基于遗传规划的多源信号交易策略组合优化

Genetic programming based trading strategy optimization with multi-sourced signals

将财经新闻事件转化为交易信号融入到策略组合中,提出基于遗传规划的多源信号股票交易策略组合优化模型.模型综合财经事件信息以及移动平均、动量等传统技术交易策略,利用遗传规划对多源信号交易策略组合进行整体优化.采用中国股市数据实验发现,典型财经新闻事件可以有效转化为交易信号并与技术交易策略融合;融合新闻事件信号与传统技术策略可以有效提升股票自动交易的收益率;基于遗传规划的组合优化模型可以生成有效的交易策略,其中新闻与指数移动平均组合的策略可以产生较高的收益率并具有较强的稳健性.

This paper integrates the financial news events as signals into trading strategies and proposes a blended trading strategy model based on genetic programming.Such a model is the exploitation of news events in traditional strategies like moving averages and momentum.Experiments on the data from news web-sites and China stock market show,that the typical financial news events can be effectively transferred into signals for trading strategies,that the integrated trading strategy will prominently improve on traditional trading strategies,while the proposed model based on genetic programming can produce effective trading strategies for application,and that the news with exponential moving average strategy will reap higher returns and are more robust than other combination strategies.