for model training and inferencing.
 
-The motivation of this module is that we want to maximize the flexibility of of different models to handle data that are suitable for themselves. This module gives the model the flexibility to process their data in an unique way. For instance, models such as ``GBDT`` may work well on data that contains `nan` or `None` value, while neural networks such as ``MLP`` will break down on such data. 
+The motivation of this module is that we want to maximize the flexibility of different models to handle data that are suitable for themselves. This module gives the model the flexibility to process their data in an unique way. For instance, models such as ``GBDT`` may work well on data that contains `nan` or `None` value, while neural networks such as ``MLP`` will break down on such data. 
 
 If user's model need process its data in a different way, user could implement his own ``Dataset`` class. If the model's
 data processing is not special, ``DatasetH`` can be used directly.

docs/component/highfreq.rst

@@ -1,120 +1,38 @@
 .. _highfreq:
 
 ============================================
-Design of hierarchical order execution framework
+Design of Nested Decision Execution Framework for High-Frequency Trading
 ============================================
 .. currentmodule:: qlib
 
 Introduction
 ===================
 
-In order to support reinforcement learning algorithms for high-frequency trading, a corresponding framework is required. None of the publicly available high-frequency trading frameworks now consider multi-layer trading mechanisms, and the currently designed algorithms cannot directly use existing frameworks.
-In addition to supporting the basic intraday multi-layer trading, the linkage with the day-ahead strategy is also a factor that affects the performance evaluation of the strategy. Different day strategies generate different order distributions and different patterns on different stocks. To verify that high-frequency trading strategies perform well on real trading orders, it is necessary to support day-frequency and high-frequency multi-level linkage trading. In addition to more accurate backtesting of high-frequency trading algorithms, if the distribution of day-frequency orders is considered when training a high-frequency trading model, the algorithm can also be optimized more for product-specific day-frequency orders.
-Therefore, innovation in the high-frequency trading framework is necessary to solve the various problems mentioned above, for which we designed a hierarchical order execution framework that can link daily-frequency and intra-day trading at different granularities.
+Daily trading (e.g. portfolio management) and intraday trading (e.g. orders execution) are two hot topics in Quant investment and usually studied separately.
+
+To get the join trading performance of daily and intraday trading, they must interact with each other and run backtest jointly.
+In order to support the joint backtest strategies in multiple levels, a corresponding framework is required. None of the publicly available high-frequency trading frameworks considers multi-level joint trading, which make the backtesting aforementioned inaccurate.
+
+Besides backtesting, the optimization of strategies from different levels is not standalone and can be affected by each other.
+For example, the best portfolio management strategy may change with the performance of order executions(e.g. a portfolio with higher turnover may becomes a better choice when we improve the order execution strategies).
+To achieve the overall good performance , it is necessary to consider the interaction of strategies in different level. 
+
+Therefore, building a new framework for trading in multiple levels becomes necessary to solve the various problems mentioned above, for which we designed a nested decision execution framework that consider the interaction of strategies.
 
 .. image:: ../_static/img/framework.svg
 
-The design of the framework is shown in the figure above. At each layer consists of Trading Agent and Execution Env. The Trading Agent has its own data processing module (Information Extractor), forecasting module (Forecast Model) and decision generator (Decision Generator). The trading algorithm generates the corresponding decisions by the Decision Generator based on the forecast signals output by the Forecast Module, and the decisions generated by the trading algorithm are passed to the Execution Env, which returns the execution results. Here the frequency of trading algorithm, decision content and execution environment can be customized by users (e.g. intra-day trading, daily-frequency trading, weekly-frequency trading), and the execution environment can be nested with finer-grained trading algorithm and execution environment inside (i.e. sub-workflow in the figure, e.g. daily-frequency orders can be turned into finer-grained decisions by splitting orders within the day). The hierarchical order execution framework is user-defined in terms of hierarchy division and decision frequency, making it easy for users to explore the effects of combining different levels of trading algorithms and breaking down the barriers between different levels of trading algorithm optimization.
-In addition to the innovation in the framework, the hierarchical order execution framework also takes into account various details of the real backtesting environment, minimizing the differences with the final real environment as much as possible. At the same time, the framework is designed to unify the interface between online and offline (e.g. data pre-processing level supports using the same set of code to process both offline and online data) to reduce the cost of strategy go-live as much as possible.
-
-Prepare Data
-===================
-.. _data:: ../../examples/highfreq/README.md
+The design of the framework is shown in the yellow part in the middle of the figure above. Each level consists of ``Trading Agent`` and ``Execution Env``. ``Trading Agent`` has its own data processing module (``Information Extractor``), forecasting module (``Forecast Model``) and decision generator (``Decision Generator``). The trading algorithm generates the decisions by the ``Decision Generator`` based on the forecast signals output by the ``Forecast Module``, and the decisions generated by the trading algorithm are passed to the ``Execution Env``, which returns the execution results. 
 
+The frequency of trading algorithm, decision content and execution environment can be customized by users (e.g. intraday trading, daily-frequency trading, weekly-frequency trading), and the execution environment can be nested with finer-grained trading algorithm and execution environment inside (i.e. sub-workflow in the figure, e.g. daily-frequency orders can be turned into finer-grained decisions by splitting orders within the day). The flexibility of nested decision execution framework makes it easy for users to explore the effects of combining different levels of trading strategies and break down the optimization barriers between different levels of trading algorithm.
 
 Example
 ===========================
 
-Here is an example of highfreq execution.
+An example of nested decision execution framework for high-frequency can be found `here <https://github.com/microsoft/qlib/blob/main/examples/nested_decision_execution/workflow.py>`_.
 
-.. code-block:: python
 
-    import qlib
-    # init qlib
-    provider_uri_day = "~/.qlib/qlib_data/cn_data"
-    provider_uri_1min = "~/.qlib/qlib_data/cn_data_1min"
-    provider_uri_map = {"1min": provider_uri_1min, "day": provider_uri_day}
-    qlib.init(provider_uri=provider_uri_day, expression_cache=None, dataset_cache=None)
+Besides, the above examples, here are some other related work about high-frequency trading in Qlib.
 
-    # data freq and backtest time
-    freq = "1min"
-    inst_list = D.list_instruments(D.instruments("all"), as_list=True)
-    start_time = "2020-01-01"
-    start_time = "2020-01-31"
-
-When initializing qlib, if the default data is used, then both daily and minute frequency data need to be passed in.
-
-.. code-block:: python
-
-    # random order strategy config
-    strategy_config = {
-        "class": "RandomOrderStrategy",
-        "module_path": "qlib.contrib.strategy.rule_strategy",
-        "kwargs": {
-            "trade_range": TradeRangeByTime("9:30", "15:00"),
-            "sample_ratio": 1.0,
-            "volume_ratio": 0.01,
-            "market": market,
-        },
-    }
-
-.. code-block:: python
-    # backtest config
-    backtest_config = {
-        "start_time": start_time,
-        "end_time": end_time,
-        "account": 100000000,
-        "benchmark": None,
-        "exchange_kwargs": {
-            "freq": freq,
-            "limit_threshold": 0.095,
-            "deal_price": "close",
-            "open_cost": 0.0005,
-            "close_cost": 0.0015,
-            "min_cost": 5,
-            "codes": market,
-        },
-        "pos_type": "InfPosition",  # Position with infinitive position
-    }
-
-please refer to "../../qlib/backtest".
-
-.. code-block:: python
-    # excutor config
-    executor_config = {
-        "class": "NestedExecutor",
-        "module_path": "qlib.backtest.executor",
-        "kwargs": {
-            "time_per_step": "day",
-            "inner_executor": {
-                "class": "SimulatorExecutor",
-                "module_path": "qlib.backtest.executor",
-                "kwargs": {
-                    "time_per_step": freq,
-                    "generate_portfolio_metrics": True,
-                    "verbose": False,
-                    # "verbose": True,
-                    "indicator_config": {
-                        "show_indicator": False,
-                    },
-                },
-            },
-            "inner_strategy": {
-                "class": "TWAPStrategy",
-                "module_path": "qlib.contrib.strategy.rule_strategy",
-            },
-            "track_data": True,
-            "generate_portfolio_metrics": True,
-            "indicator_config": {
-                "show_indicator": True,
-            },
-        },
-    }
-
-NestedExecutor represents not the innermost layer, the initialization parameters should contain inner_executor and inner_strategy. simulatorExecutor represents the current excutor is the innermost layer, the innermost strategy used here is the TWAP strategy, the framework currently also supports the VWAP strategy
-
-.. code-block:: python
-    # backtest
-    portfolio_metrics_dict, indicator_dict = backtest(executor=executor_config, strategy=strategy_config, **backtest_config)
-
-The metrics of backtest are included in the portfolio_metrics_dict and indicator_dict.
+- `Prediction with high-frequency data <https://github.com/microsoft/qlib/tree/main/examples/highfreq#benchmarks-performance-predicting-the-price-trend-in-high-frequency-data>`_
+- `Examples <https://github.com/microsoft/qlib/blob/main/examples/orderbook_data/>`_ to extract features form high-frequency data without fixed frequency.
+- `A paper <https://github.com/microsoft/qlib/tree/high-freq-execution#high-frequency-execution>`_ for high-frequency trading.

docs/component/meta.rst

@@ -0,0 +1,68 @@
+.. _meta:
+
+=================================
+Meta Controller: Meta-Task & Meta-Dataset & Meta-Model
+=================================
+.. currentmodule:: qlib
+
+
+Introduction
+=============
+``Meta Controller`` provides guidance to ``Forecast Model``, which aims to learn regular patterns among a series of forecasting tasks and use learned patterns to guide forthcoming forecasting tasks. Users can implement their own meta-model instance based on ``Meta Controller`` module.
+
+Meta Task
+=============
+
+A `Meta Task` instance is the basic element in the meta-learning framework. It saves the data that can be used for the `Meta Model`. Multiple `Meta Task` instances may share the same `Data Handler`, controlled by `Meta Dataset`. Users should use `prepare_task_data()` to obtain the data that can be directly fed into the `Meta Model`.
+
+.. autoclass:: qlib.model.meta.task.MetaTask
+    :members:
+
+Meta Dataset
+=============
+
+`Meta Dataset` controls the meta-information generating process. It is on the duty of providing data for training the `Meta Model`. Users should use `prepare_tasks` to retrieve a list of `Meta Task` instances.
+
+.. autoclass:: qlib.model.meta.dataset.MetaTaskDataset
+    :members:
+
+Meta Model
+=============
+
+General Meta Model
+------------------
+`Meta Model` instance is the part that controls the workflow. The usage of the `Meta Model` includes:
+1. Users train their `Meta Model` with the `fit` function. 
+2. The `Meta Model` instance guides the workflow by giving useful information via the `inference` function.
+
+.. autoclass:: qlib.model.meta.model.MetaModel
+    :members:
+
+Meta Task Model
+------------------
+This type of meta-model may interact with task definitions directly. Then, the `Meta Task Model` is the class for them to inherit from. They guide the base tasks by modifying the base task definitions. The function `prepare_tasks` can be used to obtain the modified base task definitions.
+
+.. autoclass:: qlib.model.meta.model.MetaTaskModel
+    :members:
+
+Meta Guide Model
+------------------
+This type of meta-model participates in the training process of the base forecasting model. The meta-model may guide the base forecasting models during their training to improve their performances.
+
+.. autoclass:: qlib.model.meta.model.MetaGuideModel
+    :members:
+
+
+Example
+=============
+``Qlib`` provides an implementation of ``Meta Model`` module, ``DDG-DA``, 
+which adapts to the market dynamics. 
+
+``DDG-DA`` includes four steps:
+
+1. Calculate meta-information and encapsulate it into ``Meta Task`` instances. All the meta-tasks form a ``Meta Dataset`` instance.
+2. Train ``DDG-DA`` based on the training data of the meta-dataset.
+3. Do the inference of the ``DDG-DA`` to get guide information.
+4. Apply guide information to the forecasting models to improve their performances.
+
+The `above example <https://github.com/microsoft/qlib/tree/main/examples/benchmarks_dynamic/DDG-DA>`_ can be found in ``examples/benchmarks_dynamic/DDG-DA/workflow.py``.

docs/component/model.rst

@@ -106,6 +106,9 @@ Example
         `SignalRecord` is the `Record Template` in ``Qlib``, please refer to `Workflow <recorder.html#record-template>`_.
 
 Also, the above example has been given in ``examples/train_backtest_analyze.ipynb``.
+Technically, the meaning of the model prediction depends on the label setting designed by user.
+By default, the meaning of the score is normally the rating of the instruments by the forecasting model. The higher the score, the more profit the instruments. 
+
 
 Custom Model
 ===================

docs/component/online.rst

@@ -23,6 +23,10 @@ The `examples <https://github.com/microsoft/qlib/tree/main/examples/online_srv>`
 
 **NOTE**: User should keep his data source updated to support online serving. For example, Qlib provides `a batch of scripts <https://github.com/microsoft/qlib/blob/main/scripts/data_collector/yahoo/README.md#automatic-update-of-daily-frequency-datafrom-yahoo-finance>`_ to help users update Yahoo daily data.
 
+Known limitations currently
+- Currently, the daily updating prediction for the next trading day is supported. But generating orders for the next trading day is not supported due to the `limitations of public data <https://github.com/microsoft/qlib/issues/215#issuecomment-766293563>_`
+
+
 Online Manager
 =============
 

docs/component/recorder.rst

@@ -37,7 +37,7 @@ Here is a general view of the structure of the system:
         
 This experiment management system defines a set of interface and provided a concrete implementation ``MLflowExpManager``, which is based on the machine learning platform: ``MLFlow`` (`link <https://mlflow.org/>`_). 
 
-If users set the implementation of ``ExpManager`` to be ``MLflowExpManager``, they can use the command `mlflow ui` to visualize and check the experiment results. For more information, pleaes refer to the related documents `here <https://www.mlflow.org/docs/latest/cli.html#mlflow-ui>`_.
+If users set the implementation of ``ExpManager`` to be ``MLflowExpManager``, they can use the command `mlflow ui` to visualize and check the experiment results. For more information, please refer to the related documents `here <https://www.mlflow.org/docs/latest/cli.html#mlflow-ui>`_.
 
