Directional Forecast with Dynamic Volatility and Time Regime Classification: An Evaluation on EUR/USD

Abstract

Predicting the directional movement of price is of utmost importance to remain profitable in financial markets. This is particularly important for the Foreign Exchange (Forex) market owing to its high volatility which is affected by both internal and external market factors. Volatility is considered to be one of the significant obstacles for accurate directional prediction in the forex market. Although research efforts have been made to couple dynamic volatility with trend prediction models, most previous studies have been conducted subjected to unrealistic assumptions pertaining to volatility which have led to unsatisfactory results. This indicates that traders still face serious challenges when deriving more accurate predictions on the direction of the forex market, in order to remain profitable in the market. This study presents a directional prediction model for the forex market incorporating the dynamic volatility inherent to the market using intraday data. This was achieved by identifying different volatility levels that exist in the market using techniques such as change point analysis and clustering while Support Vector Machines (SVMs) are utilized to capture the directional movement of the market. The proposed solution is validated using different metrics and the results indicate that it outperforms the standard trend prediction method subjected to the nature of the input variables used when constructing the SVM models.