Predicting the Nikkei 225 Using Machine Learning: An Empirical Comparison

This study compares the predictive performance of multiple machine learning models for forecasting the closing price of the Nikkei 225 on the forecast day using daily data. Five models—Random Forest Regression, Decision Tree Regression, XGBoost Regression, LightGBM Regression, and K-Nearest Neighbors Regression—are employed. In addition to baseline models that use lags of the closing price as explanatory variables, three additional testing models are constructed by adding lags of the opening, high, and low prices individually to the explanatory variables of the baseline models. Furthermore, six different forecast target periods, ranging from five to ten years, are considered, and predictive performance is evaluated using the R-squared value, mean absolute error, and root mean squared error. The results show that Random Forest Regression and Decision Tree Regression exhibit relatively high predictive performance, followed by K-Nearest Neighbors Regression. In addition, the results indicate that adding lags of the opening, high, and low prices to the explanatory variables slightly improves predictive performance in some cases. Furthermore, predictive performance tends to improve as the forecast target period becomes longer. These findings suggest that, in forecasting the closing price of the Nikkei 225, predictive performance may be influenced not only by the type of model but also by the composition of explanatory variables and the forecast target period.