Thesis On Price Forecast And Trading Strategy Of Stock Index Futures Based On XGBoost Model

The financial attributes of the stock index futures market are strong,and as an important financial tool for hedging stock market risks,it has been widely valued by various investors.When trading stock index futures,many investors choose to use tools such as fundamentals or technical indicators for trading,but these tools bring limited returns and high risks.With the concept of quantitative investment entering the domestic market,the use of quantitative models and tools for trading stock index futures has attracted attention from academia and various investors.With the development of machine learning and other research,using machine learning and other models to predict financial data prices and construct trading strategies has become an important direction for the development of quantitative investment.This thesis predicts the rise and fall of stock index futures prices,and based on the rise and fall prediction results,constructs intraday and trend trading strategies for stock index futures to improve the returns of the strategies and reduce risks.Firstly,analyze the problems of traditional time series and machine learning models in predicting the rise and fall of stock index futures prices.Secondly,design a prediction index system based on various technical indicators;Building a stock index futures price forecast model based on the XGBoost model;Adopting genetic algorithm for parameter optimization;Use AutoEncoder to process the input technical indicators.Thirdly,based on the rise and fall prediction model,construct intraday and trend trading strategies for stock index futures.Finally,collect data on the Shanghai and Shenzhen 300 Index and conduct empirical analysis on the constructed prediction model and trading strategies.The conclusions of this thesis are as follows:(1)The traditional time series models such as ARMA and machine learning models such as random forest are not effective in predicting the price rise and fall of stock index futures,and the trading strategies based on these models are risky.(2)The XGBoost model has a good predictive effect on the rise and fall of Shanghai and Shenzhen 300 stock index futures prices.Compared with the ARMA model,the precision has increased by 0.2104,the recall has increased by 0.2331,the accuracy has increased by 0.2156,the true negative rate has increased by 0.1982,and the AUC has increased by 0.2055.At the same time,due to models such as GBDT and support vector machine.And it can be seen that linear models such as logistic regression and ARMA have lower predictive ability,which is basically the same as random prediction.(3)On intraday trading,XGBoost achieved better trading results with an annualized return of 17.26%,higher than other models,and a maximum pullback rate of 16.92%,lower than other models,while the ARMA model had a return of less than 0.In trend trading,XGBoost has better trading results,with an annualized return of 19.31%,higher than other models,and a maximum pullback rate of 18.26%,indicating lower risk than other models.