 Qlib Recorder
 ===================
@@ -143,3 +143,9 @@ Here is a simple exampke of what is done in ``PortAnaRecord``, which users can r
     print(analysis_df)
 
 For more information about the APIs, please refer to `Record Template API <../reference/api.html#module-qlib.workflow.record_temp>`_.
+
+
+
+Known Limitations
+=================
+- The Python objects are saved based on pickle, which may results in issues when the environment dumping objects and loading objects are different.

docs/component/report.rst

@@ -20,6 +20,9 @@ Introduction
     - model_performance_graph
 
 
+All of the accumulated profit metrics(e.g. return, max drawdown) in Qlib are calculated by summation.
+This avoids the metrics or the plots being skewed exponentially over time.
+
 Graphical Reports
 ===================
 
@@ -101,7 +104,7 @@ Graphical Result
     - Axis Y: 
         - `ic`
             The `Pearson correlation coefficient` series between `label` and `prediction score`.
-            In the above example, the `label` is formulated as `Ref($close, -1)/$close - 1`. Please refer to `Data Feature <data.html#feature>`_ for more details.
+            In the above example, the `label` is formulated as `Ref($close, -2)/Ref($close, -1)-1`. Please refer to `Data Feature <data.html#feature>`_ for more details.
                 
         - `rank_ic`
             The `Spearman's rank correlation coefficient` series between `label` and `prediction score`.

docs/component/strategy.rst

@@ -8,11 +8,13 @@ Portfolio Strategy: Portfolio Management
 Introduction
 ===================
 
-``Portfolio Strategy`` is designed to adopt different portfolio strategies, which means that users can adopt different algorithms to generate investment portfolios based on the prediction scores of the ``Forecast Model``. Users can use the ``Portfolio Strategy`` in an automatic workflow by ``Workflow`` module, please refer to `Workflow: Workflow Management <workflow.html>`_.  
+``Portfolio Strategy`` is designed to adopt different portfolio strategies, which means that users can adopt different algorithms to generate investment portfolios based on the prediction scores of the ``Forecast Model``. Users can use the ``Portfolio Strategy`` in an automatic workflow by ``Workflow`` module, please refer to `Workflow: Workflow Management <workflow.html>`_.
 
 Because the components in ``Qlib`` are designed in a loosely-coupled way, ``Portfolio Strategy`` can be used as an independent module also.
 
-``Qlib`` provides several implemented portfolio strategies. Also, ``Qlib`` supports custom strategy, users can customize strategies according to their own needs.
+``Qlib`` provides several implemented portfolio strategies. Also, ``Qlib`` supports custom strategy, users can customize strategies according to their own requirements.
+
+After users specifying the models(forecasting signals) and strategies, running backtest will help users to check the performance of a custom model(forecasting signals)/strategy.
 
 Base Class & Interface
 ======================
@@ -20,20 +22,19 @@ Base Class & Interface
 BaseStrategy
 ------------------
 
-Qlib provides a base class ``qlib.contrib.strategy.BaseStrategy``. All strategy classes need to inherit the base class and implement its interface.
+Qlib provides a base class ``qlib.strategy.base.BaseStrategy``. All strategy classes need to inherit the base class and implement its interface.
 
-- `get_risk_degree`
-    Return the proportion of your total value you will use in investment. Dynamically risk_degree will result in Market timing.
-
-- `generate_order_list`
-    Return the order list. 
+- `generate_trade_decision`
+    generate_trade_decision is a key interface that generates trade decisions in each trading bar.
+    The frequency to call this method depends on the executor frequency("time_per_step"="day" by default). But the trading frequency can be decided by users' implementation.
+    For example, if the user wants to trading in weekly while the `time_per_step` is "day" in executor, user can return non-empty TradeDecision weekly(otherwise return empty like `this <https://github.com/microsoft/qlib/blob/main/qlib/contrib/strategy/signal_strategy.py#L132>`_ ).
 
 Users can inherit `BaseStrategy` to customize their strategy class.
 
 WeightStrategyBase
 --------------------
 
-Qlib also provides a class ``qlib.contrib.strategy.WeightStrategyBase`` that is a subclass of `BaseStrategy`. 
+Qlib also provides a class ``qlib.contrib.strategy.WeightStrategyBase`` that is a subclass of `BaseStrategy`.
 
 `WeightStrategyBase` only focuses on the target positions, and automatically generates an order list based on positions. It provides the `generate_target_weight_position` interface.
 
@@ -65,55 +66,246 @@ TopkDropoutStrategy
 - Adopt the ``Topk-Drop`` algorithm to calculate the target amount of each stock
 
     .. note::
-        ``Topk-Drop`` algorithm：
+        There are two parameters for the ``Topk-Drop`` algorithm：
 
         - `Topk`: The number of stocks held
         - `Drop`: The number of stocks sold on each trading day
+
+        In general, the number of stocks currently held is `Topk`, with the exception of being zero at the beginning period of trading.
+        For each trading day, let $d$ be the number of the instruments currently held and with a rank $\gt K$ when ranked by the prediction scores from high to low.
+        Then `d` number of stocks currently held with the worst `prediction score` will be sold, and the same number of unheld stocks with the best `prediction score` will be bought.
         
-        Currently, the number of held stocks is `Topk`.
-        On each trading day, the `Drop` number of held stocks with the worst `prediction score` will be sold, and the same number of unheld stocks with the best `prediction score` will be bought.
-        
+        In general, $d=$`Drop`, especially when the pool of the candidate instruments is large, $K$ is large, and `Drop` is small.
+         
+        In most cases, ``TopkDrop`` algorithm sells and buys `Drop` stocks every trading day, which yields a turnover rate of 2$\times$`Drop`/$K$.
+         
+        The following images illustrate a typical scenario.
         .. image:: ../_static/img/topk_drop.png
             :alt: Topk-Drop
 
-        ``TopkDrop`` algorithm sells `Drop` stocks every trading day, which guarantees a fixed turnover rate.
-        
+       
+
 - Generate the order list from the target amount
 
+EnhancedIndexingStrategy
+------------------------
+`EnhancedIndexingStrategy` Enhanced indexing combines the arts of active management and passive management,
+with the aim of outperforming a benchmark index (e.g., S&P 500) in terms of portfolio return while controlling
+the risk exposure (a.k.a. tracking error).
+
+For more information, please refer to `qlib.contrib.strategy.signal_strategy.EnhancedIndexingStrategy`
+and `qlib.contrib.strategy.optimizer.enhanced_indexing.EnhancedIndexingOptimizer`.
+
+
 Usage & Example
 ====================
-``Portfolio Strategy`` can be specified in the ``Intraday Trading(Backtest)``, the example is as follows.
+
+First, user can create a model to get trading signals(the variable name is ``pred_score`` in following cases).
+
+Prediction Score
+-----------------
+
+The `prediction score` is a pandas DataFrame. Its index is <datetime(pd.Timestamp), instrument(str)> and it must
+contains a `score` column.
+
+A prediction sample is shown as follows.
 
 .. code-block:: python
 
-    from qlib.contrib.strategy.strategy import TopkDropoutStrategy
-    from qlib.contrib.evaluate import backtest
-    STRATEGY_CONFIG = {
-        "topk": 50,
-        "n_drop": 5,
-    }
-    BACKTEST_CONFIG = {
-        "limit_threshold": 0.095,
-        "account": 100000000,
-        "benchmark": BENCHMARK,
-        "deal_price": "close",
-        "open_cost": 0.0005,
-        "close_cost": 0.0015,
-        "min_cost": 5,
-        
-    }
-    # use default strategy
-    strategy = TopkDropoutStrategy(**STRATEGY_CONFIG)
+      datetime instrument     score
+    2019-01-04   SH600000 -0.505488
+    2019-01-04   SZ002531 -0.320391
+    2019-01-04   SZ000999  0.583808
+    2019-01-04   SZ300569  0.819628
+    2019-01-04   SZ001696 -0.137140
+                 ...            ...
+    2019-04-30   SZ000996 -1.027618
+    2019-04-30   SH603127  0.225677
+    2019-04-30   SH603126  0.462443
+    2019-04-30   SH603133 -0.302460
+    2019-04-30   SZ300760 -0.126383
 
-    # pred_score is the `prediction score` output by Model
-    report_normal, positions_normal = backtest(
-        pred_score, strategy=strategy, **BACKTEST_CONFIG
-    )
+``Forecast Model`` module can make predictions, please refer to `Forecast Model: Model Training & Prediction <model.html>`_.
 
-To know more about the `prediction score` `pred_score` output by ``Forecast Model``, please refer to `Forecast Model: Model Training & Prediction <model.html>`_.
+Normally, the prediction score is the output of the models. But some models are learned from a label with a different scale. So the scale of the prediction score may be different from your expectation(e.g. the return of instruments).
+
+Qlib didn't add a step to scale the prediction score to a unified scale due to the following reasons.
+- Because not every trading strategy cares about the scale(e.g. TopkDropoutStrategy only cares about the order).  So the strategy is responsible for rescaling the prediction score(e.g. some portfolio-optimization-based strategies may require a meaningful scale).
+- The model has the flexibility to define the target, loss, and data processing. So we don't think there is a silver bullet to rescale it back directly barely based on the model's outputs. If you want to scale it back to some meaningful values(e.g. stock returns.), an intuitive solution is to create a regression model for the model's recent outputs and your recent target values.
+
+Running backtest
+-----------------
+
+- In most cases, users could backtest their portfolio management strategy  with ``backtest_daily``.
+
+    .. code-block:: python
+
+        from pprint import pprint
+
+        import qlib
+        import pandas as pd
+        from qlib.utils.time import Freq
+        from qlib.utils import flatten_dict
+        from qlib.contrib.evaluate import backtest_daily
+        from qlib.contrib.evaluate import risk_analysis
+        from qlib.contrib.strategy import TopkDropoutStrategy
+
+        # init qlib
+        qlib.init(provider_uri=<qlib data dir>)
+
+        CSI300_BENCH = "SH000300"
+        STRATEGY_CONFIG = {
+            "topk": 50,
+            "n_drop": 5,
+            # pred_score, pd.Series
+            "signal": pred_score,
+        }
+
+
+        strategy_obj = TopkDropoutStrategy(**STRATEGY_CONFIG)
+        report_normal, positions_normal = backtest_daily(
+            start_time="2017-01-01", end_time="2020-08-01", strategy=strategy_obj
+        )
+        analysis = dict()
+        # default frequency will be daily (i.e. "day")
+        analysis["excess_return_without_cost"] = risk_analysis(report_normal["return"] - report_normal["bench"])
+        analysis["excess_return_with_cost"] = risk_analysis(report_normal["return"] - report_normal["bench"] - report_normal["cost"])
+
+        analysis_df = pd.concat(analysis)  # type: pd.DataFrame
+        pprint(analysis_df)
+
+
+
+- If users would like to control their strategies in a more detailed(e.g. users have a more advanced version of executor), user could follow this example.
+
+    .. code-block:: python
+
+        from pprint import pprint
+
+        import qlib
+        import pandas as pd
+        from qlib.utils.time import Freq
+        from qlib.utils import flatten_dict
+        from qlib.backtest import backtest, executor
+        from qlib.contrib.evaluate import risk_analysis
+        from qlib.contrib.strategy import TopkDropoutStrategy
+
+        # init qlib
+        qlib.init(provider_uri=<qlib data dir>)
+
+        CSI300_BENCH = "SH000300"
+        # Benchmark is for calculating the excess return of your strategy.
+        # Its data format will be like **ONE normal instrument**. 
+        # For example, you can query its data with the code below
+        # `D.features(["SH000300"], ["$close"], start_time='2010-01-01', end_time='2017-12-31', freq='day')`
+        # It is different from the argument `market`, which indicates a universe of stocks (e.g. **A SET** of stocks like csi300)
+        # For example, you can query all data from a stock market with the code below.
+        # ` D.features(D.instruments(market='csi300'), ["$close"], start_time='2010-01-01', end_time='2017-12-31', freq='day')`
+
+        FREQ = "day"
+        STRATEGY_CONFIG = {
+            "topk": 50,
+            "n_drop": 5,
+            # pred_score, pd.Series
+            "signal": pred_score,
+        }
+
+        EXECUTOR_CONFIG = {
+            "time_per_step": "day",
+            "generate_portfolio_metrics": True,
+        }
+
+        backtest_config = {
+            "start_time": "2017-01-01",
+            "end_time": "2020-08-01",
+            "account": 100000000,
+            "benchmark": CSI300_BENCH,
+            "exchange_kwargs": {
+                "freq": FREQ,
+                "limit_threshold": 0.095,
+                "deal_price": "close",
+                "open_cost": 0.0005,
+                "close_cost": 0.0015,
+                "min_cost": 5,
+            },
+        }
+
+        # strategy object
+        strategy_obj = TopkDropoutStrategy(**STRATEGY_CONFIG)
+        # executor object
+        executor_obj = executor.SimulatorExecutor(**EXECUTOR_CONFIG)
+        # backtest
+        portfolio_metric_dict, indicator_dict = backtest(executor=executor_obj, strategy=strategy_obj, **backtest_config)
+        analysis_freq = "{0}{1}".format(*Freq.parse(FREQ))
+        # backtest info
+        report_normal, positions_normal = portfolio_metric_dict.get(analysis_freq)
+
+        # analysis
+        analysis = dict()
+        analysis["excess_return_without_cost"] = risk_analysis(
+            report_normal["return"] - report_normal["bench"], freq=analysis_freq
+        )
+        analysis["excess_return_with_cost"] = risk_analysis(
+            report_normal["return"] - report_normal["bench"] - report_normal["cost"], freq=analysis_freq
+        )
+
+        analysis_df = pd.concat(analysis)  # type: pd.DataFrame
+        # log metrics
+        analysis_dict = flatten_dict(analysis_df["risk"].unstack().T.to_dict())
+        # print out results
+        pprint(f"The following are analysis results of benchmark return({analysis_freq}).")
+        pprint(risk_analysis(report_normal["bench"], freq=analysis_freq))
+        pprint(f"The following are analysis results of the excess return without cost({analysis_freq}).")
+        pprint(analysis["excess_return_without_cost"])
+        pprint(f"The following are analysis results of the excess return with cost({analysis_freq}).")
+        pprint(analysis["excess_return_with_cost"])
+
+
+Result
+------------------
+
+The backtest results are in the following form:
+
+.. code-block:: python
+
+                                                      risk
+    excess_return_without_cost mean               0.000605
+                               std                0.005481
+                               annualized_return  0.152373
+                               information_ratio  1.751319
+                               max_drawdown      -0.059055
+    excess_return_with_cost    mean               0.000410
+                               std                0.005478
+                               annualized_return  0.103265
+                               information_ratio  1.187411
+                               max_drawdown      -0.075024
+
+
+- `excess_return_without_cost`
+    - `mean`
+        Mean value of the `CAR` (cumulative abnormal return) without cost
+    - `std`
+        The `Standard Deviation` of `CAR` (cumulative abnormal return) without cost.
+    - `annualized_return`
+        The `Annualized Rate` of `CAR` (cumulative abnormal return) without cost.
+    - `information_ratio`
+        The `Information Ratio` without cost. please refer to `Information Ratio – IR <https://www.investopedia.com/terms/i/informationratio.asp>`_.
+    - `max_drawdown`
+        The `Maximum Drawdown` of `CAR` (cumulative abnormal return) without cost, please refer to `Maximum Drawdown (MDD) <https://www.investopedia.com/terms/m/maximum-drawdown-mdd.asp>`_.
+
+- `excess_return_with_cost`
+    - `mean`
+        Mean value of the `CAR` (cumulative abnormal return) series with cost
+    - `std`
+        The `Standard Deviation` of `CAR` (cumulative abnormal return) series with cost.
+    - `annualized_return`
+        The `Annualized Rate` of `CAR` (cumulative abnormal return) with cost.
+    - `information_ratio`
+        The `Information Ratio` with cost. please refer to `Information Ratio – IR <https://www.investopedia.com/terms/i/informationratio.asp>`_.
+    - `max_drawdown`
+        The `Maximum Drawdown` of `CAR` (cumulative abnormal return) with cost, please refer to `Maximum Drawdown (MDD) <https://www.investopedia.com/terms/m/maximum-drawdown-mdd.asp>`_.
 
