Intraday Foreign Exchange Rate Forecasting Using Sparse Grids

Abstract

We present a machine learning approach using the sparsegrid combination technique for the forecasting of intradayforeign exchange rates. The aim is to learn the impact oftrading rules used by technical analysts just from theempirical behaviour of the market. To this end, the problemof analyzing a time series of transaction tick data istransformed by delay embedding into a D-dimensionalregression problem using derived measurements from severaldifferent exchange rates. Then, a grid-based approach isused to discretize the resulting high-dimensional featurespace. To cope with the curse of dimensionality we employsparse grids in the form of the combination technique.Here, the problem is discretized and solved for acollection of conventional grids. The sparse grid solutionis then obtained by linear combination of the solutions onthese grids. We give the results of this approach to FXforecasting using real historical exchange data of theEuro, the US dollar, the Japanese Yen, the Swiss Franc andthe British Pound from 2001 to 2005.