Research On Quantitative Investment Strategies Of Stock Index Futures Based On Machine Learnin

Investors in today’s market will receive a large amount of data information every day.In the face of huge and complex information,investors often feel powerless to make correct decisions quickly and timely.Moreover,people have character defects and will be affected by market emotions and make impulsive and unwise decisions.Based on this,this paper aims to use machine learning to find an accurate price prediction model,and build a quantitative investment strategy based on price prediction to solve problems that cannot be solved by human.The main content of this study is to predict the closing price of the main continuous contract of CSI 300 stock index futures(IF9999)in the future day by building BP neural network and LSTM long and short term memory network,and find the optimal closing price prediction model of stock index futures in the future day by comparison.The quantitative investment strategy of CSI 300 stock index futures is constructed on the basis of the price prediction of the optimal prediction model.The research methods are literature research and comparative analysis.Firstly,through literature review,it is found that predecessors’ prediction data of stock index futures prices are only limited to the opening price,closing price,maximum price and minimum price,and they pay little attention to domestic and foreign macro data affecting stock index futures prices.Second,moving average trading strategy automation has not been fully developed and utilized.Finally,the comparative analysis is mainly reflected in the BP neural network hidden layer and LSTM long and short memory network step size were compared and determined the BP hidden layer number and LSTM time step,and then the network pattern parameter optimization and prediction performance comparison of BP and LSTM.Due to the superior evaluation indexes in all aspects of BP neural network prediction,the predicted value of BP neural network was selected as the backtest data for constructing quantitative investment strategy of CSI 300 stock index futures.The quantitative trading strategy of CSI 300 stock index futures was developed and optimized.The returns of the optimized strategy were significantly higher than those of the basic strategy and the benchmark strategy,and the evaluation indexes in all aspects were significantly better than those of the basic strategy,and good results were obtained.In the aspect of innovation,it mainly further enriched the data of stock index futures price prediction,which included not only the basic quantity and price data and technical index data,but also the macro data at home and abroad.The random forest algorithm is used to screen characteristic factors to find the more important characteristic factors that affect the price of stock index futures.The algorithm is optimized by using the method of network style parameter,and the prediction performance is improved.The introduction of traditional moving average trading strategy has improved the basic quantitative strategy and achieved higher returns.