A Multi-objective Deep Reinforcement Learning Approach for Stock Index Future’s Intraday Trading
Authors:
- Weiyu SiWSWeiyu Si
- Jinke LiJLJinke Li
- Peng DingPDPeng Ding
- Ruonan RaoRRRuonan Rao
- 1 December 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 431-436
Abstract
Modern artificial intelligence has been widely discussed to practice in automated financial asserts trading. Automated intraday trading means that the agent can react to the market conditions automatically, while simultaneously make the right decisions. Besides, the profits will be made within a day considering transaction cost charged by the brokerage company. In this paper, we introduce a multiobjective deep reinforcement learning approach for intraday financial signal representation and trading. We design the deep neural networks to automatically discover the dynamic market features, then a reinforcement learning method implemented by a special kind of recurrent neural network (LSTM) is applied to make continuous trading decisions. In terms of balancing the profit and risk, we implement a multi-objective structure which includes two objectives with different weights. We conduct experiments on stock index futures data, and our analysis and experiments not only offer insights into financial market features mining, but also provide a straightforward and reliable method to make profits, which sheds light on its wide application on automated financial trading.Keywords
This publication has 20 references indexed in Scilit:
- Forecasting S&P 500 index using artificial neural networks and design of experimentsJournal of Industrial Engineering International, 2013
- Threshold Recurrent Reinforcement Learning Model for Automated TradingLecture Notes in Computer Science, 2010
- Learning Deep Architectures for AIFoundations and Trends in Machine Learning, 2009
- A Multiagent Approach to $Q$-Learning for Daily Stock TradingIEEE Transactions on Systems, Man, and Cybernetics - Part A: Systems and Humans, 2007
- Learning-Induced Plasticity in Deep Cerebellar NucleusThe Journal of Neuroscience, 2006
- An automated FX trading system using adaptive reinforcement learningExpert Systems with Applications, 2006
- Learning to trade via direct reinforcementIEEE Transactions on Neural Networks, 2001
- Comparative study of stock trend prediction using time delay, recurrent and probabilistic neural networksIEEE Transactions on Neural Networks, 1998
- Long Short-Term MemoryNeural Computation, 1997
- TD-Gammon: A Self-Teaching Backgammon Program1995