Return Predictability and Trends
Abstract
The gambler's fallacy predicts that trends bias investor expectations. We find that trends in prior earnings surprises are a robust predictor of risk-adjusted returns. Moreover, the underreaction of investors to trends determined exclusively by the sign of prior earnings surprises provides empirical support for the gambler's fallacy. The return predictability of trends is not attributable to the magnitude of earnings surprises nor the autocorrelation in firm-level earnings surprises. Instead, trends explain more than half of the post-earnings announcement drift in our sample. Therefore, this cross-sectional return anomaly has a significant time series component that is consistent with the gambler's fallacy.
This paper has been withdrawn.