-To know more about ``Intraday Trading``, please refer to `Intraday Trading: Model&Strategy Testing <backtest.html>`_.
 
 Reference
 ===================
-To know more about ``Portfolio Strategy``, please refer to `Strategy API <../reference/api.html#module-qlib.contrib.strategy.strategy>`_.
+To know more about the `prediction score` `pred_score` output by ``Forecast Model``, please refer to `Forecast Model: Model Training & Prediction <model.html>`_.

docs/component/workflow.rst

@@ -124,9 +124,47 @@ Configuration File
 ===================
 
 Let's get into details of ``qrun`` in this section.
-
 Before using ``qrun``, users need to prepare a configuration file. The following content shows how to prepare each part of the configuration file.
 
+The design logic of the configuration file is very simple. It predefines fixed workflows and provide this yaml interface to users to define how to initialize each component. 
+It follow the design of `init_instance_by_config <https://github.com/microsoft/qlib/blob/2aee9e0145decc3e71def70909639b5e5a6f4b58/qlib/utils/__init__.py#L264>`_ .  It defines the initialization of each component of Qlib, which typically include the class and the initialization arguments.
+
+For example, the following yaml and code are equivalent.
+
+.. code-block:: YAML
+
+    model:
+        class: LGBModel
+        module_path: qlib.contrib.model.gbdt
+        kwargs:
+            loss: mse
+            colsample_bytree: 0.8879
+            learning_rate: 0.0421
+            subsample: 0.8789
+            lambda_l1: 205.6999
+            lambda_l2: 580.9768
+            max_depth: 8
+            num_leaves: 210
+            num_threads: 20
+
+
+.. code-block:: python
+
+        from qlib.contrib.model.gbdt import LGBModel
+        kwargs = {
+            "loss": "mse" ,
+            "colsample_bytree": 0.8879,
+            "learning_rate": 0.0421,
+            "subsample": 0.8789,
+            "lambda_l1": 205.6999,
+            "lambda_l2": 580.9768,
+            "max_depth": 8,
+            "num_leaves": 210,
+            "num_threads": 20,
+        }
+        LGBModel(kwargs)
+
+
 Qlib Init Section
 --------------------
 
