Strategies / Low Volatility

Low Volatility

5 min read · Strategy guide

Buy the most boring quintile of the equity universe. The risk-adjusted premium is one of the most persistent anomalies in markets.

What it is

A low-volatility (or “min vol”) strategy ranks stocks by realised volatility — typically 60-day or 12-month — and overweights the bottom quintile. The empirical finding that drives the trade is striking: low-vol stocks systematically deliver returns close to the market average while taking dramatically less risk, which means their risk-adjusted returns dominate.

The seminal paper is Ang, Hodrick, Xing & Zhang (2006), which documented the anomaly in US equities and across international markets. It contradicts CAPM (which predicts higher beta = higher expected return) and is one of the most replicated cross-sectional findings in finance.

The core formula

for each stock in universe:
  vol_60d = std(daily_returns, 60) * sqrt(252)
rank universe by vol_60d (ascending)
hold bottom quintile, equal-weighted, rebalance monthly

# Variants:
# - vol over different windows (1y, 3y)
# - idiosyncratic vol (residual after factor model)
# - minimum variance optimisation across the universe

The cleanest version uses simple realised vol; the institutional version (e.g. MSCI Minimum Volatility) solves a constrained optimisation that minimises portfolio variance subject to sector caps. They give similar exposures.

When it works

Long-run: almost always on a Sharpe basis. Over rolling 10-year windows from 1970 forward, low-vol portfolios consistently outperformed the cap-weighted benchmark on risk-adjusted returns even when they lagged on raw return.

Acute: bear markets and corrections. Low-vol crushes in any drawdown — 2008, 2020, 2022 all saw min-vol portfolios drawdown 30–50% less than the broad market. This is the trade's defining characteristic: you give up upside in rallies to get a much shallower hole in drawdowns.

When it fails

Speculative rallies. 2020-Q2 through 2021 was brutal: low-vol underperformed the S&P by 15+ percentage points as profitless tech, meme stocks, and high-beta recovery names ran 5–10x. The 1999 dot-com peak had a similar dynamic. Whenever the market is paying for risk-on narrative rather than fundamentals, the boring-stocks portfolio gets left behind.

The other failure mode is sector concentration. A naive low-vol screen loads up on utilities and consumer staples — defensive sectors that all share the same rate-sensitivity. When rates rise sharply (e.g. 2022), the entire low-vol portfolio sells off together, losing the diversification benefit.

What to watch in the result card

Sharpe good, CAGR mediocre. This is the signature. Healthy low-vol backtests show Sharpe 0.7–1.0 with CAGR 7–10% — better risk-adjusted than the benchmark, lower absolute return. If your CAGR matches or exceeds the SPY, you're probably picking up something other than low-vol (often momentum hidden in the low-vol bucket).
MaxDD half the benchmark. The trade's reason for existing. If MaxDD is > 70% of the benchmark, the volatility filter isn't doing its job — usually because the lookback window is too short and the rankings are noisy.
Beta 0.6–0.8. Mechanically baked in by the construction. Anything above 0.9 means the universe doesn't have enough cross-sectional vol dispersion to differentiate the bottom quintile from the average.
Calmar > 1.0. Low-vol is one of the few strategies where Calmar reliably beats CAGR alone — modest returns, but a much shallower drawdown profile.

Common mistakes