Black

qlib/rl/contrib/backtest.py

@@ -278,7 +278,7 @@ def single_with_collect_data_loop(
         **{
             "codes": stocks,
             "freq": backtest_config["data_granularity"],
-        }
+        },
     }
 
     strategy, executor = get_strategy_executor(

qlib/rl/contrib/naive_config_parser.py

@@ -192,13 +192,15 @@ class TrainingConfigParser:
                 "qlib": {
                     "provider_uri_1min": config["qlib"]["provider_uri_1min"],
                 },
-                "exchange": exchange_config
+                "exchange": exchange_config,
             }
 
             return ret_config
 
+
 if __name__ == "__main__":
     parser = TrainingConfigParser("/home/huoran/exp_configs/amc4th_training_refined.yml")
 
     from pprint import pprint
+
     pprint(parser.parse())

qlib/rl/contrib/train.py

@@ -49,7 +49,7 @@ def get_executor_config(freq: int) -> dict:
                             "track_data": True,
                             "trade_type": "serial",
                             "verbose": False,
-                        }
+                        },
                     },
                     "inner_strategy": {
                         "class": "TWAPStrategy",
@@ -58,7 +58,7 @@ def get_executor_config(freq: int) -> dict:
                     },
                     "time_per_step": "30min",
                     "track_data": True,
-                }
+                },
             },
             "inner_strategy": {
                 "class": "ProxySAOEStrategy",
@@ -67,7 +67,7 @@ def get_executor_config(freq: int) -> dict:
             },
             "time_per_step": "1day",
             "track_data": True,
-        }
+        },
     }
 
 
@@ -168,6 +168,7 @@ def train_and_test(
 
     sim_type = simulator_config["type"]
     if sim_type == "simple":
+
         def _simulator_factory(order: Order) -> SingleAssetOrderExecutionSimple:
             simulator = SingleAssetOrderExecutionSimple(
                 order=order,
@@ -178,6 +179,7 @@ def train_and_test(
                 vol_threshold=simulator_config["vol_limit"],
             )
             return simulator
+
     elif sim_type == "full":
         init_qlib(simulator_config["qlib"])
         executor_config = get_executor_config(freq)
@@ -287,16 +289,13 @@ def main(config: dict, run_training: bool, run_backtest: bool) -> None:
         network_config = config["policy"]["network"]
         network_config["kwargs"] = {
             **network_config.get("kwargs", {}),
-            **{"obs_space": state_interpreter.observation_space}
+            **{"obs_space": state_interpreter.observation_space},
         }
         additional_policy_kwargs["network"] = init_instance_by_config(network_config)
 
     # Create policy
     policy_config = config["policy"]["policy"]
-    policy_config["kwargs"] = {
+    policy_config["kwargs"] = {**policy_config.get("kwargs", {}), **additional_policy_kwargs}
-        **policy_config.get("kwargs", {}),
-        **additional_policy_kwargs
-    }
     policy: BasePolicy = init_instance_by_config(policy_config)
 
     use_cuda = config["runtime"]["use_cuda"]

qlib/rl/contrib/train_onpolicy_full_simulation.py

@@ -50,7 +50,7 @@ def get_executor_config(data_granularity: int = 1) -> dict:
                             "track_data": True,
                             "trade_type": "serial",
                             "verbose": False,
-                        }
+                        },
                     },
                     "inner_strategy": {
                         "class": "TWAPStrategy",
@@ -59,7 +59,7 @@ def get_executor_config(data_granularity: int = 1) -> dict:
                     },
                     "time_per_step": "30min",
                     "track_data": True,
-                }
+                },
             },
             "inner_strategy": {
                 "class": "ProxySAOEStrategy",
@@ -68,7 +68,7 @@ def get_executor_config(data_granularity: int = 1) -> dict:
             },
             "time_per_step": "1day",
             "track_data": True,
-        }
+        },
     }
 
 

qlib/rl/data/native.py

@@ -149,6 +149,7 @@ def _load_handler_pickle(path: str) -> object:
         obj = pickle.load(fstream)
     return obj
 
+
 class HandlerIntradayProcessedData(BaseIntradayProcessedData):
     """Subclass of IntradayProcessedData. Used to handle handler (bin format) style data."""
 
@@ -189,7 +190,12 @@ def load_handler_intraday_processed_data(
     backtest: bool = False,
 ) -> HandlerIntradayProcessedData:
     return HandlerIntradayProcessedData(
-        data_dir, stock_id, date, feature_columns_today, feature_columns_yesterday, backtest,
+        data_dir,
+        stock_id,
+        date,
+        feature_columns_today,
+        feature_columns_yesterday,
+        backtest,
     )
 
 

qlib/rl/utils/profiling.py

@@ -2,6 +2,7 @@ import time
 from contextlib import contextmanager
 from line_profiler import LineProfiler
 
+
 @contextmanager
 def simple_perf(desc: str = ""):
     s = time.perf_counter()
@@ -17,4 +18,5 @@ def lprofile(func):
         res = lpw(*args, **kwargs)
         lp.print_stats()
         return res
+
     return wrapper