Using a sample of 97 stock return anomalies, we find that anomaly returns are 7 times higher on earnings announcement days and 2 times higher on corporate news days. Anomaly variables also predict analyst earnings forecast errors: analysts' earnings forecasts are too low for anomaly-longs, and too high for anomaly-shorts. We develop and conduct several unique data mining tests, and find that data mining cannot explain our findings. Our results support the view that anomaly returns are the

more »

... of biased expectations, which are at least partially corrected upon news arrival.