@@ -195,7 +233,7 @@ The meaning of each field is as follows:
 Dataset Section
 ~~~~~~~~~~~~~~~~~~~~
 
-The `dataset` field describes the parameters for the ``Dataset`` module in ``Qlib`` as well those for the module ``DataHandler``. For more information about the ``Dataset`` module, please refer to `Qlib Model <../component/data.html#dataset>`_.
+The `dataset` field describes the parameters for the ``Dataset`` module in ``Qlib`` as well those for the module ``DataHandler``. For more information about the ``Dataset`` module, please refer to `Qlib Data <../component/data.html#dataset>`_.
 
 The keywords arguments configuration of the ``DataHandler`` is as follows:
 
@@ -210,7 +248,7 @@ The keywords arguments configuration of the ``DataHandler`` is as follows:
 
 Users can refer to the document of `DataHandler <../component/data.html#datahandler>`_ for more information about the meaning of each field in the configuration.
 
-Here is the configuration for the ``Dataset`` module which will take care of data preprossing and slicing during the training and testing phase.
+Here is the configuration for the ``Dataset`` module which will take care of data preprocessing and slicing during the training and testing phase.
 
 .. code-block:: YAML
 

docs/conf.py

@@ -54,9 +54,9 @@ master_doc = "index"
 
 
 # General information about the project.
-project = u"QLib"
-copyright = u"Microsoft"
-author = u"Microsoft"
+project = "QLib"
+copyright = "Microsoft"
+author = "Microsoft"
 
 # The version info for the project you're documenting, acts as replacement for
 # |version| and |release|, also used in various other places throughout the
@@ -174,7 +174,7 @@ latex_elements = {
 # (source start file, target name, title,
 #  author, documentclass [howto, manual, or own class]).
 latex_documents = [
-    (master_doc, "qlib.tex", u"QLib Documentation", u"Microsoft", "manual"),
+    (master_doc, "qlib.tex", "QLib Documentation", "Microsoft", "manual"),
 ]
 
 
@@ -182,7 +182,7 @@ latex_documents = [
 
 # One entry per manual page. List of tuples
 # (source start file, name, description, authors, manual section).
-man_pages = [(master_doc, "qlib", u"QLib Documentation", [author], 1)]
+man_pages = [(master_doc, "qlib", "QLib Documentation", [author], 1)]
 
 
 # -- Options for Texinfo output -------------------------------------------
@@ -194,7 +194,7 @@ texinfo_documents = [
     (
         master_doc,
         "QLib",
-        u"QLib Documentation",
+        "QLib Documentation",
         author,
         "QLib",
         "One line description of project.",

docs/developer/code_standard.rst

@@ -14,9 +14,35 @@ Continuous Integration (CI) tools help you stick to the quality standards by run
 
 When you submit a PR request, you can check whether your code passes the CI tests in the "check" section at the bottom of the web page.
 
-A common error is the mixed use of space and tab. You can fix the bug by inputing the following code in the command line.
+1. Qlib will check the code format with black. The PR will raise error if your code does not align to the standard of Qlib(e.g. a common error is the mixed use of space and tab).
+ You can fix the bug by inputing the following code in the command line.
 
-.. code-block:: python
+.. code-block:: bash
 
     pip install black
     python -m black . -l 120
+
+
+2. Qlib will check your code style pylint. The checking command is implemented in [github action workflow](https://github.com/microsoft/qlib/blob/0e8b94a552f1c457cfa6cd2c1bb3b87ebb3fb279/.github/workflows/test.yml#L66). 
+   Sometime pylint's restrictions are not that reasonable. You can ignore specific errors like this
+
+.. code-block:: python
+
+    return -ICLoss()(pred, target, index)  # pylint: disable=E1130
+
+
+3. Qlib will check your code style flake8. The checking command is implemented in [github action workflow](https://github.com/microsoft/qlib/blob/0e8b94a552f1c457cfa6cd2c1bb3b87ebb3fb279/.github/workflows/test.yml#L73).
+ You can fix the bug by inputing the following code in the command line.
+
+.. code-block:: bash
+
+    flake8 --ignore E501,F541,E402,F401,W503,E741,E266,E203,E302,E731,E262,F523,F821,F811,F841,E713,E265,W291,E712,E722,W293 qlib
+
+
+4. Qlib has integrated pre-commit, which will make it easier for developers to format their code.
+ Just run the following two commands, and the code will be automatically formatted using black and flake8 when the git commit command is executed.
+
+.. code-block:: bash
+
+    pip install -e .[dev]
+    pre-commit install

docs/hidden/tuner.rst

@@ -31,7 +31,7 @@ Let's see an example,
 
 First make sure you have the latest version of `qlib` installed.
 
-Then, you need to privide a configuration to setup the experiment.
+Then, you need to provide a configuration to setup the experiment.
 We write a simple configuration example as following,
 
 .. code-block:: YAML
@@ -217,13 +217,13 @@ The tuner pipeline contains different tuners, and the `tuner` program will proce
 Each part represents a tuner, and its modules which are to be tuned. Space in each part is the hyper-parameters' space of a certain module, you need to create your searching space and modify it in `/qlib/contrib/tuner/space.py`. We use `hyperopt` package to help us to construct the space, you can see the detail of how to use it in https://github.com/hyperopt/hyperopt/wiki/FMin .
 
 - model
-    You need to provide the `class` and the `space` of the model. If the model is user's own implementation, you need to privide the `module_path`. 
+    You need to provide the `class` and the `space` of the model. If the model is user's own implementation, you need to provide the `module_path`.
 
 - trainer
-    You need to proveide the `class` of the trainer. If the trainer is user's own implementation, you need to privide the `module_path`. 
+    You need to provide the `class` of the trainer. If the trainer is user's own implementation, you need to provide the `module_path`.
 
 - strategy
-    You need to provide the `class` and the `space` of the strategy. If the strategy is user's own implementation, you need to privide the `module_path`. 
+    You need to provide the `class` and the `space` of the strategy. If the strategy is user's own implementation, you need to provide the `module_path`.
 
 - data_label
     The label of the data, you can search which kinds of labels will lead to a better result. This part is optional, and you only need to provide `space`.
@@ -273,7 +273,7 @@ You need to use the same dataset to evaluate your different `estimator` experime
 About the data and backtest
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-`data` and `backtest` are all same in the whole `tuner` experiment. Different `estimator` experiments must use the same data and backtest method. So, these two parts of config are same with that in `estimator` configuration. You can see the precise defination of these parts in `estimator` introduction. We only provide an example here.
+`data` and `backtest` are all same in the whole `tuner` experiment. Different `estimator` experiments must use the same data and backtest method. So, these two parts of config are same with that in `estimator` configuration. You can see the precise definition of these parts in `estimator` introduction. We only provide an example here.
 
 .. code-block:: YAML
 

docs/index.rst

@@ -36,10 +36,11 @@ Document Structure
    :caption: COMPONENTS:
 
    Workflow: Workflow Management <component/workflow.rst>
-   Data Layer: Data Framework&Usage <component/data.rst>
+   Data Layer: Data Framework & Usage <component/data.rst>
    Forecast Model: Model Training & Prediction <component/model.rst>
-   Strategy: Portfolio Management <component/strategy.rst>
-   Intraday Trading: Model&Strategy Testing <component/backtest.rst>
+   Portfolio Management and Backtest <component/strategy.rst>
+   Nested Decision Execution: High-Frequency Trading <component/highfreq.rst>
+   Meta Controller: Meta-Task & Meta-Dataset & Meta-Model <component/meta.rst>
    Qlib Recorder: Experiment Management <component/recorder.rst>
    Analysis: Evaluation & Results Analysis <component/report.rst>
    Online Serving: Online Management & Strategy & Tool <component/online.rst>
@@ -52,6 +53,7 @@ Document Structure
    Online & Offline mode <advanced/server.rst>
    Serialization <advanced/serial.rst>
    Task Management <advanced/task_management.rst>
+   Point-In-Time database <advanced/PIT.rst>
 
 .. toctree::
    :maxdepth: 3

docs/introduction/introduction.rst

@@ -34,9 +34,14 @@ Name                      Description
 
 `Workflow` layer          `Workflow` layer covers the whole workflow of quantitative investment.
                           `Information Extractor` extracts data for models. `Forecast Model` focuses
-                          on producing all kinds of forecast signals (e.g. _alpha_, risk) for other
-                          modules. With these signals `Portfolio Generator` will generate the target
-                          portfolio and produce orders to be executed by `Order Executor`.
+                          on producing all kinds of forecast signals (e.g. *alpha*, risk) for other
+                          modules.  With these signals `Decision Generator` will generate the target 
+                          trading decisions(i.e. portfolio, orders)  to be executed by `Execution Env`
+                          (i.e. the trading market).  There may be multiple levels of `Trading Agent`
+                          and `Execution Env` (e.g. an *order executor trading agent and intraday
+                          order execution environment* could behave like an interday trading
+                          environment and nested in  *daily portfolio management trading agent and
+                          interday trading environment*  ) 
 
 `Interface` layer         `Interface` layer tries to present a user-friendly interface for the underlying
                           system. `Analyser` module will provide users detailed analysis reports of

docs/introduction/quick.rst

@@ -31,7 +31,7 @@ Users can easily intsall ``Qlib`` according to the following steps:
         git clone https://github.com/microsoft/qlib.git && cd qlib
         python setup.py install
 
-To kown more about `installation`, please refer to `Qlib Installation <../start/installation.html>`_.
+To known more about `installation`, please refer to `Qlib Installation <../start/installation.html>`_.
 
 Prepare Data
 ==============
@@ -44,7 +44,7 @@ Load and prepare data by running the following code:
 
 This dataset is created by public data collected by crawler scripts in ``scripts/data_collector/``, which have been released in the same repository. Users could create the same dataset with it.
 
-To kown more about `prepare data`, please refer to `Data Preparation <../component/data.html#data-preparation>`_.
+To known more about `prepare data`, please refer to `Data Preparation <../component/data.html#data-preparation>`_.
 
 Auto Quant Research Workflow
 ====================================

docs/requirements.txt

@@ -3,3 +3,4 @@ cmake
 numpy
 scipy
 scikit-learn
+pandas

docs/start/getdata.rst

@@ -120,6 +120,32 @@ For more details about features, please refer `Feature API <../component/data.ht
 
 .. note:: When calling `D.features()` at the client, use parameter `disk_cache=0` to skip dataset cache, use `disk_cache=1` to generate and use dataset cache. In addition, when calling at the server, users can use `disk_cache=2` to update the dataset cache.
 
+
+When you are building complicated expressions, implementing all the expressions in a single string may not be easy.
+For example, it looks quite long and complicated:
+
+.. code-block:: python
+
+   >> from qlib.data import D
+   >> data = D.features(["sh600519"], ["(($high / $close) + ($open / $close)) * (($high / $close) + ($open / $close)) / ($high / $close) + ($open / $close)"], start_time="20200101")
+
+
+But using string is not the only way to implement the expression. You can also implement expression by code.
+Here is an exmaple which does the same thing as above examples.
+
+
+.. code-block:: python
+
+   >> from qlib.data.ops import *
+   >> f1 = Feature("high") / Feature("close")
+   >> f2 = Feature("open") / Feature("close")
+   >> f3 = f1 + f2
+   >> f4 = f3 * f3 / f3
+
+   >> data = D.features(["sh600519"], [f4], start_time="20200101")
+   >> data.head()
+
+
 API
 ====================
 To know more about how to use the Data, go to API Reference: `Data API <../reference/api.html#data>`_

docs/start/initialization.rst

@@ -27,7 +27,7 @@ Initialize Qlib before calling other APIs: run following code in python.
 
         import qlib
         # region in [REG_CN, REG_US]
-        from qlib.config import REG_CN
+        from qlib.constant import REG_CN
         provider_uri = "~/.qlib/qlib_data/cn_data"  # target_dir
         qlib.init(provider_uri=provider_uri, region=REG_CN)
     
@@ -37,17 +37,19 @@ Initialize Qlib before calling other APIs: run following code in python.
 Parameters
 -------------------
 
-Besides `provider_uri` and `region`, `qlib.init` has other parameters. The following are several important parameters of `qlib.init`:
+Besides `provider_uri` and `region`, `qlib.init` has other parameters.
+The following are several important parameters of `qlib.init` (`Qlib` has a lot of config. Only part of parameters are limited here. More detailed setting can be found `here <https://github.com/microsoft/qlib/blob/main/qlib/config.py>`_):
 
 - `provider_uri`
     Type: str. The URI of the Qlib data. For example, it could be the location where the data loaded by ``get_data.py`` are stored.
 - `region`
-    Type: str, optional parameter(default: `qlib.config.REG_CN`).
-        Currently: ``qlib.config.REG_US`` ('us') and ``qlib.config.REG_CN`` ('cn') is supported. Different value of  `region` will result in different stock market mode.
-        - ``qlib.config.REG_US``: US stock market.
-        - ``qlib.config.REG_CN``: China stock market.
+    Type: str, optional parameter(default: `qlib.constant.REG_CN`).
+        Currently: ``qlib.constant.REG_US`` ('us') and ``qlib.constant.REG_CN`` ('cn') is supported. Different value of  `region` will result in different stock market mode.
+        - ``qlib.constant.REG_US``: US stock market.
+        - ``qlib.constant.REG_CN``: China stock market.
 
         Different modes will result in different trading limitations and costs.
+        The region is just `shortcuts for defining a batch of configurations <https://github.com/microsoft/qlib/blob/528f74af099bf6156e9480bcd2bb28e453231212/qlib/config.py#L249>`_, which include minimal trading order unit (``trade_unit``),  trading limitation (``limit_threshold``) , etc.  It is not a necessary part and users can set the key configurations manually if the existing region setting can't meet their requirements.
 - `redis_host`
     Type: str, optional parameter(default: "127.0.0.1"), host of `redis`
         The lock and cache mechanism relies on redis.
@@ -87,3 +89,9 @@ Besides `provider_uri` and `region`, `qlib.init` has other parameters. The follo
             "task_url": "mongodb://localhost:27017/",  # your mongo url
             "task_db_name": "rolling_db", # the database name of Task Management
         })
+
+- `logging_level`
+    The logging level for the system.
+
+- `kernels`
+    The number of processes used when calculating features in Qlib's expression engine. It is very helpful to set it to 1 when you are debuggin an expression calculating exception

examples/benchmarks/ADARNN/README.md

@@ -0,0 +1,4 @@
+# AdaRNN
+* Code: [https://github.com/jindongwang/transferlearning/tree/master/code/deep/adarnn](https://github.com/jindongwang/transferlearning/tree/master/code/deep/adarnn)
+* Paper: [AdaRNN: Adaptive Learning and Forecasting for Time Series](https://arxiv.org/pdf/2108.04443.pdf).
+

examples/benchmarks/ADARNN/requirements.txt

@@ -0,0 +1,4 @@
+pandas==1.1.2
+numpy==1.21.0
+scikit_learn==0.23.2
+torch==1.7.0

examples/benchmarks/ADARNN/workflow_config_adarnn_Alpha360.yaml

@@ -0,0 +1,88 @@
+qlib_init:
+    provider_uri: "~/.qlib/qlib_data/cn_data"
+    region: cn
+market: &market csi300
+benchmark: &benchmark SH000300
+data_handler_config: &data_handler_config
+    start_time: 2008-01-01
+    end_time: 2020-08-01
+    fit_start_time: 2008-01-01
+    fit_end_time: 2014-12-31
+    instruments: *market
+    infer_processors:
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
+port_analysis_config: &port_analysis_config
+    strategy:
+        class: TopkDropoutStrategy
+        module_path: qlib.contrib.strategy
+        kwargs:
+            model: <MODEL>
+            dataset: <DATASET>
+            topk: 50
+            n_drop: 5
+    backtest:
+        start_time: 2017-01-01
+        end_time: 2020-08-01
+        account: 100000000
+        benchmark: *benchmark
+        exchange_kwargs:
+            limit_threshold: 0.095
+            deal_price: close
+            open_cost: 0.0005
+            close_cost: 0.0015
+            min_cost: 5
+task:
+    model:
+        class: ADARNN
+        module_path: qlib.contrib.model.pytorch_adarnn
+        kwargs:
+            d_feat: 6
+            hidden_size: 64
+            num_layers: 2
+            dropout: 0.0
+            n_epochs: 200
+            lr: 1e-3
+            early_stop: 20
+            batch_size: 800
+            metric: loss
+            loss: mse
+            GPU: 0
+    dataset:
+        class: DatasetH
+        module_path: qlib.data.dataset
+        kwargs:
+            handler:
+                class: Alpha360
+                module_path: qlib.contrib.data.handler
+                kwargs: *data_handler_config
+            segments:
+                train: [2008-01-01, 2014-12-31]
+                valid: [2015-01-01, 2016-12-31]
+                test: [2017-01-01, 2020-08-01]
+    record: 
+        - class: SignalRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            model: <MODEL>
+            dataset: <DATASET>
+        - class: SigAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            ana_long_short: False
+            ann_scaler: 252
+        - class: PortAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            config: *port_analysis_config

examples/benchmarks/ADD/README.md

@@ -0,0 +1,3 @@
+# ADD
+* Paper: [ADD: Augmented Disentanglement Distillation Framework for Improving Stock Trend Forecasting](https://arxiv.org/abs/2012.06289).
+

examples/benchmarks/ADD/requirements.txt

@@ -0,0 +1,4 @@
+numpy==1.21.0
+pandas==1.1.2
+scikit_learn==0.23.2
+torch==1.7.0

examples/benchmarks/ADD/workflow_config_add_Alpha360.yaml

@@ -0,0 +1,94 @@
+qlib_init:
+    provider_uri: "~/.qlib/qlib_data/cn_data"
+    region: cn
+market: &market csi300
+benchmark: &benchmark SH000300
+data_handler_config: &data_handler_config
+    start_time: 2008-01-01
+    end_time: 2020-08-01
+    fit_start_time: 2008-01-01
+    fit_end_time: 2014-12-31
+    instruments: *market
+    infer_processors:
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
+port_analysis_config: &port_analysis_config
+    strategy:
+        class: TopkDropoutStrategy
+        module_path: qlib.contrib.strategy
+        kwargs:
+            signal:
+                - <MODEL>
+                - <DATASET>
+            topk: 50
+            n_drop: 5
+    backtest:
+        start_time: 2017-01-01
+        end_time: 2020-08-01
+        account: 100000000
+        benchmark: *benchmark
+        exchange_kwargs:
+            limit_threshold: 0.095
+            deal_price: close
+            open_cost: 0.0005
+            close_cost: 0.0015
+            min_cost: 5
+task:
+    model:
+        class: ADD
+        module_path: qlib.contrib.model.pytorch_add
+        kwargs:
+            d_feat: 6
+            hidden_size: 64
+            num_layers: 2
+            dropout: 0.1
+            dec_dropout: 0.0
+            n_epochs: 200
+            lr: 1e-3
+            early_stop: 20
+            batch_size: 5000
+            metric: ic
+            base_model: GRU
+            gamma: 0.1
+            gamma_clip: 0.2
+            optimizer: adam
+            mu: 0.2
+            GPU: 0
+    dataset:
+        class: DatasetH
+        module_path: qlib.data.dataset
+        kwargs:
+            handler:
+                class: Alpha360
+                module_path: qlib.contrib.data.handler
+                kwargs: *data_handler_config
+            segments:
+                train: [2008-01-01, 2014-12-31]
+                valid: [2015-01-01, 2016-12-31]
+                test: [2017-01-01, 2020-08-01]
+    record:
+        - class: SignalRecord
+          module_path: qlib.workflow.record_temp
+          kwargs:
+            model: <MODEL>
+            dataset: <DATASET>
+        - class: SigAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs:
+            ana_long_short: False
+            ann_scaler: 252
+        - class: PortAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs:
+            config: *port_analysis_config

examples/benchmarks/ALSTM/README.md

@@ -6,3 +6,4 @@
 
   [https://www.ijcai.org/Proceedings/2017/0366.pdf](https://www.ijcai.org/Proceedings/2017/0366.pdf)
 
+- NOTE: Current version of implementation is just a simplified version of ALSTM. It is an LSTM with attention.

examples/benchmarks/ALSTM/requirements.txt

@@ -1,4 +1,4 @@
-numpy==1.17.4
+numpy==1.21.0
 pandas==1.1.2
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/CatBoost/requirements.txt

@@ -1,3 +1,3 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 catboost==0.24.3

examples/benchmarks/DoubleEnsemble/requirements.txt

@@ -1,3 +1,3 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 lightgbm==3.1.0

examples/benchmarks/GATs/requirements.txt

@@ -1,4 +1,4 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/GRU/README.md

@@ -0,0 +1,2 @@
+# Gated Recurrent Unit (GRU)
+* Paper: [Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation](https://aclanthology.org/D14-1179.pdf).

examples/benchmarks/GRU/requirements.txt

@@ -1,4 +1,4 @@
-numpy==1.17.4
+numpy==1.21.0
 pandas==1.1.2
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/HIST/README.md

@@ -0,0 +1,3 @@
+# HIST
+* Code: [https://github.com/Wentao-Xu/HIST](https://github.com/Wentao-Xu/HIST)
+* Paper: [HIST: A Graph-based Framework for Stock Trend Forecasting via Mining Concept-Oriented Shared InformationAdaRNN: Adaptive Learning and Forecasting for Time Series](https://arxiv.org/abs/2110.13716).

examples/benchmarks/HIST/requirements.txt

@@ -0,0 +1,4 @@
+pandas==1.1.2
+numpy==1.21.0
+scikit_learn==0.23.2
+torch==1.7.0

examples/benchmarks/HIST/workflow_config_hist_Alpha360.yaml

@@ -0,0 +1,92 @@
+qlib_init:
+    provider_uri: "~/.qlib/qlib_data/cn_data"
+    region: cn
+market: &market csi300
+benchmark: &benchmark SH000300
+data_handler_config: &data_handler_config
+    start_time: 2008-01-01
+    end_time: 2020-08-01
+    fit_start_time: 2008-01-01
+    fit_end_time: 2014-12-31
+    instruments: *market
+    infer_processors:
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
+port_analysis_config: &port_analysis_config
+    strategy:
+        class: TopkDropoutStrategy
+        module_path: qlib.contrib.strategy
+        kwargs:
+            signal:
+                - <MODEL> 
+                - <DATASET>
+            topk: 50
+            n_drop: 5
+    backtest:
+        start_time: 2017-01-01
+        end_time: 2020-08-01
+        account: 100000000
+        benchmark: *benchmark
+        exchange_kwargs:
+            limit_threshold: 0.095
+            deal_price: close
+            open_cost: 0.0005
+            close_cost: 0.0015
+            min_cost: 5
+task:
+    model:
+        class: HIST
+        module_path: qlib.contrib.model.pytorch_hist
+        kwargs:
+            d_feat: 6
+            hidden_size: 64
+            num_layers: 2
+            dropout: 0
+            n_epochs: 200
+            lr: 1e-4
+            early_stop: 20
