Forecasting the Financial Market Using Random Forest -A Case Study

摘要

In this paper, we attempt to model and forecast the movement of the excess return of a Chinese closed-ended fund. We introduce a new non-parametric random forest method to forecast the signs of excess return, and compare the results with some conventional linear models such as the random walk model (RW), ARIMA model and Support Vector Machine (SVM). The out-of-sample results show that the performance of random forest is superior to others. Moreover, we also construct the various trading strategies based on different forecasts methods. The performance of different trading strategies are contrasted and also compared with a simple buy-and-hold strategy. The results indicate that the trading strategy based on random forest methods earns significantly higher returns than other alternatives.