AbstractThis study provides evidence that the price‐earnings (PE) ratio effect is not homogeneous across firms with similar PE ratios. Instead, firms with the lowest PE ratios and those with the lowest expected annual earnings per share outperform all other groups in January. These results can be partially attributed to security analysts consistently underestimating reported earnings of firms with the lowest level of expected earnings and the lowest PE ratios. A negative October effect is also found for the same‐firms, which appears to be caused by downward revisions in analysts' forecasts between September 16 and November