Abstract

Betting against beta (BAB) can be seen as the combination of three investable component portfolios: Two cross-sectional components exploiting the beta anomaly attributable to stock selection and rank weighting scheme, and one time-series component with a dynamic net-long position due to “beta-parity.” Virtually all superior performance of BAB stems from the time-series component. The two cross-sectional components only provide hedging benefits in market downturns. The time-series component has modest portfolio turnover. Betting against correlation (BAC) yields similar findings, except that the two cross-sectional components in BAC outperform on a risk-adjusted basis. However, this effect arises purely from the positive association between firm size and stock correlation. Excluding micro-cap stocks, the performance of BAC shrinks more than that of BAB. Overall, only the time-series component remains as the robust source for the profits of the BA-type strategies.