+            metric: ic
+            loss: mse
+            base_model: LSTM
+            model_path: "benchmarks/LSTM/model_lstm_csi300.pkl"
+            stock2concept: "benchmarks/HIST/qlib_csi300_stock2concept.npy"
+            stock_index: "benchmarks/HIST/qlib_csi300_stock_index.npy"
+            GPU: 0
+    dataset:
+        class: DatasetH
+        module_path: qlib.data.dataset
+        kwargs:
+            handler:
+                class: Alpha360
+                module_path: qlib.contrib.data.handler
+                kwargs: *data_handler_config
+            segments:
+                train: [2008-01-01, 2014-12-31]
+                valid: [2015-01-01, 2016-12-31]
+                test: [2017-01-01, 2020-08-01]
+    record: 
+        - class: SignalRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            model: <MODEL>
+            dataset: <DATASET>
+        - class: SigAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            ana_long_short: False
+            ann_scaler: 252
+        - class: PortAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            config: *port_analysis_config

examples/benchmarks/IGMTF/README.md

@@ -0,0 +1,4 @@
+# IGMTF
+* Code: [https://github.com/Wentao-Xu/IGMTF](https://github.com/Wentao-Xu/IGMTF)
+* Paper: [IGMTF: An Instance-wise Graph-based Framework for
+Multivariate Time Series Forecasting](https://arxiv.org/abs/2109.06489).

examples/benchmarks/IGMTF/requirements.txt

@@ -0,0 +1,4 @@
+pandas==1.1.2
+numpy==1.21.0
+scikit_learn==0.23.2
+torch==1.7.0

examples/benchmarks/IGMTF/workflow_config_igmtf_Alpha360.yaml

@@ -0,0 +1,89 @@
+qlib_init:
+    provider_uri: "~/.qlib/qlib_data/cn_data"
+    region: cn
+market: &market csi300
+benchmark: &benchmark SH000300
+data_handler_config: &data_handler_config
+    start_time: 2008-01-01
+    end_time: 2020-08-01
+    fit_start_time: 2008-01-01
+    fit_end_time: 2014-12-31
+    instruments: *market
+    infer_processors:
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
+port_analysis_config: &port_analysis_config
+    strategy:
+        class: TopkDropoutStrategy
+        module_path: qlib.contrib.strategy
+        kwargs:
+            model: <MODEL>
+            dataset: <DATASET>
+            topk: 50
+            n_drop: 5
+    backtest:
+        start_time: 2017-01-01
+        end_time: 2020-08-01
+        account: 100000000
+        benchmark: *benchmark
+        exchange_kwargs:
+            limit_threshold: 0.095
+            deal_price: close
+            open_cost: 0.0005
+            close_cost: 0.0015
+            min_cost: 5
+task:
+    model:
+        class: IGMTF
+        module_path: qlib.contrib.model.pytorch_igmtf
+        kwargs:
+            d_feat: 6
+            hidden_size: 64
+            num_layers: 2
+            dropout: 0
+            n_epochs: 200
+            lr: 1e-4
+            early_stop: 20
+            metric: ic
+            loss: mse
+            base_model: LSTM
+            model_path: "benchmarks/LSTM/model_lstm_csi300.pkl"
+            GPU: 0
+    dataset:
+        class: DatasetH
+        module_path: qlib.data.dataset
+        kwargs:
+            handler:
+                class: Alpha360
+                module_path: qlib.contrib.data.handler
+                kwargs: *data_handler_config
+            segments:
+                train: [2008-01-01, 2014-12-31]
+                valid: [2015-01-01, 2016-12-31]
+                test: [2017-01-01, 2020-08-01]
+    record: 
+        - class: SignalRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            model: <MODEL>
+            dataset: <DATASET>
+        - class: SigAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            ana_long_short: False
+            ann_scaler: 252
+        - class: PortAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            config: *port_analysis_config

examples/benchmarks/LSTM/README.md

@@ -0,0 +1,2 @@
+# Long Short-Term Memory (LSTM)
+* Paper: [Long Short-Term Memory](https://direct.mit.edu/neco/article-abstract/9/8/1735/6109/Long-Short-Term-Memory?redirectedFrom=fulltext).

examples/benchmarks/LSTM/requirements.txt

@@ -1,4 +1,4 @@
-numpy==1.17.4
+numpy==1.21.0
 pandas==1.1.2
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/LightGBM/requirements.txt

@@ -1,3 +1,3 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 lightgbm==3.1.0

examples/benchmarks/Linear/workflow_config_linear_Alpha158.yaml

@@ -22,7 +22,6 @@ data_handler_config: &data_handler_config
         - class: CSRankNorm
           kwargs:
               fields_group: label
-    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
 port_analysis_config: &port_analysis_config
     strategy:
         class: TopkDropoutStrategy

examples/benchmarks/Localformer/README.md

@@ -0,0 +1 @@
+# Localformer

examples/benchmarks/Localformer/requirements.txt

@@ -1,3 +1,3 @@
-numpy==1.17.4
+numpy==1.21.0
 pandas==1.1.2
 torch==1.2.0

examples/benchmarks/MLP/README.md

@@ -0,0 +1 @@
+# Multi-Layer Perceptron (MLP)

examples/benchmarks/MLP/requirements.txt

@@ -1,4 +1,4 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/MLP/workflow_config_mlp_Alpha158.yaml

@@ -63,8 +63,6 @@ task:
         module_path: qlib.contrib.model.pytorch_nn
         kwargs:
             loss: mse
-            input_dim: 157
-            output_dim: 1
             lr: 0.002
             lr_decay: 0.96
             lr_decay_steps: 100
@@ -73,6 +71,8 @@ task:
             batch_size: 8192
             GPU: 0
             weight_decay: 0.0002
+            pt_model_kwargs:
+              input_dim: 157
     dataset:
         class: DatasetH
         module_path: qlib.data.dataset

examples/benchmarks/MLP/workflow_config_mlp_Alpha360.yaml

@@ -51,8 +51,6 @@ task:
         module_path: qlib.contrib.model.pytorch_nn
         kwargs:
             loss: mse
-            input_dim: 360
-            output_dim: 1
             lr: 0.002
             lr_decay: 0.96
             lr_decay_steps: 100
@@ -60,6 +58,8 @@ task:
             max_steps: 8000
             batch_size: 4096
             GPU: 0
+            pt_model_kwargs:
+              input_dim: 360
     dataset:
         class: DatasetH
         module_path: qlib.data.dataset

examples/benchmarks/README.md

@@ -1,21 +1,30 @@
 # Benchmarks Performance
+This page lists a batch of methods designed for alpha seeking. Each method tries to give scores/predictions for all stocks each day(e.g. forecasting the future excess return of stocks). The scores/predictions of the models will be used as the mined alpha. Investing in stocks with higher scores is expected to yield more profit.  
+
+The alpha is evaluated in two ways.
+1. The correlation between the alpha and future return.
+1. Constructing portfolio based on the alpha and evaluating the final total return.
+   - The explanation of metrics can be found [here](https://qlib.readthedocs.io/en/latest/component/report.html#id4)
 
 Here are the results of each benchmark model running on Qlib's `Alpha360` and `Alpha158` dataset with China's A shared-stock & CSI300 data respectively. The values of each metric are the mean and std calculated based on 20 runs with different random seeds.
 
 The numbers shown below demonstrate the performance of the entire `workflow` of each model. We will update the `workflow` as well as models in the near future for better results.
 <!-- 
-> If you need to reproduce the results below, please use the **v1** dataset: `python scripts/get_data.py qlib_data --target_dir ~/.qlib/qlib_data/qlib_cn_1d --region cn --version v1`
+> If you need to reproduce the results below, please use the **v1** dataset: `python scripts/get_data.py qlib_data --target_dir ~/.qlib/qlib_data/cn_data --region cn --version v1`
 >
 > In the new version of qlib, the default dataset is **v2**. Since the data is collected from the YahooFinance API (which is not very stable), the results of *v2* and *v1* may differ -->
 
 > NOTE:
 > The backtest start from 0.8.0 is quite different from previous version. Please check out the changelog for the difference.
 
+> NOTE:
+> We have very limited resources to implement and finetune the models. We tried our best effort to fairly compare these models.  But some models may have greater potential than what it looks like in the table below.  Your contribution is highly welcomed to explore their potential.
 
 ## Alpha158 dataset
 
 | Model Name                               | Dataset                             | IC          | ICIR        | Rank IC     | Rank ICIR   | Annualized Return | Information Ratio | Max Drawdown |
 |------------------------------------------|-------------------------------------|-------------|-------------|-------------|-------------|-------------------|-------------------|--------------|
+| TCN(Shaojie Bai, et al.)                 | Alpha158                            | 0.0275±0.00 | 0.2157±0.01 | 0.0411±0.00 | 0.3379±0.01 | 0.0190±0.02       | 0.2887±0.27       | -0.1202±0.03 |
 | TabNet(Sercan O. Arik, et al.)           | Alpha158                            | 0.0204±0.01 | 0.1554±0.07 | 0.0333±0.00 | 0.2552±0.05 | 0.0227±0.04       | 0.3676±0.54       | -0.1089±0.08 |
 | Transformer(Ashish Vaswani, et al.)      | Alpha158                            | 0.0264±0.00 | 0.2053±0.02 | 0.0407±0.00 | 0.3273±0.02 | 0.0273±0.02       | 0.3970±0.26       | -0.1101±0.02 |
 | GRU(Kyunghyun Cho, et al.)               | Alpha158(with selected 20 features) | 0.0315±0.00 | 0.2450±0.04 | 0.0428±0.00 | 0.3440±0.03 | 0.0344±0.02       | 0.5160±0.25       | -0.1017±0.02 |
@@ -35,7 +44,6 @@ The numbers shown below demonstrate the performance of the entire `workflow` of
 | DoubleEnsemble(Chuheng Zhang, et al.)    | Alpha158                            | 0.0544±0.00 | 0.4340±0.00 | 0.0523±0.00 | 0.4284±0.01 | 0.1168±0.01       | 1.3384±0.12       | -0.1036±0.01 |
 
 
-
 ## Alpha360 dataset
 
 | Model Name                                | Dataset  | IC          | ICIR        | Rank IC     | Rank ICIR   | Annualized Return | Information Ratio | Max Drawdown |
@@ -48,13 +56,42 @@ The numbers shown below demonstrate the performance of the entire `workflow` of
 | XGBoost(Tianqi Chen, et al.)              | Alpha360 | 0.0394±0.00 | 0.2909±0.00 | 0.0448±0.00 | 0.3679±0.00 | 0.0344±0.00       | 0.4527±0.02       | -0.1004±0.00 |
 | DoubleEnsemble(Chuheng Zhang, et al.)     | Alpha360 | 0.0404±0.00 | 0.3023±0.00 | 0.0495±0.00 | 0.3898±0.00 | 0.0468±0.01       | 0.6302±0.20       | -0.0860±0.01 |
 | LightGBM(Guolin Ke, et al.)               | Alpha360 | 0.0400±0.00 | 0.3037±0.00 | 0.0499±0.00 | 0.4042±0.00 | 0.0558±0.00       | 0.7632±0.00       | -0.0659±0.00 |
+| TCN(Shaojie Bai, et al.)                  | Alpha360 | 0.0441±0.00 | 0.3301±0.02 | 0.0519±0.00 | 0.4130±0.01 | 0.0604±0.02       | 0.8295±0.34       | -0.1018±0.03 |
 | ALSTM (Yao Qin, et al.)                   | Alpha360 | 0.0497±0.00 | 0.3829±0.04 | 0.0599±0.00 | 0.4736±0.03 | 0.0626±0.02       | 0.8651±0.31       | -0.0994±0.03 |
 | LSTM(Sepp Hochreiter, et al.)             | Alpha360 | 0.0448±0.00 | 0.3474±0.04 | 0.0549±0.00 | 0.4366±0.03 | 0.0647±0.03       | 0.8963±0.39       | -0.0875±0.02 |
+| ADD                                       | Alpha360 | 0.0430±0.00 | 0.3188±0.04 | 0.0559±0.00 | 0.4301±0.03 | 0.0667±0.02       | 0.8992±0.34       | -0.0855±0.02 |
 | GRU(Kyunghyun Cho, et al.)                | Alpha360 | 0.0493±0.00 | 0.3772±0.04 | 0.0584±0.00 | 0.4638±0.03 | 0.0720±0.02       | 0.9730±0.33       | -0.0821±0.02 |
+| AdaRNN(Yuntao Du, et al.)                 | Alpha360 | 0.0464±0.01 | 0.3619±0.08 | 0.0539±0.01 | 0.4287±0.06 | 0.0753±0.03       | 1.0200±0.40       | -0.0936±0.03 |
 | GATs (Petar Velickovic, et al.)           | Alpha360 | 0.0476±0.00 | 0.3508±0.02 | 0.0598±0.00 | 0.4604±0.01 | 0.0824±0.02       | 1.1079±0.26       | -0.0894±0.03 |
 | TCTS(Xueqing Wu, et al.)                  | Alpha360 | 0.0508±0.00 | 0.3931±0.04 | 0.0599±0.00 | 0.4756±0.03 | 0.0893±0.03       | 1.2256±0.36       | -0.0857±0.02 |
 | TRA(Hengxu Lin, et al.)                   | Alpha360 | 0.0485±0.00 | 0.3787±0.03 | 0.0587±0.00 | 0.4756±0.03 | 0.0920±0.03       | 1.2789±0.42       | -0.0834±0.02 |
+| IGMTF(Wentao Xu, et al.)                  | Alpha360 | 0.0480±0.00 | 0.3589±0.02 | 0.0606±0.00 | 0.4773±0.01 | 0.0946±0.02       | 1.3509±0.25       | -0.0716±0.02 |
+| HIST(Wentao Xu, et al.)                   | Alpha360 | 0.0522±0.00 | 0.3530±0.01 | 0.0667±0.00 | 0.4576±0.01 | 0.0987±0.02       | 1.3726±0.27       | -0.0681±0.01 |
+
 
 - The selected 20 features are based on the feature importance of a lightgbm-based model.
 - The base model of DoubleEnsemble is LGBM.
 - The base model of TCTS is GRU.
+- About the datasets
+  - Alpha158 is a tabular dataset. There are less spatial relationships between different features. Each feature are carefully desgined by human (a.k.a feature engineering)
+  - Alpha360 contains raw price and volue data without much feature engineering. There are strong strong spatial relationships between the features in the time dimension.
+- The metrics can be categorized into two
+   - Signal-based evaluation:  IC, ICIR, Rank IC, Rank ICIR
+   - Portfolio-based metrics:  Annualized Return, Information Ratio, Max Drawdown
+
+
+# Contributing
+
+Your contributions to new models are highly welcome!
+
+If you want to contribute your new models, you can follow the steps below.
+1. Create a folder for your model
+2. The folder contains following items(you can refer to [this example](https://github.com/microsoft/qlib/tree/main/examples/benchmarks/TCTS)).
+    - `requirements.txt`: required dependencies.
+    - `README.md`: a brief introduction to your models
+    - `workflow_config_<model name>_<dataset>.yaml`: a configuration which can read by `qrun`. You are encouraged to run your model in all datasets.
+3. You can integrate your model as a module [in this folder](https://github.com/microsoft/qlib/tree/main/qlib/contrib/model).
+4. Please updated your results in the benchmark tables, e.g. [Alpha360](#alpha158-dataset), [Alpha158](#alpha158-dataset)(the values of each metric are the mean and std calculated based on 20 runs with different random seeds, if you don't have enough computational resource, you can ask for help in the PR).
+5. Update the info in the index page in the [news list](https://github.com/microsoft/qlib#newspaper-whats-new----sparkling_heart) and [model list](https://github.com/microsoft/qlib#quant-model-paper-zoo).
+
+Finally, you can send PR for review. ([here is an example](https://github.com/microsoft/qlib/pull/1040))

examples/benchmarks/SFM/requirements.txt

@@ -1,4 +1,4 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/TCN/README.md

@@ -0,0 +1,4 @@
+# TCN
+* Code: [https://github.com/locuslab/TCN](https://github.com/locuslab/TCN)
+* Paper: [An Empirical Evaluation of Generic Convolutional and Recurrent Networks for Sequence Modeling](https://arxiv.org/abs/1803.01271).
+

examples/benchmarks/TCN/requirements.txt

@@ -0,0 +1,4 @@
+numpy==1.21.0
+pandas==1.1.2
+scikit_learn==0.23.2
+torch==1.7.0

examples/benchmarks/TCN/workflow_config_tcn_Alpha158.yaml

@@ -0,0 +1,100 @@
+qlib_init:
+    provider_uri: "~/.qlib/qlib_data/cn_data"
+    region: cn
+market: &market csi300
+benchmark: &benchmark SH000300
+data_handler_config: &data_handler_config
+    start_time: 2008-01-01
+    end_time: 2020-08-01
+    fit_start_time: 2008-01-01
+    fit_end_time: 2014-12-31
+    instruments: *market
+    infer_processors:
+        - class: FilterCol
+          kwargs:
+              fields_group: feature
+              col_list: ["RESI5", "WVMA5", "RSQR5", "KLEN", "RSQR10", "CORR5", "CORD5", "CORR10", 
+                            "ROC60", "RESI10", "VSTD5", "RSQR60", "CORR60", "WVMA60", "STD5", 
+                            "RSQR20", "CORD60", "CORD10", "CORR20", "KLOW"
+                        ]
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
+
+port_analysis_config: &port_analysis_config
+    strategy:
+        class: TopkDropoutStrategy
+        module_path: qlib.contrib.strategy
+        kwargs:
+            model: <MODEL>
+            dataset: <DATASET>
+            topk: 50
+            n_drop: 5
+    backtest:
+        start_time: 2017-01-01
+        end_time: 2020-08-01
+        account: 100000000
+        benchmark: *benchmark
+        exchange_kwargs:
+            limit_threshold: 0.095
+            deal_price: close
+            open_cost: 0.0005
+            close_cost: 0.0015
+            min_cost: 5
+task:
+    model:
+        class: TCN
+        module_path: qlib.contrib.model.pytorch_tcn_ts
+        kwargs:
+            d_feat: 20
+            num_layers: 5
+            n_chans: 32
+            kernel_size: 7
+            dropout: 0.5
+            n_epochs: 200
+            lr: 1e-4
+            early_stop: 20
+            batch_size: 2000
+            metric: loss
+            loss: mse
+            optimizer: adam
+            n_jobs: 20
+            GPU: 0
+    dataset:
+        class: TSDatasetH
+        module_path: qlib.data.dataset
+        kwargs:
+            handler:
+                class: Alpha158
+                module_path: qlib.contrib.data.handler
+                kwargs: *data_handler_config
+            segments:
+                train: [2008-01-01, 2014-12-31]
+                valid: [2015-01-01, 2016-12-31]
+                test: [2017-01-01, 2020-08-01]
+            step_len: 20
+    record: 
+        - class: SignalRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            model: <MODEL>
+            dataset: <DATASET>
+        - class: SigAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            ana_long_short: False
+            ann_scaler: 252
+        - class: PortAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            config: *port_analysis_config

examples/benchmarks/TCN/workflow_config_tcn_Alpha360.yaml

@@ -0,0 +1,90 @@
+qlib_init:
+    provider_uri: "~/.qlib/qlib_data/cn_data"
+    region: cn
+market: &market csi300
+benchmark: &benchmark SH000300
+data_handler_config: &data_handler_config
+    start_time: 2008-01-01
+    end_time: 2020-08-01
+    fit_start_time: 2008-01-01
+    fit_end_time: 2014-12-31
+    instruments: *market
+    infer_processors:
+        - class: RobustZScoreNorm
+          kwargs:
+              fields_group: feature
+              clip_outlier: true
+        - class: Fillna
+          kwargs:
+              fields_group: feature
+    learn_processors:
+        - class: DropnaLabel
+        - class: CSRankNorm
+          kwargs:
+              fields_group: label
+    label: ["Ref($close, -2) / Ref($close, -1) - 1"]
+port_analysis_config: &port_analysis_config
+    strategy:
+        class: TopkDropoutStrategy
+        module_path: qlib.contrib.strategy
+        kwargs:
+            model: <MODEL>
+            dataset: <DATASET>
+            topk: 50
+            n_drop: 5
+    backtest:
+        start_time: 2017-01-01
+        end_time: 2020-08-01
+        account: 100000000
+        benchmark: *benchmark
+        exchange_kwargs:
+            limit_threshold: 0.095
+            deal_price: close
+            open_cost: 0.0005
+            close_cost: 0.0015
+            min_cost: 5
+task:
+    model:
+        class: TCN
+        module_path: qlib.contrib.model.pytorch_tcn
+        kwargs:
+            d_feat: 6
+            num_layers: 5
+            n_chans: 128
+            kernel_size: 3
+            dropout: 0.5
+            n_epochs: 200
+            lr: 1e-3
+            early_stop: 20
+            batch_size: 2000
+            metric: loss
+            loss: mse
+            optimizer: adam
+            GPU: 0
+    dataset:
+        class: DatasetH
+        module_path: qlib.data.dataset
+        kwargs:
+            handler:
+                class: Alpha360
+                module_path: qlib.contrib.data.handler
+                kwargs: *data_handler_config
+            segments:
+                train: [2008-01-01, 2014-12-31]
+                valid: [2015-01-01, 2016-12-31]
+                test: [2017-01-01, 2020-08-01]
+    record: 
+        - class: SignalRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            model: <MODEL>
+            dataset: <DATASET>
+        - class: SigAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            ana_long_short: False
+            ann_scaler: 252
+        - class: PortAnaRecord
+          module_path: qlib.workflow.record_temp
+          kwargs: 
+            config: *port_analysis_config

examples/benchmarks/TCTS/requirements.txt

@@ -1,4 +1,4 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/TCTS/workflow_config_tcts_Alpha360.yaml

@@ -95,4 +95,4 @@ task:
         - class: PortAnaRecord
           module_path: qlib.workflow.record_temp
           kwargs: 
-            config: *port_analysis_config
+            config: *port_analysis_config

examples/benchmarks/TFT/data_formatters/base.py

@@ -32,7 +32,7 @@ import abc
 import enum
 
 
-# Type defintions
+# Type definitions
 class DataTypes(enum.IntEnum):
     """Defines numerical types of each column."""
 

examples/benchmarks/TFT/libs/hyperparam_opt.py

@@ -254,9 +254,9 @@ class DistributedHyperparamOptManager(HyperparamOptManager):
           param_ranges: Discrete hyperparameter range for random search.
           fixed_params: Fixed model parameters per experiment.
           root_model_folder: Folder to store optimisation artifacts.
-          worker_number: Worker index definining which set of hyperparameters to
+          worker_number: Worker index defining which set of hyperparameters to
             test.
-          search_iterations: Maximum numer of random search iterations.
+          search_iterations: Maximum number of random search iterations.
           num_iterations_per_worker: How many iterations are handled per worker.
           clear_serialised_params: Whether to regenerate hyperparameter
             combinations.
@@ -330,7 +330,7 @@ class DistributedHyperparamOptManager(HyperparamOptManager):
         if os.path.exists(self.serialised_ranges_folder):
             df = pd.read_csv(self.serialised_ranges_path, index_col=0)
         else:
-            print("Unable to load - regenerating serach ranges instead")
+            print("Unable to load - regenerating search ranges instead")
             df = self.update_serialised_hyperparam_df()
 
         return df

examples/benchmarks/TFT/libs/tft_model.py

@@ -342,7 +342,7 @@ class TFTDataCache:
 
     @classmethod
     def contains(cls, key):
-        """Retuns boolean indicating whether key is present in cache."""
+        """Returns boolean indicating whether key is present in cache."""
 
         return key in cls._data_cache
 
@@ -1120,10 +1120,10 @@ class TemporalFusionTransformer:
         Args:
           df: Input dataframe
           return_targets: Whether to also return outputs aligned with predictions to
-            faciliate evaluation
+            facilitate evaluation
 
         Returns:
-          Input dataframe or tuple of (input dataframe, algined output dataframe).
+          Input dataframe or tuple of (input dataframe, aligned output dataframe).
         """
 
         data = self._batch_data(df)

examples/benchmarks/TFT/tft.py

@@ -209,7 +209,6 @@ class TFTModel(ModelFT):
         fixed_params = self.data_formatter.get_experiment_params()
         params = self.data_formatter.get_default_model_params()
 
-        # Wendi: 合并调优的参数和非调优的参数
         params = {**params, **fixed_params}
 
         if not os.path.exists(self.model_folder):
@@ -295,7 +294,7 @@ class TFTModel(ModelFT):
     def to_pickle(self, path: Union[Path, str]):
         """
         Tensorflow model can't be dumped directly.
-        So the data should be save seperatedly
+        So the data should be save separately
 
         **TODO**: Please implement the function to load the files
 

examples/benchmarks/TRA/README.md

@@ -57,7 +57,7 @@ And here are two ways to run the model:
   python example.py --config_file configs/config_alstm.yaml
   ```
 
