Forecasting directional changes in the FX markets

Most of existing studies sample markets' prices as time series when developing models to predict market's trend. Directional Changes (DC) is an approach to summarize market prices other than time series. DC marks the market as downtrend or uptrend based on the magnitude of prices changes. In this paper we address the problem of forecasting trend's direction in the foreign exchange (FX) market under the DC framework. In particularly we aim to answer the question of whether the current trend will continue for a specific percentage before the trend ends. We propose one single independent variable to make the forecast. We assess the accuracy of our approach using three currency pairs in the FX market; namely EUR/CHF, GBP/CHF, and USD/JPY. The experimental results show that the accuracy of the proposed forecasting model is very good; in some cases, forecasting accuracy was over 80%. However, under particular settings the accuracy may not outperform dummy prediction. The results confirm that directional changes are predictable, and the identified independent variable is useful for forecasting under the DC framework.