adelorenzo/stable
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

expose environment reward parameters to the user config

Browse Source
This commit is contained in:
robcaulk
2022-08-21 20:33:09 +02:00
parent d88a0dbf82
commit 29f0e01c4a
5 changed files with 28 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
freqtrade/freqai/prediction_models/ReinforcementLearner.py
View File

@@ -57,26 +57,20 @@ class MyRLEnv(Base5ActionRLEnv):
# close long
if action == Actions.Long_exit.value and self._position == Positions.Long:
last_trade_price = self.add_buy_fee(self.prices.iloc[self._last_trade_tick].open)
current_price = self.add_sell_fee(self.prices.iloc[self._current_tick].open)
return float(np.log(current_price) - np.log(last_trade_price))
if action == Actions.Long_exit.value and self._position == Positions.Long:
if self.close_trade_profit[-1] > self.profit_aim * self.rr:
last_trade_price = self.add_buy_fee(self.prices.iloc[self._last_trade_tick].open)
current_price = self.add_sell_fee(self.prices.iloc[self._current_tick].open)
return float((np.log(current_price) - np.log(last_trade_price)) * 2)
last_trade_price = self.add_entry_fee(self.prices.iloc[self._last_trade_tick].open)
current_price = self.add_exit_fee(self.prices.iloc[self._current_tick].open)
factor = 1
if self.close_trade_profit and self.close_trade_profit[-1] > self.profit_aim * self.rr:
factor = self.rl_config['model_reward_parameters'].get('win_reward_factor', 2)
return float((np.log(current_price) - np.log(last_trade_price)) * factor)
# close short
if action == Actions.Short_exit.value and self._position == Positions.Short:
last_trade_price = self.add_sell_fee(self.prices.iloc[self._last_trade_tick].open)
current_price = self.add_buy_fee(self.prices.iloc[self._current_tick].open)
return float(np.log(last_trade_price) - np.log(current_price))
if action == Actions.Short_exit.value and self._position == Positions.Short:
if self.close_trade_profit[-1] > self.profit_aim * self.rr:
last_trade_price = self.add_sell_fee(self.prices.iloc[self._last_trade_tick].open)
current_price = self.add_buy_fee(self.prices.iloc[self._current_tick].open)
return float((np.log(last_trade_price) - np.log(current_price)) * 2)
last_trade_price = self.add_exit_fee(self.prices.iloc[self._last_trade_tick].open)
current_price = self.add_entry_fee(self.prices.iloc[self._current_tick].open)
factor = 1
if self.close_trade_profit and self.close_trade_profit[-1] > self.profit_aim * self.rr:
factor = self.rl_config['model_reward_parameters'].get('win_reward_factor', 2)
return float(np.log(last_trade_price) - np.log(current_price) * factor)
return 0.

4
freqtrade/freqai/prediction_models/ReinforcementLearner_multiproc.py
View File

@@ -62,12 +62,12 @@ class ReinforcementLearner_multiproc(BaseReinforcementLearningModel):
env_id = "train_env"
num_cpu = int(self.freqai_info["data_kitchen_thread_count"] / 2)
self.train_env = SubprocVecEnv([make_env(env_id, i, 1, train_df, prices_train,
self.reward_params, self.CONV_WIDTH) for i
self.reward_params, self.CONV_WIDTH, config=self.config) for i
in range(num_cpu)])
eval_env_id = 'eval_env'
self.eval_env = SubprocVecEnv([make_env(eval_env_id, i, 1, test_df, prices_test,
self.reward_params, self.CONV_WIDTH, monitor=True) for i
self.reward_params, self.CONV_WIDTH, monitor=True, config=self.config) for i
in range(num_cpu)])
self.eval_callback = EvalCallback(self.eval_env, deterministic=True,
render=False, eval_freq=eval_freq,