Research on Multimodal Futures Price Prediction Method based on DCC-GARCH Model

As the economy and technology rapidly advance, financial investment has increasingly become a means for the general public to generate and accumulate wealth. Futures trading, a method in which buyers and sellers enter a standardized contract to exchange a commodity at a predetermined date, is one such avenue. While futures investments require minimal capital and permit investors to sell prior to buying, they also subject traders to greater investment risks. In recent years, as computer science and its interdisciplinary fields have developed, employing mathematical and statistical approaches to analyze and predict the futures market for wise investments has gained popularity. Considering the uncertainty and multimodal characteristics of the futures trading market, this study categorizes futures trading into five modes and investigates the relationship between these modes and futures prices. By utilizing the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model for predictions in varying mode environments, this research applies a mode correction function to adjust the base step in GARCH forecasts to address market fluctuations. Analyzing results derived from actual futures trading data reveals that the multimodal futures price prediction model demonstrates higher accuracy compared to the conventional GARCH model.