Exchange Rate Forecasting with Support Vector Machines

Part II examines and analyzes the general ability of support vector machine (SVM) models to correctly predict and trade daily EUR/GBP, EUR/JPY, and EUR/USD exchange rate return directions. When computers are applied to solve a practical problem, it is usually the case that the method of deriving the required output from a set of inputs can be described explicitly. As computers are applied to solve more complex problems, however, situations can arise in which there is no known method for computing the desired output from a set of inputs, or where that computation may be very expensive. Forecasting financial time series events such as daily exchange rate directions, for instance, is a problem that is very relevant for the financial community and known to be very difficult in practice. We formally represent this problem as a classification task which is described by the linear separability problem. In the special case of finding whether two sets of points (namely exchange rate ups and downs) in general space can be separated, the linear separability problem becomes the binary classification problem whose most general form, the case of whether two sets of points in general space can be separated by k hyperplanes, is known to be NP-complete. It is generally believed that NP-complete problems cannot be solved efficiently.