Jane Street Stock prediction model based on LightGBM

In recent years, the application of deep learning prediction models in predicting stock price trends has attracted extensive attention from researchers, which can help investors make better investment decisions. Therefore, we aims to use the data set provided by Jane Street to establish a stock forecast model and develop a trading strategy. In this paper, we proposed a stock prediction model based on LightGBM. Experiments show that our model has better predictive power and higher returns compared with other machine learning models. Specifically, our LightGBM-based stock prediction model has a return rate of 24.7% higher than that of the Xgboost model and 8.9% higher than that of the Nerual Network.