Svm And Chaos Theory-based Exchange Rate Forecast

With the economic globalization of the world and the integrated development of the financial field, the exchange rate has more and more important influence on the economical operation, trade contacts, and international status of various countries. So, it is obviously very important to analysis prediction of exchange rate fluctuate. Meanwhile, the economic system is a complicated huge system, and there are intricate relations between each economic variable inside it. The fluctuation of the exchange rate is even more like this. After the breakdown of the Breton-Woods System, exchange rate's fluctuation becomes more and more frequent and unstable. The forecast of exchange rate also becomes more and more difficult. And the traditional theories of exchange rate determination including"purchasing power parity","the balance of international payments of exchange rate determination"and"the method of asset market analysis"based on linear model didn't explain the rule of exchange rate's change well. Since the 1980s, more and more economists and mathematician have been exploring and looking for some nonlinear methods to explain the complicated fluctuation of exchange rate phenomenon and give the effective suggestion to the adjustment of the exchange rate. So, researching the exchange rate in terms of nonlinearity, have very wide space and great realistic meaning.In the papers, theories and methods of chaos and nonlinearity is used to test exchange rate data. Thus daily series of rate of US dollars to one Euro were analyzed using correlation dimension (2.559) and the max Lyapunov exponent (0.0686>0) as well as its surrogate data. Apparent difference of test statistics from the daily series and its surrogate data rejects the null hypothesis that the daily return series is derived from a linear system,which confirms that the fractal dimension of exchange rate time series stems from an inner nonlinear dynamics. Then the combined method of the reconstructed phase space and support vector machines is used to forecast the exchange rate. The result showed that the model above can track the short-term exchange rate's change well with high accurate rate and speed.