We investigate whether the well-known positive association between past stock returns and analysts' earnings forecast revisions differs for stocks that have experienced extreme positive (or negative) price changes. We document an asymmetry in this association depending on whether a stock experienced consecutive rounds of capital gains (winner) or losses (loser): the association is strong for losers, but weak for winners. We also find that earnings forecasts are less accurate for analysts who treat winners and losers differently from moderately-performing stocks. However, the degree of such aberrant reactions to winners and losers and the consequent detrimental effect on the forecasting accuracy are less acute for analysts who have more forecasting experience with winners and losers. Finally, consistent with the efficient market hypothesis (EMH), we find that investors discount earnings forecast revisions by analysts who treat winners and losers differently, particularly after 2003, coinciding with the Global Settlement that aims to improve the integrity of analyst research and investors' financial literacy.