计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (15): 86-91.DOI: 10.3778/j.issn.1002-8331.1908-0450

• 大数据与云计算 • 上一篇    下一篇

基于gcForest的多因子量化选股策略

王伦,李路   

  1. 上海工程技术大学 数理与统计学院,上海 201620
  • 出版日期:2020-08-01 发布日期:2020-07-30

Multi-factor Quantitative Stock Selection Strategy Based on gcForest

WANG Lun, LI Lu   

  1. School of Mathematics and Statistics, Shanghai University of Engineering Science, Shanghai 201620, China
  • Online:2020-08-01 Published:2020-07-30

摘要:

为了获取股票市场更高的超额收益,提高股票涨跌预测准确率,将gcForest(深度森林)算法引入了股票投资市场,建立基于gcForest多因子量化投资策略,每月月末在沪深300成分股中买入gcForest算法预测上涨的前30支股票,并进行回测。研究结果表明,gcForest算法的年化收益率为29.2%,远超基准年化收益率15.0%,并且获得了15.8%的超额收益。同时还将gcForest算法同随机森林和支持向量机算法进行了比较,从各项技术指标综合分析来看,gcForest算法在股市行情平稳和上涨时期都较其他算法有着明显的优势。

关键词: 多因子选股, 多粒度扫描, gcForest算法

Abstract:

In order to obtain higher excess returns in the stock market and improve the accuracy of stocks’ ups and downs, this paper introduces the gcForest(deep forest) algorithm into the stock investment market, and establishes a multi-factor quantitative investment strategy based on gcForest, which is composed of 300 constituent stocks at the end of each month. The gcForest algorithm is used to predict the top 30 stocks and the backtest is conducted. The research results show that the annualized rate of return of the gcForest algorithm is 29. 2%, far exceeding the benchmark annualized rate of return of 15. 0%, and an excess of 15. 8% is obtained. At the same time, the gcForest algorithm is compared with the random forest and support vector machine algorithm. From the comprehensive analysis of various technical indicators, the gcForest algorithm has obvious advantages over other algorithms in the stable and rising stock market.

Key words: multi-factor stock selection, multi-granularity scanning, gcForest algorithm