I remember running my first cross-sectional regression in graduate school and staring at the results in disbelief. The coefficient on volatility was negative, which meant higher risk was associated with lower subsequent returns. This directly contradicted the capital asset pricing model I had just spent months studying. What drew me in was not just the statistical significance but the persistence of this pattern across different time periods, countries, and asset classes. The anomaly seemed counterintuitive because we are trained to believe that rational investors demand compensation for bearing additional risk. Yet here was evidence that the riskiest assets were delivering the worst risk-adjusted outcomes. I realized that either markets were profoundly irrational in a systematic way, or our models were missing something fundamental about how real investors behave. That tension between theory and evidence became the driving force of my research career.

The behavioral explanation centers on what we call the lottery ticket effect. Many investors, both retail and institutional, are drawn to assets with high upside potential even when the expected value is poor. High volatility stocks resemble lottery tickets because they offer a small chance of extraordinary gains. This preference creates excess demand that drives prices up and expected returns down. Another crucial factor is representativeness bias, where investors extrapolate recent performance too far into the future. A stock that has recently surged attracts attention regardless of its fundamentals. Analysts also play a role through their tendency to issue overly optimistic forecasts for glamorous growth companies, which tend to be more volatile. The combination of these biases creates a systematic overvaluation of risky assets that arbitrageurs cannot easily correct due to short-selling constraints and the fear of momentum continuing against rational positions.

Several institutional frictions make arbitrage costly and risky. First, many institutional investors face explicit or implicit benchmarks that discourage significant deviations from market capitalization weights. A pension fund manager who underweights volatile growth stocks might outperform over the long term but risks underperforming during speculative bubbles and losing clients or employment. Second, short-selling high volatility stocks is expensive and sometimes impossible due to limited borrow availability. The stocks most likely to be overpriced are also the hardest to short. Third, arbitrage requires capital and patience, and volatile positions can move against rational investors for extended periods. As John Maynard Keynes famously noted, markets can remain irrational longer than you can remain solvent. These structural barriers mean that even when sophisticated investors recognize the mispricing, they may be unable or unwilling to trade against it in sufficient size to eliminate the anomaly.

The standard approach begins by sorting stocks into portfolios based on their historical volatility, usually estimated from daily returns over the past year or month. Researchers then form quintiles or deciles, with the lowest volatility stocks in one group and the highest in another. The critical test compares the subsequent returns of these portfolios, typically on a risk-adjusted basis using metrics like the Sharpe ratio or alpha from factor regressions. More sophisticated studies use cross-sectional regressions where volatility is one of many predictive variables, allowing researchers to control for size, value, momentum, and other known effects. Fama-MacBeth regressions are particularly popular because they handle the time-series and cross-sectional dimensions simultaneously. Robustness checks involve varying the estimation window, using different volatility measures such as idiosyncratic versus total volatility, and testing across international markets. The most convincing studies also examine transaction costs and liquidity to ensure the anomaly is implementable in practice, not just a statistical artifact.

The most important insight is that risk is not a single dimension and does not always translate linearly into expected returns. Investors should examine what type of risk they are being compensated for and whether that compensation is adequate. The evidence suggests that volatility risk, at least as traditionally measured, is not priced in a way that rewards bearing more of it. This does not mean investors should avoid all risk, but rather that they should be selective about which risks they accept. Another practical consideration is that low volatility approaches tend to have sector and regional tilts, often overweighting utilities and consumer staples while underweighting technology and biotech. Investors need to understand these implicit bets and decide whether they align with their broader objectives. Finally, the anomaly has received considerable attention in recent years, which may have reduced its magnitude. Past performance does not guarantee future results, and any strategy based on historical patterns should be implemented with appropriate skepticism and ongoing monitoring.