Deep Reinforcement Learning for Cryptocurrency Trading: Practical Approach to Address Backtest Overfitting

DOI: 10.48550/arXiv.2209.05559

ABSTRACT: Designing profitable and reliable trading strategies is challenging in the highly volatile cryptocurrency market. Existing works applied deep reinforcement learning methods and optimistically reported increased profits in backtesting, which may suffer from the false positive issue due to overfitting. In this paper, we propose a practical approach to address backtest overfitting for cryptocurrency trading using deep reinforcement learning. First, we formulate the detection of backtest overfitting as a hypothesis test. Then, we train the DRL agents, estimate the probability of overfitting, and reject the overfitted agents, increasing the chance of good trading performance. Finally, on 10 cryptocurrencies over a testing period from 05/01/2022 to 06/27/2022 (during which the crypto market crashed two times), we show that the less overfitted deep reinforcement learning agents have a higher return than that of more overfitted agents, an equal weight strategy, and the S&P DBM Index (market benchmark), offering confidence in possible deployment to a real market.

– Less overfitted deep reinforcement learning agents have higher returns.
– The proposed approach offers confidence in possible deployment to a real market.

– Less overfitted deep reinforcement learning agents have higher returns.
– The proposed approach offers confidence in possible deployment to a real market.

– Less overfitted deep reinforcement learning agents have higher returns.
– The proposed approach offers confidence in possible deployment to a real market.

Methods used:

– Proposed approach addresses backtest overfitting in cryptocurrency trading.
– Less overfitted agents have higher returns, offering confidence for real market deployment.

– Less overfitted deep reinforcement learning agents have higher returns.
– The proposed approach offers confidence in possible deployment to a real market.

– Paper proposes a practical approach to address backtest overfitting in cryptocurrency trading using deep reinforcement learning.
– Less overfitted agents have higher returns than more overfitted agents.

– Paper addresses backtest overfitting in cryptocurrency trading using deep reinforcement learning.
– Proposes a practical approach to detect and reject overfitted agents.

This paper proposes a practical approach to address backtest overfitting for cryptocurrency trading using deep reinforcement learning, and shows that the less overfittedDeep reinforcement learning agents have a higher return than that of more overfitted agents, an equal weight strategy, and the S&P DBM Index (market benchmark), offering confidence in possible deployment to a real market.