Trend Forecast And Demonstration Of US Dollar Index Based On COT

The US dollar,is an intermediate currency and plays the important role in the economies of various countries.The US dollar index shows the strength and weakness of the dollar that relative to other non-US currencies.Predicting the trend of the US dollar index provides early warning of abnormal fluctuations in exchange rates of various countries and avoids major financial risks caused by exchange rate changes.Therefore,it is of practical significance to predict the trend of the US dollar index.Scholars mostly used fundamental and technical methods to predict exchange rates at home and abroad.However,economic variables in fundamental analysis are difficult to obtain because of the long publication cycle and small amount of data.It is not very convincing to predict the future price by using the technical method.This paper uses the data of the commitment of Traders(COT)to predict the price trend of the US dollar index.The COT of dollar index is made up of three types of position reports and five types of position reports.By analyzing of correction,direction consistency and gray correlation,a set of suitable data is chosen.The results show that the swap dealer's net position and commercial net position are the best data to predict the price trend of the US dollar index in next 7 weeks.Therefore,a VAR model is established with the data of the USD index price,swap dealer's net position and commercial net position.After logarithmic processing and first-order difference of the data,all three sets of data were stationary sequences and were first-order single integer sequences.Granger causality test results showed that these two sets of data are both Granger causes of the dollar index price,and there is no Granger cause between the two sets of COT data.The VAR impulse response graph shows that the impact on the US dollar index price reaches its peak after the lag period of 5,and then gradually becomes to zero over time.Equation decomposition experiments show that the two sets of data affect the price of the US dollar index with increasing of time.Therefore,the changed in the commercial net position and swap dealer's net position will affect the price of the US dollar after 7 weeks.The empirical results of static prediction shows that the data generated by the prediction are in good agreement with the real value and the error between them was relatively small.The presented model is an effective way to forecast of the US dollar index futures price by using the commercial net position and the swap dealer's net position.Because the trading volume of US dollar futures is much smaller than the trading volume of non-US currency futures,two new COTs of US dollar are presented by combination of the COT of non-US currency.In the first new method,it starts from the definition of the US dollar index.The COT of US dollar is combined by the COT of Euro,Japanese Yen,British Pound,Canadian Dollar and Swiss Franc.Then the corresponding VAR model are established.This model is more effective to predict the future trend of the US dollar index futures price.In the second new method,The COT of US dollar is combined by the COT of Australian dollar,Mexican peso,New Zealand dollar,which are currencies in emerging country.The VAR model shows that the second new method is more effective and accurate to predict the future trend of the US dollar relative to the currencies of the emerging countries.This dissertation develops the quantitative analysis methods for the US dollar index by using of The COT data.Three methods are presented,which can effectively predict the trend of US dollar index.In this article,the research provides a new method and ideas for the forecast of the trend of the US dollar.And it is helpful for China's exchange rate warning and foreign exchange policy formulation.