-Here we trained TRA on a pretrained backbone model. Therefore we run `*_init.yaml` before TRA's scipts.
+Here we trained TRA on a pretrained backbone model. Therefore we run `*_init.yaml` before TRA's scripts.
 
 ### Results
 

examples/benchmarks/TRA/requirements.txt

@@ -1,5 +1,5 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 scikit_learn==0.23.2
 torch==1.7.0
 seaborn

examples/benchmarks/TRA/src/dataset.py

@@ -6,8 +6,7 @@ import torch
 import numpy as np
 import pandas as pd
 
-from qlib.utils import init_instance_by_config
-from qlib.data.dataset import DatasetH, DataHandler
+from qlib.data.dataset import DatasetH
 
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
@@ -95,7 +94,7 @@ class MTSDatasetH(DatasetH):
         shuffle=True,
         pin_memory=False,
         drop_last=False,
-        **kwargs
+        **kwargs,
     ):
 
         assert horizon > 0, "please specify `horizon` to avoid data leakage"
@@ -150,8 +149,15 @@ class MTSDatasetH(DatasetH):
 
     def _prepare_seg(self, slc, **kwargs):
         fn = _get_date_parse_fn(self._index[0][1])
-        start_date = fn(slc.start)
-        end_date = fn(slc.stop)
+
+        if isinstance(slc, slice):
+            start, stop = slc.start, slc.stop
+        elif isinstance(slc, (list, tuple)):
+            start, stop = slc
+        else:
+            raise NotImplementedError(f"This type of input is not supported")
+        start_date = fn(start)
+        end_date = fn(stop)
         obj = copy.copy(self)  # shallow copy
         # NOTE: Seriable will disable copy `self._data` so we manually assign them here
         obj._data = self._data

examples/benchmarks/TRA/src/model.py

@@ -124,13 +124,13 @@ class TRAModel(Model):
             loss = (pred - label).pow(2).mean()
 
             L = (all_preds.detach() - label[:, None]).pow(2)
-            L -= L.min(dim=-1, keepdim=True).values  # normalize & ensure postive input
+            L -= L.min(dim=-1, keepdim=True).values  # normalize & ensure positive input
 
             data_set.assign_data(index, L)  # save loss to memory
 
             if prob is not None:
                 P = sinkhorn(-L, epsilon=0.01)  # sample assignment matrix
-                lamb = self.lamb * (self.rho ** self.global_step)
+                lamb = self.lamb * (self.rho**self.global_step)
                 reg = prob.log().mul(P).sum(dim=-1).mean()
                 loss = loss - lamb * reg
 
@@ -165,7 +165,7 @@ class TRAModel(Model):
 
             L = (all_preds - label[:, None]).pow(2)
 
-            L -= L.min(dim=-1, keepdim=True).values  # normalize & ensure postive input
+            L -= L.min(dim=-1, keepdim=True).values  # normalize & ensure positive input
 
             data_set.assign_data(index, L)  # save loss to memory
 
@@ -484,7 +484,7 @@ class TRA(nn.Module):
 
     """Temporal Routing Adaptor (TRA)
 
-    TRA takes historical prediction erros & latent representation as inputs,
+    TRA takes historical prediction errors & latent representation as inputs,
     then routes the input sample to a specific predictor for training & inference.
 
