Footnote:
Nach Informationen von SSRN wurde die ursprüngliche Fassung des Dokuments October 19, 2022 erstellt
Description:
Deep reinforcement learning (DRL) has shown great potential in financial tasks. However, existing works optimistically reported profitable results through backtesting, suffering the \textit{look-ahead bias} and \textit{overfitting} issues. Therefore, these promising research results do not necessarily lead to a good performance in real-world markets. In this paper, we provide the first empirical demonstration of the stock paper trading task using deep reinforcement learning. First, we employ a ``training-validation-trading" pipeline where an agent always learns the latest market daily data; meanwhile, the pipeline helps avoid information leakage. Then, we show that the trained DRL agent outperforms a conventional machine learning method (random forest) and the Dow Jones Industrial Average (DJIA) index during stock paper trading. Our codes are available online at: \url{https://github.com/AI4Finance-Foundation/FinRL-Meta}