Quantitative Stock Selection Based On Xgboost Exponential Enhancement Strategy Design

The index enhancement strategy is an investment strategy that seeks to exceed the performance of the benchmark index by optimizing the portfolio allocation on the premise of controlling the tracking error of the benchmark index,and is an organic combination of active investment and passive investment.am.Based on the Index Enhancement Strategy,Index Enhanced Funds are attracting attention from institutional and private investors due to their low fees,excellent long-term returns and high transparency in position adjustments.As of the end of December 2021,my country has 144 index-based expansion funds with a market size of 141.7 billion yuan,a growth of more than 20%year-on-year in both quantity and size.It has been a good development momentum and is now an important index fund in my country.However,in today’s financial markets,stock price-related factors are increasingly being discovered,and issues such as colinearity,endogeneity,and non-linearity are building traditional multi-factor stock selection models based on multiple regression...Enhanced strategic portfolios can no longer get excess returns.With the continuous development of mathematical statistical theory related to artificial intelligence technology,some excellent machine learning algorithms have been studied,and their excellent non-linear data processing capacity predicts inventory increase and decrease more accurately.,You can build a yield.A relatively large index expansion strategy portfolio.Based on the above background,in this paper,we introduce a good Xgboost model of machine learning algorithms into our index enhancement strategy,and use its good classification ability to select high quality stocks and then in equal weight position.Run backtests and get a good and stable return on investment.This paper first studied and analyzed the theoretical basis and applicability of Xgboost multifactorial strain selection,and Xgboost has higher predictive accuracy compared to traditional multiple linear regression models and other machine learning models.We believe it has the advantage of strong interpretability.Then build an element library for this paper by choosing 69 elements from rating,scale,growth,leverage,finance,momentum,technology,and more.Then split the training set and test set to train and optimize the parameters of the Xgboost model.All models have AUC values above0.5 over the entire backtest period,indicating that the Xgboost model can predict an increase.And the stock price fell.Finally,the stocks in the stock pool are scored using the predicted up and down probability values of the stock price output by the Xgboost classification algorithm.Selected for index expansion portfolio.The selected stocks will be backtested with the same weight.During the backtest period from 2016-12-31 to 2021-12-31,the total return of the strategic portfolio was 71% and the annual rate of return was15.43%...The yield alpha is 9.2%.Over the same period,the CSI 300 Index returned45.98% and the strategic portfolio returns 35.95% higher than the CSI 300 benchmark returns.This shows that an index enhancement strategy built on Xgboost’s quantitative equity selection is effective.