Research on Futures Price Forecast of WVMD-LSTM Model

The research on futures price prediction is of great significance to society and enterprises. In order to cope with the increasingly complex financial market environment, this paper proposes a new hybrid model, WVMD-LSTM, to improve the accuracy of futures price prediction by capturing the information of multiple dimensions in futures data. A series of single models and mixed models are used to comprehensively compare the forecasting performance of VMD-LSTM model, and the gold futures price is selected to verify the effectiveness of WVMDLSTM in futures price forecasting. The performance of the WVMD-LSTM model is comprehensively evaluated by fitting degree indicators (RMSE, MAE, MAPE, R²) and direction accuracy indicators with economic significance $(\mathrm{D}_{\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}})$. The results show that compared with ARIMA, VMD-LSTM, WVMD-LSTM has the best prediction ability.