We develop a four-state regime-switching model for optimal foreign exchange (FX) hedging using forward contracts. The states reflect four possible market conditions, defined by the direction and magnitude of deviation of the prevailing FX spot rate from its long-term trends. The model’s performance is tested for five currencies against pound sterling for various horizons. Our analysis compares the hedging outcomes of the proposed model to those of other frequently used hedging approaches. The empirical results suggest that our model demonstrates the highest level of risk reduction for the US dollar, euro, Japanese yen and Turkish lira and the second-best performance for the Indian rupee. The risk reduction is significantly higher for lira, which suggests that the proposed model might be able to provide much more effective hedging for highly volatile currencies. The improved performance of the model can be attributed to the adjustability of the estimation horizon for the optimal hedge ratio based on the prevailing market conditions. This, in turn, allows it to better capture fat？tail properties frequently observed in FX returns. Our findings suggest that FX investors tend to use short-term memory (focus more on recent price movements) during low market conditions (relative to trend) and long-term memory in high ones. It would be also useful to build a better understanding of how investor behaviour depends on market conditions and mitigate the adverse behavioural implications of short-term memory, such as panic.