Emotions Matter : Sentiment and Momentum in Foreign Exchange

We introduce news sentiment as a variable that can explain and predict subsequent changes in the USD/EUR exchange rate, and therefore close a gap in the foreign exchange literature. By applying the concept of frequency filtering from the domain of electrical engineering, we show an innovative way of filtering for noise in both news sentiment, but also in price momentum. We find that news sentiment is not correlated to price momentum, and that trading strategies based on news sentiment achieve around twice as high information ratios (up to 0.9) than with trading strategies based on price momentum