Is Betting-Against-Beta a Construction Artefact?, Explained

Frazzini and Pedersen's Betting-Against-Beta (BAB) factor buys low-beta stocks, leverages that basket up to a beta of 1, and shorts high-beta stocks de-levered down to a beta of 1 — a market-neutral bet that low-beta stocks are "underpriced" because leverage-constrained investors overpay for high-beta stocks to get more market exposure without borrowing. The strategy showed strong historical returns. But a later critique argued that much of that return comes from how the portfolio is built, not from the risk story it tells.

BAB rebalances monthly using each stock's most recently estimated beta and re-levers the whole portfolio back to market-neutral every month. Novy-Marx and Velikov showed that this frequent re-leveraging, applied on top of illiquid small-cap stocks, generates a large chunk of the measured return through rebalancing mechanics rather than a persistent low-beta mispricing.

Where the critique bites

Two mechanical choices matter a lot. First, BAB weights stocks by the inverse of their estimated beta within each side of the portfolio, which overweights stocks with beta estimates near zero — often illiquid small caps with noisy beta estimates, since a beta estimated from thin trading is easy to get wrong in either direction. Second, monthly rebalancing to hold market beta constant means the strategy is constantly trading based on last month's beta estimate, and if that estimate is partly just noise, the rebalancing itself manufactures turnover-driven returns that look like alpha in a backtest but partly evaporate against realistic trading costs.

The debate isn't whether BAB has zero economic content — it's how much of the reported number survives once construction choices and costs are stripped out.

Worked example

In the original BAB paper, US equities show an annualized Sharpe ratio around 0.7-0.8. Novy-Marx and Velikov re-ran the strategy while excluding stocks below the NYSE 20th percentile of market cap (removing the illiquid small caps that get the largest inverse-beta weights) and found the Sharpe ratio fell by roughly half. They then applied realistic transaction costs to the monthly rebalancing turnover and found a further meaningful chunk of the remaining return disappeared, leaving a much more modest, though not zero, net effect.

What this means in practice

This doesn't mean BAB is fake — leverage-constrained investors bidding up high-beta stocks is a real, well-documented phenomenon. It means the number reported in the original paper mixes a real economic effect with construction choices (inverse-beta weighting concentrated in small caps, monthly re-leveraging) that inflate the measured return relative to what's actually capturable after costs. Anyone sizing a BAB-style allocation should test the strategy with capacity-realistic weighting and turnover assumptions rather than taking the headline Sharpe at face value.

This is a recurring pattern across the factor zoo: a real economic story can still have a backtested return that's substantially inflated by how the portfolio is constructed. "The effect is real" and "the reported magnitude is trustworthy" are separate claims, and BAB is one of the clearest documented cases where they diverge.

Is Betting-Against-Beta a Construction Artefact?

Where the critique bites

Worked example

What this means in practice

Related concepts

Practice in interviews

Further reading