An application of adjusted AR model and Markov chains in the forecast of exchange rate of RMB

This article used statistical data of recent two years. Under the condition that RMB satisfies Markov property, we used Markov chain to compute the frequency matrix and transition probability matrix and moreover, we predicted the future trend of exchange rate. We also simulated the original data using AR model, and added residual factor that is from analysis of Markov Chain. Finally, we got a model which is better than traditional ARIMA model. According to the errors of our result, we analyzed the reason of these errors and its effect to exchange rate, and proposed suggestions to improve the method.