     Args:
@@ -547,7 +547,7 @@ def evaluate(pred):
     score = pred.score
     label = pred.label
     diff = score - label
-    MSE = (diff ** 2).mean()
+    MSE = (diff**2).mean()
     MAE = (diff.abs()).mean()
     IC = score.corr(label)
     return {"MSE": MSE, "MAE": MAE, "IC": IC}

examples/benchmarks/TabNet/README.md

@@ -0,0 +1,3 @@
+# TabNet
+* Code: [https://github.com/dreamquark-ai/tabnet](https://github.com/dreamquark-ai/tabnet)
+* Paper: [TabNet: Attentive Interpretable Tabular Learning](https://arxiv.org/pdf/1908.07442.pdf).

examples/benchmarks/TabNet/requirements.txt

@@ -1,4 +1,4 @@
 pandas==1.1.2
-numpy==1.17.4
+numpy==1.21.0
 scikit_learn==0.23.2
 torch==1.7.0

examples/benchmarks/Transformer/README.md

@@ -0,0 +1,3 @@
+# Transformer
+* Code: [https://github.com/tensorflow/tensor2tensor](https://github.com/tensorflow/tensor2tensor)
+* Paper: [Attention is All you Need](https://proceedings.neurips.cc/paper/2017/file/3f5ee243547dee91fbd053c1c4a845aa-Paper.pdf).

examples/benchmarks/Transformer/requirements.txt

@@ -1,3 +1,3 @@
-numpy==1.17.4
+numpy==1.21.0
 pandas==1.1.2
 torch==1.2.0

examples/benchmarks/XGBoost/requirements.txt

@@ -1,3 +1,3 @@
-numpy==1.17.4
+numpy==1.21.0
 pandas==1.1.2
 xgboost==1.2.1

examples/benchmarks_dynamic/DDG-DA/README.md

@@ -0,0 +1,33 @@
+# Introduction
+This is the implementation of `DDG-DA` based on `Meta Controller` component provided by `Qlib`.
+
+Please refer to the paper for more details: *DDG-DA: Data Distribution Generation for Predictable Concept Drift Adaptation* [[arXiv](https://arxiv.org/abs/2201.04038)]
+
+
+# Background
+In many real-world scenarios, we often deal with streaming data that is sequentially collected over time. Due to the non-stationary nature of the environment, the streaming data distribution may change in unpredictable ways, which is known as concept drift. To handle concept drift, previous methods first detect when/where the concept drift happens and then adapt models to fit the distribution of the latest data. However, there are still many cases that some underlying factors of environment evolution are predictable, making it possible to model the future concept drift trend of the streaming data, while such cases are not fully explored in previous work.
+
+Therefore, we propose a novel method `DDG-DA`, that can effectively forecast the evolution of data distribution and improve the performance of models. Specifically, we first train a predictor to estimate the future data distribution, then leverage it to generate training samples, and finally train models on the generated data.
+
+# Dataset
+The data in the paper are private. So we conduct experiments on Qlib's public dataset.
+Though the dataset is different, the conclusion remains the same. By applying `DDG-DA`, users can see rising trends at the test phase both in the proxy models' ICs and the performances of the forecasting models.
+
+# Run the Code
+Users can try `DDG-DA` by running the following command:
+```bash
+    python workflow.py run_all
+```
+
+The default forecasting models are `Linear`. Users can choose other forecasting models by changing the `forecast_model` parameter when `DDG-DA` initializes. For example, users can try `LightGBM` forecasting models by running the following command:
+```bash
+    python workflow.py --forecast_model="gbdt" run_all
+```
+
+# Results
+The results of related methods in Qlib's public dataset can be found [here](../)
+
+# Requirements
+Here is the minimal hardware requirements to run the ``workflow.py`` of DDG-DA.
+* Memory: 45G
+* Disk: 4G

examples/benchmarks_dynamic/DDG-DA/requirements.txt

@@ -0,0 +1 @@
+torch==1.10.0 

examples/benchmarks_dynamic/DDG-DA/workflow.py

@@ -0,0 +1,259 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+from pathlib import Path
+from qlib.model.meta.task import MetaTask
+from qlib.contrib.meta.data_selection.model import MetaModelDS
+from qlib.contrib.meta.data_selection.dataset import InternalData, MetaDatasetDS
+from qlib.data.dataset.handler import DataHandlerLP
+
+import pandas as pd
+import fire
+import sys
+import pickle
+from qlib import auto_init
+from qlib.model.trainer import TrainerR
+from qlib.utils import init_instance_by_config
+from qlib.workflow import R
+from qlib.tests.data import GetData
+
+DIRNAME = Path(__file__).absolute().resolve().parent
+sys.path.append(str(DIRNAME.parent / "baseline"))
+from rolling_benchmark import RollingBenchmark  # NOTE: sys.path is changed for import RollingBenchmark
+
+
+class DDGDA:
+    """
+    please run `python workflow.py run_all` to run the full workflow of the experiment
+
+    **NOTE**
+    before running the example, please clean your previous results with following command
+    - `rm -r mlruns`
+    """
+
+    def __init__(self, sim_task_model="linear", forecast_model="linear"):
+        self.step = 20
+        # NOTE:
+        # the horizon must match the meaning in the base task template
+        self.horizon = 20
+        self.meta_exp_name = "DDG-DA"
+        self.sim_task_model = sim_task_model  # The model to capture the distribution of data.
+        self.forecast_model = forecast_model  # downstream forecasting models' type
+
+    def get_feature_importance(self):
+        # this must be lightGBM, because it needs to get the feature importance
+        rb = RollingBenchmark(model_type="gbdt")
+        task = rb.basic_task()
+
+        with R.start(experiment_name="feature_importance"):
+            model = init_instance_by_config(task["model"])
+            dataset = init_instance_by_config(task["dataset"])
+            model.fit(dataset)
+
+        fi = model.get_feature_importance()
+
+        # Because the model use numpy instead of dataframe for training lightgbm
+        # So the we must use following extra steps to get the right feature importance
+        df = dataset.prepare(segments=slice(None), col_set="feature", data_key=DataHandlerLP.DK_R)
+        cols = df.columns
+        fi_named = {cols[int(k.split("_")[1])]: imp for k, imp in fi.to_dict().items()}
+
+        return pd.Series(fi_named)
+
+    def dump_data_for_proxy_model(self):
+        """
+        Dump data for training meta model.
+        The meta model will be trained upon the proxy forecasting model.
+        This dataset is for the proxy forecasting model.
+        """
+        topk = 30
+        fi = self.get_feature_importance()
+        col_selected = fi.nlargest(topk)
+
+        rb = RollingBenchmark(model_type=self.sim_task_model)
+        task = rb.basic_task()
+        dataset = init_instance_by_config(task["dataset"])
+        prep_ds = dataset.prepare(slice(None), col_set=["feature", "label"], data_key=DataHandlerLP.DK_L)
+
+        feature_df = prep_ds["feature"]
+        label_df = prep_ds["label"]
+
+        feature_selected = feature_df.loc[:, col_selected.index]
+
+        feature_selected = feature_selected.groupby("datetime").apply(lambda df: (df - df.mean()).div(df.std()))
+        feature_selected = feature_selected.fillna(0.0)
+
+        df_all = {
+            "label": label_df.reindex(feature_selected.index),
+            "feature": feature_selected,
+        }
+        df_all = pd.concat(df_all, axis=1)
+        df_all.to_pickle(DIRNAME / "fea_label_df.pkl")
+
+        # dump data in handler format for aligning the interface
+        handler = DataHandlerLP(
+            data_loader={
+                "class": "qlib.data.dataset.loader.StaticDataLoader",
+                "kwargs": {"config": DIRNAME / "fea_label_df.pkl"},
+            }
+        )
+        handler.to_pickle(DIRNAME / "handler_proxy.pkl", dump_all=True)
+
+    @property
+    def _internal_data_path(self):
+        return DIRNAME / f"internal_data_s{self.step}.pkl"
+
+    def dump_meta_ipt(self):
+        """
+        Dump data for training meta model.
+        This function will dump the input data for meta model
+        """
+        # According to the experiments, the choice of the model type is very important for achieving good results
+        rb = RollingBenchmark(model_type=self.sim_task_model)
+        sim_task = rb.basic_task()
+
+        if self.sim_task_model == "gbdt":
+            sim_task["model"].setdefault("kwargs", {}).update({"early_stopping_rounds": None, "num_boost_round": 150})
+
+        exp_name_sim = f"data_sim_s{self.step}"
+
+        internal_data = InternalData(sim_task, self.step, exp_name=exp_name_sim)
+        internal_data.setup(trainer=TrainerR)
+
+        with self._internal_data_path.open("wb") as f:
+            pickle.dump(internal_data, f)
+
+    def train_meta_model(self):
+        """
+        training a meta model based on a simplified linear proxy model;
+        """
+
+        # 1) leverage the simplified proxy forecasting model to train meta model.
+        # - Only the dataset part is important, in current version of meta model will integrate the
+        rb = RollingBenchmark(model_type=self.sim_task_model)
+        sim_task = rb.basic_task()
+        proxy_forecast_model_task = {
+            # "model": "qlib.contrib.model.linear.LinearModel",
+            "dataset": {
+                "class": "qlib.data.dataset.DatasetH",
+                "kwargs": {
+                    "handler": f"file://{(DIRNAME / 'handler_proxy.pkl').absolute()}",
+                    "segments": {
+                        "train": ("2008-01-01", "2010-12-31"),
+                        "test": ("2011-01-01", sim_task["dataset"]["kwargs"]["segments"]["test"][1]),
+                    },
+                },
+            },
+            # "record": ["qlib.workflow.record_temp.SignalRecord"]
+        }
+        # the proxy_forecast_model_task will be used to create meta tasks.
+        # The test date of first task will be 2011-01-01. Each test segment will be about 20days
+        # The tasks include all training tasks and test tasks.
+
+        # 2) preparing meta dataset
+        kwargs = dict(
+            task_tpl=proxy_forecast_model_task,
+            step=self.step,
+            segments=0.62,  # keep test period consistent with the dataset yaml
+            trunc_days=1 + self.horizon,
+            hist_step_n=30,
+            fill_method="max",
+            rolling_ext_days=0,
+        )
+        # NOTE:
+        # the input of meta model (internal data) are shared between proxy model and final forecasting model
+        # but their task test segment are not aligned! It worked in my previous experiment.
+        # So the misalignment will not affect the effectiveness of the method.
+        with self._internal_data_path.open("rb") as f:
+            internal_data = pickle.load(f)
+        md = MetaDatasetDS(exp_name=internal_data, **kwargs)
+
+        # 3) train and logging meta model
+        with R.start(experiment_name=self.meta_exp_name):
+            R.log_params(**kwargs)
+            mm = MetaModelDS(step=self.step, hist_step_n=kwargs["hist_step_n"], lr=0.001, max_epoch=200, seed=43)
+            mm.fit(md)
+            R.save_objects(model=mm)
+
+    @property
+    def _task_path(self):
+        return DIRNAME / f"tasks_s{self.step}.pkl"
+
+    def meta_inference(self):
+        """
+        Leverage meta-model for inference:
+        - Given
+            - baseline tasks
+            - input for meta model(internal data)
+            - meta model (its learnt knowledge on proxy forecasting model is expected to transfer to normal forecasting model)
+        """
+        # 1) get meta model
+        exp = R.get_exp(experiment_name=self.meta_exp_name)
+        rec = exp.list_recorders(rtype=exp.RT_L)[0]
+        meta_model: MetaModelDS = rec.load_object("model")
+
+        # 2)
+        # we are transfer to knowledge of meta model to final forecasting tasks.
+        # Create MetaTaskDataset for the final forecasting tasks
+        # Aligning the setting of it to the MetaTaskDataset when training Meta model is necessary
+
+        # 2.1) get previous config
+        param = rec.list_params()
+        trunc_days = int(param["trunc_days"])
+        step = int(param["step"])
+        hist_step_n = int(param["hist_step_n"])
+        fill_method = param.get("fill_method", "max")
+
+        rb = RollingBenchmark(model_type=self.forecast_model)
+        task_l = rb.create_rolling_tasks()
+
+        # 2.2) create meta dataset for final dataset
+        kwargs = dict(
+            task_tpl=task_l,
+            step=step,
+            segments=0.0,  # all the tasks are for testing
+            trunc_days=trunc_days,
+            hist_step_n=hist_step_n,
+            fill_method=fill_method,
+            task_mode=MetaTask.PROC_MODE_TRANSFER,
+        )
+
+        with self._internal_data_path.open("rb") as f:
+            internal_data = pickle.load(f)
+        mds = MetaDatasetDS(exp_name=internal_data, **kwargs)
+
+        # 3) meta model make inference and get new qlib task
+        new_tasks = meta_model.inference(mds)
+        with self._task_path.open("wb") as f:
+            pickle.dump(new_tasks, f)
+
+    def train_and_eval_tasks(self):
+        """
+        Training the tasks generated by meta model
+        Then evaluate it
+        """
+        with self._task_path.open("rb") as f:
+            tasks = pickle.load(f)
+        rb = RollingBenchmark(rolling_exp="rolling_ds", model_type=self.forecast_model)
+        rb.train_rolling_tasks(tasks)
+        rb.ens_rolling()
+        rb.update_rolling_rec()
+
+    def run_all(self):
+        # 1) file: handler_proxy.pkl
+        self.dump_data_for_proxy_model()
+        # 2)
+        # file: internal_data_s20.pkl
+        # mlflow: data_sim_s20, models for calculating meta_ipt
+        self.dump_meta_ipt()
+        # 3) meta model will be stored in `DDG-DA`
+        self.train_meta_model()
+        # 4) new_tasks are saved in "tasks_s20.pkl" (reweighter is added)
+        self.meta_inference()
+        # 5) load the saved tasks and train model
+        self.train_and_eval_tasks()
+
+
+if __name__ == "__main__":
+    GetData().qlib_data(exists_skip=True)
+    auto_init()
+    fire.Fire(DDGDA)

examples/benchmarks_dynamic/README.md

@@ -0,0 +1,18 @@
+# Introduction
+Due to the non-stationary nature of the environment of the financial market, the data distribution may change in different periods, which makes the performance of models build on training data decays in the future test data.
+So adapting the forecasting models/strategies to market dynamics is very important to the model/strategies' performance.
+
+The table below shows the performances of different solutions on different forecasting models.
+
+## Alpha158 dataset
+
+| Model Name       | Dataset | IC | ICIR | Rank IC | Rank ICIR | Annualized Return | Information Ratio | Max Drawdown |
+|------------------|---------|----|------|---------|-----------|-------------------|-------------------|--------------|
+| RR[Linear]       |Alpha158 |0.088|0.570|0.102    |0.622      |0.077              |1.175              |-0.086        |
+| DDG-DA[Linear]   |Alpha158 |0.093|0.622|0.106    |0.670      |0.085              |1.213              |-0.093        |
+| RR[LightGBM]     |Alpha158 |0.079|0.566|0.088    |0.592      |0.075              |1.226              |-0.096        |
+| DDG-DA[LightGBM] |Alpha158 |0.084|0.639|0.093    |0.664      |0.099              |1.442              |-0.071        |
+
+- The label horizon of the `Alpha158` dataset is set to 20.
+- The rolling time intervals are set to 20 trading days.
+- The test rolling periods are from January 2017 to August 2020.

