We revisit the assumption that options-implied signals robustly predict cross-sectional stock returns. Although these signals robustly predict returns from 1996 to 2008, we find a marked decline in performance thereafter. Furthermore, we identify a look-ahead bias in the construction of these signals, arising from the non-synchronous observation of options and stock price data. Correcting for this bias by lagging the options data substantially reduces the signals’ predictive ability before 2008. These findings question the efficacy of options-implied information at forecasting monthly stock returns and the robustness of trading strategies predicated on this information.