The Expansion Of The Quantitative Investment Strategy Parameters Constructed By Naive Bayesian Method

The naive Bayesian method is a classification algorithm based on Bayes' theorem,mainly used in the field of machine learning and artificial intelligence.Its application in the financial market is still in its preliminary stage,mainly in CTA strategy.At present,parameters of the investment strategy constructed by this classification algorithm are still the traditional index and fundamental indicators,which lack flexibility.Starting from the shortterm fluctuation law of the A-share market,this paper divides the stock historical closing price data,and obtains the data sets of non-overlapping equal-length sub-sequences.Using the index of elements in the subsequence as a new parameter of the naive Bayesian classification algorithm,build a quantitative investment strategy,and test its effectiveness.This paper use following steps to find this parameter: Firstly,take the GEM as the research object,and process the data to find its distribution during January 1,2016 to June 1,2018 for it's distribution feather.Secondly,according to the results,segment the historical closing price data to obtain the parameter value.Then an index trading strategy is initially constructed Based on the naive Bayesian classification algorithm.Test the strategy to see if the investment strategy based on this parameter can achieve the gains beyond the market performance.Finally,parameters are optimized.The length of the subsequence is set to a variable value,and the range of the sample data is expanded.Through the program traversal,find out the length of the optimal sub-sequences of the Shanghai main board,small and medium board,and the GEM from January 2016 to January 2019.Taking the index of elements in the subsequence as one of the characteristic parameters of the naive Bayesian classification algorithm,the portfolio is constructed with the latest transaction data in 2019.So explore the investment strategy after adding this parameter,whether it is more effective than the previous strategy.By comparing the performance of the portfolio constructed by the investment strategy under different characteristic parameters.Conclusion comes that take the index of elements in the sub-sequence as a parameter,which is divided from historical closing price,can be used as an extension of the quantitative investment strategy parameters constructed by the naive Bayesian method,which has a relatively good empirical effect.