Gold Price Forecast Based On Grey Neural Network Model

Gold has been constantly made as a hot topic global wide,due to its characteristics as not only an expensive medal,but also a hard currency that behaves as a world money.The scale of Gold's value,the measure of circulation,storage and payment,all these traits also enable investors to avoid the erosion of inflation,as well as the risk from uncertainties of real estate and stock market.Hence,holding gold behaves as a much better value-retaining approach,and its hedging function outstands others especially among unstable markets.The value of gold has a predictable change on a macroscopical trend for a long run,and some frequent fluctuations on a microscopical trend in the meantime.The ability to precisely foresee it's next move has significant meanings to both individuals and at large,entire nation.This dissertation focuses on predicting the price of gold as a target,after analyzing the existing prediction model's defects and downsides,which enables optimized results through two aspects: excavating factors and perfecting model.By utilizing relevant ruling data digging technology to explore the affected factors of gold price,not only the traditional economical conclusions in strong correlation factors such as gold,cruel oil,and CPI have been found,more factors such as pork price,bronze price,and Baidu index,in which were never considered,have been considered in to correlated results.Also,by taking advantage of Grey correlation and its main ingredient analysis' s filtration,factor's reliability has been greatly proved.In the meantime,when combining the thoughts of Grey System and Neural Networks,creating an integrated model of Grey neural networks,very little data is needed to reach consequences as well as non-linear reflection traits,which in the end,further improves the accuracy of prediction.This dissertation respectively build GM(1,1)and BP neural networks models,parallel integration models,series integration models,and insertion integration models to predict the price of gold,and later on compare with the result of the other two external models in order to calculate the absolute difference and analyze the result,which show that series integration model processes the highest accuracy.Finally,this dissertation systematically summarizes the solutions and hypothesis claimed,and by comparing two groups of individual factors to evaluate the prediction ability of the entire system,proving that this dissertation provides high definition and strong reliability of this predicting system.