examples/benchmarks_dynamic/baseline/README.md

@@ -0,0 +1,15 @@
+# Introduction
+
+This is the framework of periodically Rolling Retrain (RR) forecasting models. RR adapts to market dynamics by utilizing the up-to-date data periodically.
+
+## Run the Code
+Users can try RR by running the following command:
+```bash
+    python rolling_benchmark.py run_all
+```
+
+The default forecasting models are `Linear`. Users can choose other forecasting models by changing the `model_type` parameter.
+For example, users can try `LightGBM` forecasting models by running the following command:
+```bash
+    python rolling_benchmark.py --model_type="gbdt" run_all
+```

examples/benchmarks_dynamic/baseline/rolling_benchmark.py

@@ -0,0 +1,114 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+from qlib.model.ens.ensemble import RollingEnsemble
+from qlib.utils import init_instance_by_config
+import fire
+import yaml
+from qlib import auto_init
+from pathlib import Path
+from tqdm.auto import tqdm
+from qlib.model.trainer import TrainerR
+from qlib.workflow import R
+from qlib.tests.data import GetData
+
+DIRNAME = Path(__file__).absolute().resolve().parent
+from qlib.workflow.task.gen import task_generator, RollingGen
+from qlib.workflow.task.collect import RecorderCollector
+from qlib.workflow.record_temp import PortAnaRecord, SigAnaRecord
+
+
+class RollingBenchmark:
+    """
+    **NOTE**
+    before running the example, please clean your previous results with following command
+    - `rm -r mlruns`
+
+    """
+
+    def __init__(self, rolling_exp="rolling_models", model_type="linear") -> None:
+        self.step = 20
+        self.horizon = 20
+        self.rolling_exp = rolling_exp
+        self.model_type = model_type
+
+    def basic_task(self):
+        """For fast training rolling"""
+        if self.model_type == "gbdt":
+            conf_path = DIRNAME.parent.parent / "benchmarks" / "LightGBM" / "workflow_config_lightgbm_Alpha158.yaml"
+            # dump the processed data on to disk for later loading to speed up the processing
+            h_path = DIRNAME / "lightgbm_alpha158_handler_horizon{}.pkl".format(self.horizon)
+        elif self.model_type == "linear":
+            conf_path = DIRNAME.parent.parent / "benchmarks" / "Linear" / "workflow_config_linear_Alpha158.yaml"
+            h_path = DIRNAME / "linear_alpha158_handler_horizon{}.pkl".format(self.horizon)
+        else:
+            raise AssertionError("Model type is not supported!")
+        with conf_path.open("r") as f:
+            conf = yaml.safe_load(f)
+
+        # modify dataset horizon
+        conf["task"]["dataset"]["kwargs"]["handler"]["kwargs"]["label"] = [
+            "Ref($close, -{}) / Ref($close, -1) - 1".format(self.horizon + 1)
+        ]
+
+        task = conf["task"]
+
+        if not h_path.exists():
+            h_conf = task["dataset"]["kwargs"]["handler"]
+            h = init_instance_by_config(h_conf)
+            h.to_pickle(h_path, dump_all=True)
+
+        task["dataset"]["kwargs"]["handler"] = f"file://{h_path}"
+        task["record"] = ["qlib.workflow.record_temp.SignalRecord"]
+        return task
+
+    def create_rolling_tasks(self):
+        task = self.basic_task()
+        task_l = task_generator(
+            task, RollingGen(step=self.step, trunc_days=self.horizon + 1)
+        )  # the last two days should be truncated to avoid information leakage
+        return task_l
+
+    def train_rolling_tasks(self, task_l=None):
+        if task_l is None:
+            task_l = self.create_rolling_tasks()
+        trainer = TrainerR(experiment_name=self.rolling_exp)
+        trainer(task_l)
+
+    COMB_EXP = "rolling"
+
+    def ens_rolling(self):
+        rc = RecorderCollector(
+            experiment=self.rolling_exp,
+            artifacts_key=["pred", "label"],
+            process_list=[RollingEnsemble()],
+            # rec_key_func=lambda rec: (self.COMB_EXP, rec.info["id"]),
+            artifacts_path={"pred": "pred.pkl", "label": "label.pkl"},
+        )
+        res = rc()
+        with R.start(experiment_name=self.COMB_EXP):
+            R.log_params(exp_name=self.rolling_exp)
+            R.save_objects(**{"pred.pkl": res["pred"], "label.pkl": res["label"]})
+
+    def update_rolling_rec(self):
+        """
+        Evaluate the combined rolling results
+        """
+        for rid, rec in R.list_recorders(experiment_name=self.COMB_EXP).items():
+            for rt_cls in SigAnaRecord, PortAnaRecord:
+                rt = rt_cls(recorder=rec, skip_existing=True)
+                rt.generate()
+        print(f"Your evaluation results can be found in the experiment named `{self.COMB_EXP}`.")
+
+    def run_all(self):
+        # the results will be  save in mlruns.
+        # 1) each rolling task is saved in rolling_models
+        self.train_rolling_tasks()
+        # 2) combined rolling tasks and evaluation results are saved in rolling
+        self.ens_rolling()
+        self.update_rolling_rec()
+
+
+if __name__ == "__main__":
+    GetData().qlib_data(exists_skip=True)
+    auto_init()
+    fire.Fire(RollingBenchmark)

examples/data_demo/README.md

@@ -0,0 +1,2 @@
+# Introduction
+The examples in this folder try to demonstrate some common usage of data-related modules of Qlib

examples/data_demo/data_cache_demo.py

@@ -0,0 +1,53 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+"""
+    The motivation of this demo
+    - To show the data modules of Qlib is Serializable, users can dump processed data to disk to avoid duplicated data preprocessing
+"""
+
+from copy import deepcopy
+from pathlib import Path
+import pickle
+from pprint import pprint
+import subprocess
+import yaml
+from qlib.log import TimeInspector
+
+from qlib import init
+from qlib.data.dataset.handler import DataHandlerLP
+from qlib.utils import init_instance_by_config
+
+# For general purpose, we use relative path
+DIRNAME = Path(__file__).absolute().resolve().parent
+
+if __name__ == "__main__":
+    init()
+
+    config_path = DIRNAME.parent / "benchmarks/LightGBM/workflow_config_lightgbm_Alpha158.yaml"
+
+    # 1) show original time
+    with TimeInspector.logt("The original time without handler cache:"):
+        subprocess.run(f"qrun {config_path}", shell=True)
+
+    # 2) dump handler
+    task_config = yaml.safe_load(config_path.open())
+    hd_conf = task_config["task"]["dataset"]["kwargs"]["handler"]
+    pprint(hd_conf)
+    hd: DataHandlerLP = init_instance_by_config(hd_conf)
+    hd_path = DIRNAME / "handler.pkl"
+    hd.to_pickle(hd_path, dump_all=True)
+
+    # 3) create new task with handler cache
+    new_task_config = deepcopy(task_config)
+    new_task_config["task"]["dataset"]["kwargs"]["handler"] = f"file://{hd_path}"
+    new_task_config["sys"] = {"path": [str(config_path.parent.resolve())]}
+    new_task_path = DIRNAME / "new_task.yaml"
+    print("The location of the new task", new_task_path)
+
+    # save new task
+    with new_task_path.open("w") as f:
+        yaml.safe_dump(new_task_config, f, indent=4, sort_keys=False)
+
+    # 4) train model with new task
+    with TimeInspector.logt("The time for task with handler cache:"):
+        subprocess.run(f"qrun {new_task_path}", shell=True)

examples/data_demo/data_mem_resuse_demo.py

@@ -0,0 +1,59 @@
+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+"""
+    The motivation of this demo
+    - To show the data modules of Qlib is Serializable, users can dump processed data to disk to avoid duplicated data preprocessing
+"""
+
+from copy import deepcopy
+from pathlib import Path
+import pickle
+from pprint import pprint
+import subprocess
+
+import yaml
+
+from qlib import init
+from qlib.data.dataset.handler import DataHandlerLP
+from qlib.log import TimeInspector
+from qlib.model.trainer import task_train
+from qlib.utils import init_instance_by_config
+
+# For general purpose, we use relative path
+DIRNAME = Path(__file__).absolute().resolve().parent
+
+if __name__ == "__main__":
+    init()
+
+    repeat = 2
+    exp_name = "data_mem_reuse_demo"
+
+    config_path = DIRNAME.parent / "benchmarks/LightGBM/workflow_config_lightgbm_Alpha158.yaml"
+    task_config = yaml.safe_load(config_path.open())
+
+    # 1) without using processed data in memory
+    with TimeInspector.logt("The original time without reusing processed data in memory:"):
+        for i in range(repeat):
+            task_train(task_config["task"], experiment_name=exp_name)
+
+    # 2) prepare processed data in memory.
+    hd_conf = task_config["task"]["dataset"]["kwargs"]["handler"]
+    pprint(hd_conf)
+    hd: DataHandlerLP = init_instance_by_config(hd_conf)
+
+    # 3) with reusing processed data in memory
+    new_task = deepcopy(task_config["task"])
+    new_task["dataset"]["kwargs"]["handler"] = hd
+    print(new_task)
+
+    with TimeInspector.logt("The time with reusing processed data in memory:"):
+        # this will save the time to reload and process data from disk(in `DataHandlerLP`)
+        # It still takes a lot of time in the backtest phase
+        for i in range(repeat):
+            task_train(new_task, experiment_name=exp_name)
+
+    # 4) User can change other parts exclude processed data in memory(handler)
+    new_task = deepcopy(task_config["task"])
+    new_task["dataset"]["kwargs"]["segments"]["train"] = ("20100101", "20131231")
+    with TimeInspector.logt("The time with reusing processed data in memory:"):
+        task_train(new_task, experiment_name=exp_name)

examples/highfreq/README.md

@@ -1,15 +1,20 @@
-# High-Frequency Dataset
+# Introduction
+This folder contains 2 examples
+- A high-frequency dataset example
+- An example of predicting the price trend in high-frequency data
+
+## High-Frequency Dataset
 
 This dataset is an example for RL high frequency trading.
 
-## Get High-Frequency Data
+### Get High-Frequency Data
 
 Get high-frequency data by running the following command:
 ```bash
     python workflow.py get_data
 ```
 
-## Dump & Reload & Reinitialize the Dataset
+### Dump & Reload & Reinitialize the Dataset
 
 
 The High-Frequency Dataset is implemented as `qlib.data.dataset.DatasetH` in the `workflow.py`. `DatatsetH` is the subclass of [`qlib.utils.serial.Serializable`](https://qlib.readthedocs.io/en/latest/advanced/serial.html), whose state can be dumped in or loaded from disk in `pickle` format.
@@ -27,9 +32,10 @@ Run the example by running the following command:
     python workflow.py dump_and_load_dataset
 ```
 
-## Benchmarks Performance
-### Signal Test
-Here are the results of signal test for benchmark models. We will keep updating benchmark models in future.
+## Benchmarks Performance (predicting the price trend in high-frequency data)
+
+Here are the results of models for predicting the price trend in high-frequency data. We will keep updating benchmark models in future.
+
 | Model Name | Dataset | IC | ICIR | Rank IC | Rank ICIR | Long precision| Short Precision | Long-Short Average Return | Long-Short Average Sharpe |
 |---|---|---|---|---|---|---|---|---|---|
-| LightGBM | Alpha158 | 0.3042±0.00 | 1.5372±0.00| 0.3117±0.00 | 1.6258±0.00 | 0.6720±0.00 | 0.6870±0.00 | 0.000769±0.00 | 1.0190±0.00 |
+| LightGBM | Alpha158 | 0.0349±0.00 | 0.3805±0.00| 0.0435±0.00 | 0.4724±0.00 | 0.5111±0.00 | 0.5428±0.00 | 0.000074±0.00 | 0.2677±0.00 |

examples/highfreq/highfreq_ops.py

@@ -150,7 +150,7 @@ class Cut(ElemOperator):
         self.l = l
         self.r = r
         if (self.l is not None and self.l <= 0) or (self.r is not None and self.r >= 0):
-            raise ValueError("Cut operator l shoud > 0 and r should < 0")
+            raise ValueError("Cut operator l should > 0 and r should < 0")
 
         super(Cut, self).__init__(feature)
 

examples/highfreq/highfreq_processor.py

@@ -1,5 +1,6 @@
 import numpy as np
 import pandas as pd
+from qlib.constant import EPS
 from qlib.data.dataset.processor import Processor
 from qlib.data.dataset.utils import fetch_df_by_index
 
@@ -27,7 +28,7 @@ class HighFreqNorm(Processor):
                 part_values = np.log1p(part_values)
             self.feature_med[name] = np.nanmedian(part_values)
             part_values = part_values - self.feature_med[name]
-            self.feature_std[name] = np.nanmedian(np.absolute(part_values)) * 1.4826 + 1e-12
+            self.feature_std[name] = np.nanmedian(np.absolute(part_values)) * 1.4826 + EPS
             part_values = part_values / self.feature_std[name]
             self.feature_vmax[name] = np.nanmax(part_values)
             self.feature_vmin[name] = np.nanmin(part_values)

examples/highfreq/workflow.py

@@ -5,7 +5,8 @@ import fire
 
 import qlib
 import pickle
-from qlib.config import REG_CN, HIGH_FREQ_CONFIG
+from qlib.constant import REG_CN
+from qlib.config import HIGH_FREQ_CONFIG
 
 from qlib.utils import init_instance_by_config
 from qlib.data.dataset.handler import DataHandlerLP
@@ -82,7 +83,7 @@ class HighfreqWorkflow:
 
     def _init_qlib(self):
         """initialize qlib"""
-        # use yahoo_cn_1min data
+        # use cn_data_1min data
         QLIB_INIT_CONFIG = {**HIGH_FREQ_CONFIG, **self.SPEC_CONF}
         provider_uri = QLIB_INIT_CONFIG.get("provider_uri")
         GetData().qlib_data(target_dir=provider_uri, interval="1min", region=REG_CN, exists_skip=True)

examples/highfreq/workflow_config_High_Freq_Tree_Alpha158.yaml

@@ -59,7 +59,7 @@ task:
     record: 
         - class: "SignalRecord"
           module_path: "qlib.workflow.record_temp"
-          kwargs: 
+          kwargs: {}
         - class: "HFSignalRecord"
           module_path: "qlib.workflow.record_temp"
           kwargs: {}

examples/hyperparameter/LightGBM/hyperparameter_158.py

@@ -1,6 +1,6 @@
 import qlib
 import optuna
-from qlib.config import REG_CN
+from qlib.constant import REG_CN
 from qlib.utils import init_instance_by_config
 from qlib.tests.config import CSI300_DATASET_CONFIG
 from qlib.tests.data import GetData