Gold Price Forecasting Model Based On SVM And Its Parameter Optimization

With the recovery of global economy since 2016,commodities ended the bear market for five consecutive years.And market trading volume gradually increased and the market gradually active.Commodities return to the investor's insight.CTA strategy for nearly two years of extraordinary performance and the rise of machine learning,so that researchers?investors have a lot of interest of the application of machine learning to commodity transactions.Based on the support vector machine model in machine learning,this paper chooses the gold with the strongest financial attributes in the commodity,forecasts the price of gold,and tries to construct a time strategy.This paper is divided into five parts:The first part mainly introduces the background of this topic,the significance of the study and the framework of this study.In the second part,the attributes and functions of the gold varieties are expounded.Based on this analysis,the main factors influencing the gold price are analyzed.Finally,seven influencing factors are selected as the support vector in combination with the practical significance,the correlation analysis and the Granger causality.The third part of this paper is the core of the article.Firstly,the basic principle and application of SVM algorithm are introduced,and the algorithm flow is given.Then the basic principle and algorithm of grid searching and particle swarm optimization algorithm are introduced.,The third chapter is based on MATLAB based on the above-mentioned principle algorithm to predict the price of gold,and to minimize the MSE as the goal to optimize the SVM model of the penalty parameters and the kernel function parameters.In the fourth chapter,based on the third chapter to forecast the gold price of SVM model,the initial construction of a time trading strategy,and gives the backhaul in the back of the test results.In the fifth chapter,the paper summarizes the shortcomings and innovations of the model,and finally gives a brief evaluation of the model and the optimization results,and makes a prospect and pointed out the direction of further research.The data involved in this article are from the Bloomberg client,data processing and model implementation based on MATLAB software.