An Innovative Deep Learning Futures Price Prediction Method with Fast and Strong Generalization and High-Accuracy Research

Abstract

Futures commodity prices are affected by many factors, and traditional forecasting methods require close attention from professionals and suffer from high subjectivity, slowness, and low forecasting accuracy. In this paper, we propose a new method for predicting the fluctuation in futures commodity prices accurately. We solve the problem of the slow convergence of ordinary artificial bee colony algorithms by introducing a population chaotic mapping initialization operator and use the resulting chaotic mapping artificial bee colony algorithm as a trainer to learn long short-term memory neural network hyperparameters. With the combination of gate structures learned by the algorithm, the long short-term memory network can accurately characterize the basic rules of futures market prices. Finally, we conduct a series of backtesting experiments on gold and natural gas futures commodity prices to demonstrate the effectiveness of the proposed model. The experimental results show that, compared with various existing optimization models, our proposed model is able to obtain the lowest mean absolute error, mean square error, and root mean square error in the least number of iterations. In summary, the model can be used to predict the prices of a wide range of futures commodities.

Keywords

• chaotic mapping
• artificial bee colony algorithm
• long short-term memory
• futures price prediction