Garch To Predict Currency Index Dynamic Portfolio Analysis

According as the basic form of the exchange rate and USD index, this paper proposes the concepts of currency index and implicit world base currency, as the choice of assets and the basis of valuation while we establish currency portfolio in foreign exchange market. Furthermore, this paper focuses on two questions:firstly, in the dynamic time scale, which parameters the assets’weight of currency portfolio relying on; secondly, can the mean-variance model be improved, if the GARCH model predictors be joined. Then, this paper does an empirical research with eight major currencies of the foreign exchange market.In the introduction part of this paper, we give a brief introduction of portfolio theory in the securities market, especially in the foreign exchange market, then introduce the currency index, the implicit base currency and it’s possibility and necessity to be the currency portfolio base currency. The second chapter reviews the modern portfolio theory’s development of60-years from the establishment till now and research status. Multi-period portfolio model and the optimal investment consumption model are described in detail. Then this paper builds a single portfolio of multi-currency index model applying the classic Markowitz mean-variance model. Afterwards, two exogenous variables the moving window’s width L and the frequency of portfolio re-adjust T are introduced in the model which has multi-phase conditions, and this paper explores the suitable range of these two variable’s values in general case. Based on the reliability requirements of the mean-variance model’s input data, this paper attempts to introduce predictable factors instead of the expected rate of return using the simple average, so the ARCH model’s prediction power is introduced in. In the chapter IV of this article, we do the ARCH effect test of eight currency indexes after the introduction of the origin of ARCH models, the test appears that eight currency indexes’fluctuations have obvious conditional heteroscedasticity, and then we found that application of GARCH (1,1) model can solve this problem, which can better fit the historical price data for currency indexes. On this basis, the predictive power of the GARCH model is introduced in the currency indexes portfolio model. Using the predicted logarithm price data of currency indexes, we can get these predicted rate of return further. So, we rebuild the currency indexes portfolio model and do the empirical research. Finally, we summarize the whole paper, then do some extended discussion of the multi-period portfolio model, and the portfolio model by adding predictors to.