Deep Reinforcement Learning Strategies for Futures Trading

Summary

This study applies deep reinforcement learning to continuous futures trading, examining both discrete and continuous action spaces. It incorporates volatility scaling into reward functions so that position sizes respond to market volatility. The algorithms are evaluated on the 50 most liquid futures contracts from 2011 to 2019, covering commodities, equity indices, fixed income, and foreign exchange.

The reported experiments compare the learned strategies with time-series momentum baselines and find that the reinforcement-learning methods perform better, including after heavy transaction costs. The authors describe strategies that can retain positions through large trends and reduce exposure or wait during consolidation. The document does not specify the precise algorithms, contract construction, validation design, or detailed performance measures, so the findings should be understood within the stated dataset and comparison framework rather than as a guarantee of future results.

Key ideas

• The study trains reinforcement-learning strategies with both discrete and continuous trading actions.
• Volatility-scaled rewards adjust position exposure to changing market volatility.
• The evaluation spans liquid futures across several asset classes over the stated historical period.
• The methods are compared with classical time-series momentum and reportedly outperform after transaction costs.
• The summary does not provide enough detail to assess validation choices or performance metrics.

Tags

# Deep Reinforcement Learning for Trading


# Deep Reinforcement Learning for Trading









We adopt Deep Reinforcement Learning algorithms to design trading strategies for continuous futures contracts. Both discrete and continuous action spaces are considered and volatility scaling is incorporated to create reward functions which scale trade positions based on market volatility. We test our algorithms on the 50 most liquid futures contracts from 2011 to 2019, and investigate how performance varies across different asset classes including commodities, equity indices, fixed income and FX markets. We compare our algorithms against classical time series momentum strategies, and show that our method outperforms such baseline models, delivering positive profits despite heavy transaction costs. The experiments show that the proposed algorithms can follow large market trends without changing positions and can also scale down, or hold, through consolidation periods.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.