Abstract We provide a novel explanation for the empirical failure of the capital asset pricing model (CAPM) despite its widespread practical use. In a rational-expectations economy in which information is dispersed, variation in expected returns over time and across investors creates an informational gap between investors and the empiricist. The CAPM holds for investors, but the securities market line appears flat to the empiricist. Variation in expected returns across investors accounts for the larger part of this distortion, which is empirically substantial; it offers a new interpretation of why “betting against beta” (BAB) works: BAB really bets on true beta. The empiricist retrieves a stronger CAPM on days when public information reduces disagreement among investors.