An Empirical Study On Random Forest Stock Picking Strategy

At present,artificial intelligence is highly valued in the financial industry,and quantitative stock selection can efficiently perform data mining analysis and standardize investment according to models.After the US Trump administration started the trade war,the Chinese stock market plunged on March 23 and June 19,2018,and investors lost confidence in the market.In order to encourage investors to return to the market,this paper uses the random forest stock picking strategy based on principal component analysis to conduct empirical research on the 50 constituents of SSE.This paper uses Python programming software,Choice data interface,JQData local quantitative financial data to select SSE 50 constituent stocks as stock pool,from January 1,2015 to December 31,2018 as a backtest interval,the alternative feature factor pool contains estimates.Value indicators,risk analysis,profitability,trend indicators,energy indicators,and volatility indicators.The backtest interval is divided into four sub-intervals according to the year,and the current year is pushed forward into the sample by using the staged back test method.The principal component analysis method is used to reduce the feature and eliminate the correlation between the features.Sexuality,combined with the 10-fold cross-validation method and the grid search method to optimize the model parameters to train the decision tree and form a random forest model.Enter the backtest data into the model and select the top 5,10 and 15 stocks as the next month's strategy portfolio.The backtest results show that the three strategy combinations are better than the SSE 50 ETF in terms of market value and logarithmic yield.The strategy combination with the number of positions is the best,and the principal component analysis method can effectively improve the strategy.Investment effect.In the Sino-US trade war period,the strategy maintained its effectiveness,and the effectiveness was stronger than the full back-test period.The maximum revaluation rate and the Beta value were lower than the full back-test period.This paper combines machine learning to conduct empirical analysis and concludes that investors do not have to fear the market in the face of a sudden bear market.They should remain calm and still obtain excess returns by using quantitative investment tools.