Using LightGBM + SHAP to Analyze the Impact of Factors on Stock Market Investment

In recent years, the number of studies using artificial intelligence (AI) in financial investment has increased rapidly, and it has become an important tool for financial asset pricing prediction. However, when AI models perform predictive analysis, the calculation process is black box processing, and it is difficult to determine the contribution of each feature in the model.Therefore, this study uses a rolling window to build a model, and uses LightGBM (Light Gradient Boosting Machine) the SHapley Additive exPlanations (SHAP) suite to analyze the contribution of factors constructed by the model to understand the importance and contribution of each factor in the investment portfolio constructed by the model.The research period is from 2017 to 2023. The model built with 78 rolling windows selects several stocks with high expected returns for back testing. The research results show that the contribution of market capitalization has increased significantly. The most influence factors in the model are stock price to net value ratio and market capitalization. Among them, market capitalization has the best contribution.