Data Cross-Segmentation for Improved Generalization in Reinforcement Learning Based Algorithmic Trading
Proposes a reinforcement learning algorithm for trading in illiquid markets, tested on 20+ years of equity data.
What it examines
This paper proposes a reinforcement learning algorithm for trading in illiquid markets with high friction costs, combining supervised prediction with reinforcement learning. The approach aims to improve generalization by training prediction and RL policy models on non-overlapping datasets, tested on 20+ years of equity data from Bursa Malaysia.
What it concludes
The study concludes that RL algorithms can effectively train trading policies based on predictive models, with data cross-segmentation enhancing performance. Future work could explore multi-horizon predictions and newer network architectures. Potential applications include improved trading strategies in illiquid markets.
Evidence objects
The study concludes that RL algorithms can effectively train trading policies based on predictive models, with data cross-segmentation enhancing performance. Future work could explore multi-horizon predictions and newer network architectures. Potential applications include improved trading strategies in illiquid markets.
key_findings bullet 1 · key_findings · validation V0
Raw abstract and provenance
Data Cross-Segmentation for Improved Generalization in Reinforcement Learning Based Algorithmic Trading Vikram Duvvur* Carnegie Mellon University vduvvur@cs.cmu.eduAashay Mehta* Carnegie Mellon University aashaypm@cs.cmu.eduEdward Sun* Carnegie Mellon University edwardsu@cs.cmu.edu Bo Wu* Carnegie Mellon University bw1@cs.cmu.eduKen Yew Chan Kenanga Investment Bank Berhad kychan@kenanga.com.myJeff Schneider Carnegie Mellon University jeff.schneider@cs.cmu.edu ABSTRACT The use of machine learning in algorithmic trading systems is increasingly common. In a typical set-up, supervised learning is used to predict the future prices of assets, and those predictions drive a simple trading and execution strategy. This is quite effective when the predictions have sufficient signal, markets are liquid, and transaction costs are low. However, those conditions often do not hold in thinly traded financial markets and markets for differentiated assets such as real estate or vehicles. In these markets
Source row: 494 · abstract type: unknown