Can exchange rate models outperform the random walk? Magnitude, direction and profitability as criteria

Abstract

While many explanations have been put forward for the failure of exchange rate models to outperform the random walk in out-of-sample forecasting, a simple explanation is the use of measures of forecasting accuracy that depend entirely on the magnitude of the forecasting error. By using simulated data representing the forecasts of eight models, it is demonstrated that the random walk can be outperformed if forecasting power is judged by measures of direction accuracy, by adjusting the root mean square error to take into account direction accuracy, and by using the risk-adjusted return obtained from a trading strategy based on the forecasts.