FinRL icon indicating copy to clipboard operation
FinRL copied to clipboard

FinRL: Financial Reinforcement Learning. 🔥

Results 250 FinRL issues
Sort by recently updated
recently updated
newest added

I tried to solve the error for the NaN value according to this [reference](https://github.com/AI4Finance-Foundation/FinRL/issues/353#issuecomment-975188649) but after the preprocessing is done correctly I got this error in training. Can anyone please...

help_wanted

trained_a2c = agent.train_model(model=model_a2c, tb_log_name='a2c', total_timesteps=100000) Logging to tensorboard_log/a2c/a2c_3 --------------------------------------------------------------------------- ValueError Traceback (most recent call last) /tmp/ipykernel_37/2087801270.py in 1 trained_a2c = agent.train_model(model=model_a2c, 2 tb_log_name='a2c', ----> 3 total_timesteps=100000) /opt/conda/lib/python3.7/site-packages/finrl/drl_agents/stablebaselines3/models.py in train_model(self, model,...

bug

Referencing the env_stocktrading_stoploss.py , it states that "Reward at any step is given as follows **r_i = (sum(cash, asset_value) + additional_reward - total_penalty - initial_cash) / initial_cash / days_elapsed"**. But...

discussion

print("==============Compare to DJIA===========") %matplotlib inline # S&P 500: ^GSPC # Dow Jones Index: ^DJI # NASDAQ 100: ^NDX backtest_plot(df_account_value, baseline_ticker = '^DJI', baseline_start = '2019-01-01', baseline_end = '2021-01-01') Exception in...

bug

I have tried several kinds of models such as ddpg, ppo. However, when I check the training log, It seems the agent have lost almost all initial amounts. environment/ |...

bug

I believe it may help agent avoid getting negative asset when agent cannot afford transaction cost in the extreme

Closes #220 For review: @BruceYanghy, @XiaoYangLiu-FinRL

Dear all, Hope this pull requests finds you all well and in good health, Sorry for the delay, as requested this is a smaller pull request, with pep8 styling and...

Hello all, so I fixed the minor bugs working with the current repository, I will try to make this as well documented as I possibly can as per @spencerR1992's request,...