Gold Futures Price Prediction Based On CNN-BiGRU Model

As a special kind of futures in the futures market,gold futures plays an important role in the financial market because of its hedging function,price discovery function and risk investment function.Either it is hedging from the perspective of investors or maintaining the stability of the financial market from the state level,it is necessary to deeply study the law of gold futures price and forecast the price of gold futures.This thesis aims to forecast the settlement price of gold futures in Shanghai Futures Exchange.It mainly starts from two aspects: On the one hand,starting from the gold futures settlement price sequence itself,the machine learning model is used to fully learn the time sequence dependence contained in the settlement price sequence itself.On the basis of the existing model,CNN and bidirectional GRU are combined to construct the CNN-Bi GRU combination model,and the existing model and CNN-Bi GRU model are used to forecast the settlement price of gold futures in Shanghai Gold Futures Exchange from February 2008 to December 2022,the results show that CNN-Bi GRU model has a good fitting effect,indicating that the combination model has certain advantages in settlement price prediction.On the other hand,multiple indexes that affecting the settlement price of gold futures are introduced,such as the spot price of gold,Dow Jones Industrial Index,S&P 500 index,etc.,and correlation analysis is carried out on these indexes,so as to screen out the influential factors which have a strong correlation with the settlement price of gold futures,then,the principal component analysis method is used to reduce the dimension of these factors,and the extracted principal components are used as the input of the model together with the historical data of gold futures settlement price.Finally,the existing model and CNN-Bi GRU model are used to predict the settlement price,the prediction results also show that the CNN-Bi GRU model has small prediction error and high goodness of fit.In general,compared with the comparison model,CNN-Bi GRU model has the best prediction accuracy effect when only the gold futures settlement price itself series and the price series of the influencing factors of gold futures price are considered comprehensively.In addition,compared with only considering the sequence of gold futures settlement price itself,the fitting effect of the model is better when the factors affecting the gold futures settlement price are comprehensively considered,which indicates that the analysis of settlement price combined with influencing factors can effectively improve the prediction ability of the model.