Stock Index Futures Investor Sentiment Strategy Based On GA-PLS

Investor sentiment is the general attitude of investors towards the expected price development.Behavioral finance takes investor sentiment as one of the factors of asset pricing model and points out that investor sentiment has significant influence on asset price.At the same time,many empirical studies show that investor sentiment plays a certain role in predicting future asset prices: information is extracted from multiple market variables,to construct sentiment index and reflect the trend of future underlying prices in some aspects.Therefore,this paper constructs the corresponding investor sentiment index to predict the future trend of the underlying stock index futures.In this paper,the genetic algorithm is mainly used to search for the optimal combination of indirect proxy variables of investor sentiment,and the information extraction technology is used to extract the emotional information from the proxy variables to construct the investor sentiment index with the optimal prediction performance.Among them,partial least square method is adopted to extract information from the combination of proxy variables and construct the index,so as to overcome the collinearity problem in the 63 proxy variables and eliminate some variation information irrelevant to the underlying stock index futures.In order to compare the prediction performance of different variable combination models,the fitness function of genetic algorithm is set as the sum of prediction errors in the cross test of partial least square method.Data from January 2015 to August 2019 were selected for strategic backtesting.Because there are only SSE 50 ETF options in the options market,the stock index futures used by the backtesting subject are SSE 50 stock index futures.The results showed that SSE 50 index futures performed better on the investor sentiment strategy under the 250-day window,with a total return of 401% and an annualized return of 125%,while SSE 50 index had a total return of 22.1% and an annualized return of only 6.8%.The total return of the Shanghai composite index was-7.47%,and the annualized return was-9%.The total income of shenzhen component index was-15.55%,and the annualized income was-4.85%,both of which suffered losses.The total return of the CSI 300 index is 9.24%,and the annualized return is only 2.88%.Strategic returns are better than the market index,benchmark index and the underlying index.Meanwhile,the maximum risk indicator retractions,sharpe ratio and information ratio were 10.58%,6.46 and 4.78,respectively,with excellent performance.The results of the other two stock index futures are poor.The total return of the 250-day strategy of CSI 300 stock index futures and CSI 500 stock index futures is 153.4%,179.92%,and the annualized return is 47.29% and 55.42%,respectively.This is because the importance analysis of variables shows that the option market is the most important in the strategy model,and in the variable pool is the proxy market variable classification that can best reflect investor sentiment.However,there are only sse 50 ETF options in the options market in China,and there are no corresponding options,so the strategic returns of these two stock index futures are not good.Moreover,in the results of unilateral strategy,the return of long-only strategy is 154.22%,the annualized return is 48.07%,and the return of short-selling strategy is 97.24%,and the annualized return is 30.31%.The advantage of long strategy over short strategy indicates that the existing indirect proxy variables lack accurate measurement of short market.Therefore,China should vigorously develop the short selling financial derivatives market represented by the options market,develop short selling financial instruments and panic index,provide tools and indicators for hedging the short selling market risk,and promote the healthy development of the securities market.