Combining Traditional Stock Selection Models and Multi-factor Stock Selection Models: Achieving Dynamic Adjustments of Machine Learning Algorithms

Stock selection models which use machine learning (ML) techniques can capture the nonlinear influences of stock trends. However, there is a lack of comprehensive research on the combined use of various machine learning algorithms. This study bridges this gap by developing a new stock selection strategy that combines traditional stock selection models with machine learning multi-factor models. this study by using various machine learning algorithms to build a multi-factor stock selection model, and on each trading day, assigning corresponding weights based on the market performance of the previous \(n\) days according to the performance of different algorithms. The best-performing algorithms in each period can be selected through this method, and then the corresponding stock investments are selected. Finally, the experimental results show that the newly constructed stock selection strategy has excellent market performance.