Research on Investment Strategy Based on ARIMA Model and Boruta Random Forest Algorithm

Abstract

Gold has long been a financial asset used by investors to fight inflation and escape financial crises, and Bitcoin is an emerging financial asset that is gaining prominence in the modern economy. This paper analyzes the price trends of gold and bitcoin, which builds a price prediction model and a trading model. First, to achieve a higher yield, we need to make a forecast for gold and bitcoin and then design a trading strategy based on the forecast. In this paper, we use two models to make forecasts. Using ARIMA Time Series Analysis, we need to predict past price movements with 1826 cycles. We need to find the characteristics of current and past price movements that have an impact on future price movements. In this paper, we use Boruta Random Forest Algorithm to filter out the four factors that have the most influence on the price of gold and bitcoin, respectively, from a total of 21 variables, and finally, we use the price trend prediction values and get an investment strategy. Traders can get $92,115.43 on 9/10/2021 for $1,000 on 9/11/2016 with the strategy. The five-year return reached 9111.54%, with an average annualized return of 147.10%.