一种基于机器学习和蜡烛图的股市投资策略研究  

A Novel Stock Investment Strategy Using Fusion of Machine Learning Techniques and Candlestick Charting

作　　者：林耀虎 刘善存[1] 杨海军[1] LIN Yaohu;LIU Shancun;YANG Haijun(School of Economics and Management,Beihang University,Beijing 100191,China)

机构地区：[1]北京航空航天大学经济管理学院,北京100191

出　　处：《计量经济学报》2022年第1期126-140,共15页China Journal of Econometrics

基　　金：国家自然科学基金(71771008)。

摘　　要：蜡烛图模式能否有效预测市场走向至今没有一致的结论.本文讨论了8种常见的两天蜡烛图模式在中国股票市场做多的预测能力,并在此基础上引入了包括LR,KNN,RF,SVM,RBM等5种常见的机器学习方法和9种技术指标试图增强这些模式的预测能力.基于该方法进而构建了一种超短期的投资策略.实证结果表明,其中6种蜡烛图模式在中国股票市场具有较好的预测做多能力,在引入了机器学习方法后有5种模式的预测准确率得到不同程度的增强,并且可以有效减少最大回撤,产生更好的稳定性.此外,我们还得出,频繁交易会降低投资收益.There is no consistent conclusion whether the candlestick pattern can predict the market trend effectively.This paper discusses 8 common two-day candlestick patterns to make long position predictions in Chinese stock market,and then 5 common machine learning methods including LR,KNN,RF,SVM and RBM are applied to enhance the predictive power of the model.Based on this method,a short-term investment strategy is constructed.Empirical results show that six of these candlestick patterns have predictive power of long position in Chinese stock market.Accuracy of five patterns has been enhanced to varying degrees and produces better stability after the processing of machine learning methods.In addition,we conclude that investment returns would be reduced if trading frequently.

关 键 词：机器学习 蜡烛图 投资策略 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] O157.5[自动化与计算机技术—控制科学与工程] F832.51[理学—数学]