Non-linear Chaotic Time Series Gold Futures High-frequency Trading Price Forecasting Plan Planning

Complementary coordination between China's gold futures market and other financial factor markets is an essential part of China's financial market system.Gold futures have many features,including large trading volume,strong liquidity,long industry chain,multiple prices affecting factors,and strong price volatility.Price prediction of gold futures is a research hotspot.The financial time series composed of gold futures prices are affected by various factors,which make it chaotic and difficult to predict.According to chaos theory,the behavior of a very complex economic time series,which appears to be random,can be explained by a deterministic nonlinear system.This article uses 5minutes of continuous gold futures data from January 3^rd,2017 to November 29^th,2019and wavelet analysis to smooth the noise on data,and proposes three prediction schemes.The first prediction is using the traditional linear ARIMA model to make an ordinary linear prediction.The second prediction is based on the non-linearity of the gold futures price time series,and use the RBF neural network to predict.The third prediction is based on the non-linearity of the data to check whether it has chaotic characteristics,the golden time must be the price sequence is reconstructed in phase space,and then predicted based on the nature of nonlinear chaos with RBF neural network.The empirical results show that the prices fluctuation of gold futures is very complicated;the traditional linear structure model is not very effective in predicting a complex time series like gold futures prices.The nonlinear time series model has some results in predicting gold futures prices;nonetheless,there is still improvement space.Gold futures prices follow a complex nonlinear dynamic process;when the optimal embedding dimension m is 47 and the delay time?is 5,the prediction accuracy using chaotic-RBF neural network reaches the highest.