liuyu/qlib
1
0
Fork 0
mirror of https://github.com/microsoft/qlib.git synced 2026-07-10 06:20:57 +08:00
Code Activity

Refine Qlib RL data format (#1480)

Browse Source 
* wip

* wip

* wip

* Fix naming errors

* Backtest test passed

* Why training stuck?

* Minor

* Refine train configs

* Use dummy in training

* Remove pickle_dataframe

* CI

* CI

* Add more strict condition to filter orders

* Pass test

* Add TODO in example

---------

Co-authored-by: Young <afe.young@gmail.com>
This commit is contained in:
Huoran Li
2023-04-26 21:14:30 +08:00
committed by GitHub
parent 46264dfec9
commit 7f1e8c5206
17 changed files with 237 additions and 250 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
tests/rl/test_saoe_simple.py
View File

@@ -31,7 +31,6 @@ FEATURE_DATA_DIR = DATA_DIR / "processed"
ORDER_DIR = DATA_DIR / "order" / "valid_bidir"
CN_DATA_DIR = DATA_ROOT_DIR / "cn"
CN_BACKTEST_DATA_DIR = CN_DATA_DIR / "backtest"
CN_FEATURE_DATA_DIR = CN_DATA_DIR / "processed"
CN_ORDER_DIR = CN_DATA_DIR / "order" / "test"
CN_POLICY_WEIGHTS_DIR = CN_DATA_DIR / "weights"
@@ -49,7 +48,7 @@ def test_pickle_data_inspect():
def test_simulator_first_step():
order = Order("AAL", 30.0, 0, pd.Timestamp("2013-12-11 00:00:00"), pd.Timestamp("2013-12-11 23:59:59"))
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
state = simulator.get_state()
assert state.cur_time == pd.Timestamp("2013-12-11 09:30:00")
assert state.position == 30.0
@@ -83,7 +82,7 @@ def test_simulator_first_step():
def test_simulator_stop_twap():
order = Order("AAL", 13.0, 0, pd.Timestamp("2013-12-11 00:00:00"), pd.Timestamp("2013-12-11 23:59:59"))
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
for _ in range(13):
simulator.step(1.0)
@@ -106,10 +105,10 @@ def test_simulator_stop_early():
order = Order("AAL", 1.0, 1, pd.Timestamp("2013-12-11 00:00:00"), pd.Timestamp("2013-12-11 23:59:59"))
with pytest.raises(ValueError):
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
simulator.step(2.0)
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
simulator.step(1.0)
with pytest.raises(AssertionError):
@@ -119,7 +118,7 @@ def test_simulator_stop_early():
def test_simulator_start_middle():
order = Order("AAL", 15.0, 1, pd.Timestamp("2013-12-11 10:15:00"), pd.Timestamp("2013-12-11 15:44:59"))
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
assert len(simulator.ticks_for_order) == 330
assert simulator.cur_time == pd.Timestamp("2013-12-11 10:15:00")
simulator.step(2.0)
@@ -138,7 +137,7 @@ def test_simulator_start_middle():
def test_interpreter():
order = Order("AAL", 15.0, 1, pd.Timestamp("2013-12-11 10:15:00"), pd.Timestamp("2013-12-11 15:44:59"))
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
assert len(simulator.ticks_for_order) == 330
assert simulator.cur_time == pd.Timestamp("2013-12-11 10:15:00")
@@ -219,7 +218,7 @@ def test_network_sanity():
# we won't check the correctness of networks here
order = Order("AAL", 15.0, 1, pd.Timestamp("2013-12-11 9:30:00"), pd.Timestamp("2013-12-11 15:59:59"))
simulator = SingleAssetOrderExecutionSimple(order, BACKTEST_DATA_DIR)
simulator = SingleAssetOrderExecutionSimple(order, DATA_DIR)
assert len(simulator.ticks_for_order) == 390
class EmulateEnvWrapper(NamedTuple):
@@ -259,7 +258,7 @@ def test_twap_strategy(finite_env_type):
csv_writer = CsvWriter(Path(__file__).parent / ".output")
backtest(
partial(SingleAssetOrderExecutionSimple, data_dir=BACKTEST_DATA_DIR, ticks_per_step=30),
partial(SingleAssetOrderExecutionSimple, data_dir=DATA_DIR, ticks_per_step=30),
state_interp,
action_interp,
orders,
@@ -290,7 +289,7 @@ def test_cn_ppo_strategy():
csv_writer = CsvWriter(Path(__file__).parent / ".output")
backtest(
partial(SingleAssetOrderExecutionSimple, data_dir=CN_BACKTEST_DATA_DIR, ticks_per_step=30),
partial(SingleAssetOrderExecutionSimple, data_dir=CN_DATA_DIR, ticks_per_step=30),
state_interp,
action_interp,
orders,
@@ -319,7 +318,7 @@ def test_ppo_train():
policy = PPO(network, state_interp.observation_space, action_interp.action_space, 1e-4)
train(
partial(SingleAssetOrderExecutionSimple, data_dir=CN_BACKTEST_DATA_DIR, ticks_per_step=30),
partial(SingleAssetOrderExecutionSimple, data_dir=CN_DATA_DIR, ticks_per_step=30),
state_interp,
action_interp,
orders,