Prophet Algorithm Optimization with Particle Swarm Optimization for Adjusted Closing Price Forecasting in Xiaomi Corporation

Stock trading is a crucial global investment activity, relying heavily on accurate price forecasting to enhance profits and mitigate risks. However, forecasting stock prices is challenging due to the volatility of stock prices fluctuations and complexity of economic factors. This research addresses these challenges by developing a forecasting model to forecast Xiaomi Corporation’s adjusted closing prices, combining Prophet and Particle Swarm Optimization (PSO) algorithms. PSO optimizes Prophet’s hyperparameters to enhance forecasting accuracy. Utilizing a dataset spanning five years, divided into 1,054 and 264 data points for training and testing, the Prophet model incorporates additional regressors such as Open, High, Low, Close, and Volume. The PSO-Prophet model, tested across four scenarios, achieved a best mean absolute percentage error of 0.00000008522 and a best coefficient of determination of 0.9999999999991, with optimized Prophet’s hyperparameters and additional regressors.