---
title: Deep Reinforcement Learning for Active High Frequency Trading
url: https://www.ml-quant.com/papers/arxiv/2101.07107/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: arXiv:2101.07107
source_url: http://dx.doi.org/10.48550/arxiv.2101.07107
featured: 2023-08-24
citations: 51
topic: Trading, Microstructure & Execution
---


# Deep Reinforcement Learning for Active High Frequency Trading

A new Deep Reinforcement Learning framework has been developed for high frequency stock trading, showing potential for profitable long-term strategies.

- Source: http://dx.doi.org/10.48550/arxiv.2101.07107
- Identifier: arXiv:2101.07107
- Released: 2021-01-18
- First featured: Quant Letter No. 13 (2023-08-24): https://www.ml-quant.com/issues/2023-08-24/
- Citations (Semantic Scholar): 51
- Published in: not yet
- Topic: Trading, Microstructure & Execution

## Related

- [Robust Market Making with Hawkes Order Flow and Price Impact via Adversarial Reinforcement Learning](https://www.ml-quant.com/papers/arxiv/2609.22785/): The research extends adversarial reinforcement learning for market making to handle self-exciting order arrivals and price impact, using an LSTM module to improve robustness in complex microstructure environments.
- [FAST: Efficient Action Tokenization for Vision-Language-Action Models](https://www.ml-quant.com/papers/arxiv/2501.09747/): A new tokenization scheme, Frequency-space Action Sequence Tokenization (FAST), has been proposed for robot actions, facilitating the training of vision-language action policies for complex and high-frequency tasks.
- [Identifying High Frequency Trading Activity without Proprietary Data](https://www.ml-quant.com/papers/ssrn/4551238/): The article evaluates the reliability of commonly used indicators to identify high frequency traders, showing variations in performance and suggesting that unscaled proxies are more effective at indicating true HFT activity.
- [JAX-LOB: A GPU-Accelerated limit order book simulator to unlock large scale reinforcement learning for trading](https://www.ml-quant.com/papers/arxiv/2308.13289/): JAX-LOB: The paper introduces JAX-LOB, the first GPU-powered limit order book simulator capable of processing multiple books simultaneously, designed for efficient large-scale simulations of LOB dynamics for research, calibration, and reinforcement learning training.
- [Microstructure-Empowered Stock Factor Extraction and Utilization](https://www.ml-quant.com/papers/arxiv/2308.08135/): A new model has been suggested to better predict stock trends and improve order execution by analyzing high-frequency order flow data in stock investment.
- [C Design Patterns for Low-Latency Applications Including High-Frequency Trading](https://www.ml-quant.com/papers/ssrn/4565813/): The research focuses on improving high-frequency trading systems by optimizing latency-critical code, resulting in a Low Latency Programming Repository and an optimized trading strategy.