Currency exchange rate predictability: The new power of Bitcoin prices

Abstract

We show that Bitcoin prices have surprisingly predictive power for nominal currency exchange rates, both in-sample and out-of-sample. The predictability follows from the fact that Bitcoin prices are forward-looking: Bitcoin efficiently incorporates expectations of currency exchange rates and their drivers, as exchange rates serve as a fundamental of Bitcoin. We examine the Bitcoin-based exchange rate prediction model in the autoregressive distributed lag (ADL) specification and the error correction specification. Forecasts based on both specifications outperform different benchmarks for some of the exchange rates. The outperformance is most pronounced at the daily horizon using the ADL model. Bitcoin-based forex trading strategies generate Sharpe ratio gains relative to the US risk-free rate and the carry trade. Bitcoin returns incorporate extra knowledge of future interest rate differentials after controlling for lagged exchange rate movements. Our result is inspiring for currency market participants, given the well-documented difficulty in exchange rate prediction.