Source: Moskowitz, T. J. & Grinblatt, M. (1999) · Journal of Finance 54(4), 1249–1290 · DOI 10.1111/0022-1082.00146
TL;DR
Industries themselves exhibit strong momentum: a strategy that is long the past-winning industries and short the past-losing industries earns about 0.43% per month (value-weighted, July 1963–July 1995) — essentially the same magnitude as individual-stock momentum. Once you control for industry momentum, much of the Jegadeesh-Titman (1993) individual-stock momentum weakens, so a large part of "stock momentum" is really momentum in the stock's industry.
What anomaly it documents
Predictor: trailing industry return.
Direction: positive — past winning industries keep winning over the next 6–12 months; losing industries keep losing.
Shape: cross-sectional (long winner industries / short loser industries); strongest at short horizons, decaying and reversing over multi-year windows. Industry momentum is distinct from, and partly subsumes, individual-stock momentum, size, value, and cross-sectional dispersion in mean returns.
OSAP predictor: IndMom (industry momentum).
How to construct it
Universe / building blocks: 20 value-weighted industry portfolios built from CRSP/Compustat (the paper uses 20; Ken-French 12- or 49-industry portfolios are the standard public proxy).
Sorting variable: trailing six-month industry return.
Portfolio: long the top-3 industries, short the bottom-3, equal-weighting the legs (the paper's headline is the (6,6) strategy — rank on past 6 months, hold 6 months); stock-level momentum comparisons use 30% breakpoints.
Rebalancing: monthly with overlapping 6-month holding periods.
ConvexPi replication: the 12 Ken-French industry portfolios, ranked on the trailing 12-month return skipping the most recent month, long the top 3 / short the bottom 3, rebalanced monthly.
Evidence and replication
Period
Sharpe / return
Source
IS (1963–1995, (6,6) value-weighted industry momentum)
0.43%/month, highly significant
this paper
IS (equal-weighted industry momentum)
0.81%/month ≈ 10.2%/yr (t = 7.71)
this paper
OOS (post-1999, ConvexPi 12-industry version)
Sharpe 0.31 (vs 0.47 pre-1999)
ConvexPi benchmark
The 0.43%/month is robust to DGTW size/BE-ME/momentum adjustment and survives controlling for individual-stock momentum. Out of sample it loses roughly a third of its in-sample Sharpe — milder decay than the size or value premia (consistent with McLean & Pontiff, 2016).
Why it might work
Slow information diffusion: industry-wide news (commodity prices, regulation, demand shocks) is incorporated gradually across the sector, so recent industry returns predict near-term returns.
Behavioural underreaction to sector fundamentals plus delayed sector rotation.
Risk-based readings are weaker than for value; the authors argue cross-sectional dispersion in mean industry returns is too small to explain the profits (rejecting a Conrad-Kaul explanation).
Limitations and risks
Turnover and transaction costs: monthly rebalancing of concentrated sector bets, though cheaper than single-name momentum.
Crash risk: like all momentum, vulnerable to sharp reversals after market bottoms.
Industry definitions matter: results shift with the number and construction of industries.
Crowding: widely known since publication; sector-rotation products may have compressed the edge.
Key references
Jegadeesh, N. & Titman, S. (1993) — Returns to Buying Winners and Selling Losers — Journal of Finance
Moskowitz, T. & Grinblatt, M. (1999) — Do Industries Explain Momentum? — Journal of Finance
Grundy, B. & Martin, J. S. (2001) — Understanding the Nature of the Risks and the Source of the Rewards to Momentum Investing — Review of Financial Studies
Asness, C., Moskowitz, T. & Pedersen, L. (2013) — Value and Momentum Everywhere — Journal of Finance
Daniel, K. & Moskowitz, T. (2016) — Momentum Crashes — Journal of Financial Economics
Reference replication on ConvexPi
An open, verified replication of this strategy is maintained at convexpi/replications. It recomputes the strategy from underlying building blocks and scores it out of sample (the McLean & Pontiff test):
