Realised volatility and industry momentum returns

The study examines whether realised volatility predicts and enhances industry-level momentum returns. Momentum strategies, which buy past winners and sell past losers, are pervasive across markets and asset classes, but linking volatility to industry momentum has been limited. Given the growing prevalence of sector/industry investing and evidence that industry volatility plays a significant role in stock and market volatility, the authors explore how industry-level realised volatility, including its idiosyncratic and systematic components, relates to momentum profits. The purpose is to determine if higher past volatility is associated with larger subsequent industry momentum returns and whether this relationship holds after controlling for common risk factors. The study also evaluates a volatility-managed industry momentum strategy to assess whether managing volatility can improve performance and reduce crash risk. This work is important as it extends firm-level evidence to industry portfolios, which are often undiversified and may better capture idiosyncratic risks shared across constituent firms.

Prior research documents pervasive momentum profits across markets and assets (e.g., Asness et al., 2013; Jegadeesh and Titman, 1993) and shows momentum portfolios often exhibit higher volatility (Barroso and Santa-Clara, 2015). Individual stock volatility is related to industry and market volatility (Campbell et al., 2001), with industry volatility often more influential than market volatility (Black et al., 2002; Morana and Sawkins, 2004; Ferreira and Gama, 2005). Firm-level studies link volatility—particularly idiosyncratic volatility—to momentum returns (Arena et al., 2008; Liu et al., 2016; Wang and Xu, 2015), and recent work shows industry-adjusted volatility relates to firm momentum (Badreddine and Clark, 2021). However, industry-level momentum linked directly to industry volatility has received limited attention, with an exception in China’s market via oil volatility (Chen et al., 2017). Given industry portfolios’ comovement and relative lack of diversification, examining realised, systematic, and idiosyncratic volatility at the industry level can clarify volatility’s role in momentum profits.

Data: Daily value-weighted returns for 48 US industry portfolios from the Kenneth French Data Library, spanning July 1969–June 2021. Common risk factors (market, SMB, HML, RMW, CMA) are from the same source. Returns are transformed to daily log returns. Trading strategy: Sequential double sorts are conducted each month. First, industries are sorted into three terciles by realised volatility computed over an estimation window E months (E ∈ {1, 3, 6, 9, 12, 24}). Within each volatility tercile, industries are sorted into four groups by their mean returns over the same estimation window. Momentum portfolios (winner minus loser, equal-weighted) are then formed based on the second sort and held for H months (H ∈ {1, 3, 6, 9, 12, 24}). Raw and risk-adjusted returns are computed for overlapping holding periods as applicable. Volatility measurement: Realised volatility is the standard deviation of daily log returns within the estimation window. Total realised volatility is further decomposed into idiosyncratic and systematic components by estimating CAPM within each estimation window: excess industry returns are regressed on the market excess return; the residuals represent idiosyncratic returns whose realised volatility defines idiosyncratic volatility. Systematic volatility is computed as total minus idiosyncratic realised volatility. As a robustness check, drift-adjusted realised volatility (demeaning daily returns within the window before computing volatility) is also used. Returns measurement: Estimation- and holding-period log excess returns are computed by aggregating daily excess log returns within the period to form average monthly excess returns. Risk adjustments: Portfolio alphas are estimated using CAPM, Fama-French three-factor, and five-factor models applied to holding-period monthly excess returns to assess abnormal performance after controlling for market, size, value, profitability, and investment factors. Volatility-managed momentum: Following Barroso and Santa-Clara (2015), an industry momentum portfolio is formed by ranking industries on past 12–2 month returns into six groups (long highest, short lowest). The resulting momentum return is scaled each month by the ratio of a fixed target annualised volatility (~13%) to the realised volatility of the momentum portfolio computed from daily returns over the past six months, to obtain a constant-volatility strategy.

- Realised total volatility and momentum: Industry momentum returns generally increase with higher past realised volatility across most estimation/holding horizons, except the shortest 1/1 case. For example, in the 12/1 strategy, the high-volatility group yields a 1.004% monthly momentum return (significant at 5%). Sharpe ratios are typically higher in high-volatility terciles, reinforcing the positive relationship. - Risk-adjusted performance: After controlling for CAPM, Fama-French three- and five-factor models, high-volatility terciles still exhibit larger and more significant momentum alphas across most horizons (e.g., strong t-statistics for 12/12, 24/12, 24/24), indicating common factors do not explain the volatility-momentum link. - Decomposed volatility: Both idiosyncratic and systematic realised volatility positively relate to industry momentum returns. High idiosyncratic-volatility groups produce higher raw and adjusted momentum returns across most horizons (except 1/1). High systematic-volatility groups also show higher returns and Sharpe ratios across most horizons; 24/24 results are weaker/insignificant in some cases. - Volatility-managed industry momentum: Scaling momentum to a target volatility significantly improves performance and reduces tail risk. Compared to unscaled momentum (mean return 0.833%, t=2.479; skewness -0.254; kurtosis 3.036; Sharpe 0.167), the volatility-adjusted strategy delivers a 9.611% mean return (t=5.490), with skewness -0.015, kurtosis 0.546, and Sharpe 0.222. - Robustness (drift-adjusted volatility): Results are qualitatively similar when using drift-adjusted volatility. The strongest momentum effect occurs for the 9/9 strategy with a 0.797% monthly return for the high drift-adjusted volatility group.

The findings show that industries with higher past realised volatility yield stronger subsequent momentum profits, and this relationship persists after controlling for standard risk factors, implying that conventional factor models do not capture the volatility-momentum nexus at the industry level. Decomposition demonstrates that both systematic and idiosyncratic volatility matter for industry momentum, consistent with industry portfolios being relatively undiversified and thus retaining idiosyncratic risks that influence returns. The enhanced alphas for high-volatility groups suggest momentum gains represent compensation for bearing higher volatility risk. Potential mechanisms include underreaction to information (especially in more volatile environments), limits to arbitrage that discourage risk-averse arbitrageurs from correcting mispricing in high-volatility industries, and stronger price persistence due to firm- or industry-specific news. Volatility management further improves industry momentum performance while mitigating crash risk, aligning with the interpretation that risk-managed exposure to volatile momentum can enhance risk-adjusted returns.

Using 48 US industry portfolios (1969–2021), the study shows realised volatility predicts and enhances industry momentum returns across multiple horizons. The relationship strengthens after adjusting for CAPM and Fama-French factors, indicating standard risk factors do not explain the effect. Both idiosyncratic and systematic components of realised volatility positively relate to momentum profits, consistent with industry portfolios retaining undiversified risks. A volatility-managed industry momentum strategy markedly improves returns and reduces left-skewness and kurtosis relative to unscaled momentum. Results are robust to an alternative, drift-adjusted volatility measure. Future research could examine cross-country industry portfolios, alternative decomposition frameworks beyond CAPM, microstructure influences at daily frequencies, and interactions with liquidity and transaction costs.