Source: Asness, C. S., Moskowitz, T. J. & Pedersen, L. H. (2013) · Journal of Finance 68(3), 929–985 · doi:10.1111/jofi.12021
TL;DR
Value and momentum are not equity quirks: both earn significant premia in every one of eight diverse markets and asset classes — individual stocks in the US, UK, Europe, and Japan, plus country equity index futures, government bonds, currencies, and commodities. Value strategies are positively correlated with each other across asset classes, momentum strategies likewise — pointing to common global factors — yet value and momentum are negatively correlated (≈ −0.60) within and across classes. A 50/50 value+momentum combination across all asset classes achieves an annualized Sharpe ratio of 1.42. Global funding-liquidity risk is a partial common driver.
What anomaly it documents
Predictor: two signals studied jointly — value (cheap minus expensive) and momentum (recent winners minus losers) — across markets, not US stocks alone.
Direction: high value (cheap) and high momentum (winners) earn positive premia in each asset class.
Shape: monotone across value/momentum-sorted portfolios; the joint structure shows strong positive comovement among same-style strategies across classes and strong negative value-vs-momentum correlation.
Value signal: for individual stocks, book-to-market BE/ME (book lagged six months, current market value). For other classes, analogous "long-run reversal" cheapness measures (e.g., negative 5-year return; real exchange rates for currencies; 5-year yield change / yield-curve measures for bonds; spot-vs-past for commodities).
Momentum signal: the past 12-month cumulative return skipping the most recent month (MOM2-12), used uniformly across all asset classes to avoid the 1-month reversal.
Within each market form long-short (P3 − P1) spread and signal-weighted factor portfolios; then combine value and momentum 50/50.
Universe: large/liquid securities; individual-stock sample January 1972 – July 2011 (ADRs, REITs, financials, closed-end funds, sub-$1 stocks excluded).
Evidence and replication (IS/OOS)
In-sample (1972–2011), per Table I:
Significant value and momentum premia in all eight markets/asset classes.
Value–momentum correlation ≈ −0.60, so combining them sharply raises Sharpe ratios; in every market the combination beats either alone.
The all-asset-class 50/50 value+momentum combination Sharpe ≈ 1.42 — a hurdle far above the US-equity value/momentum premia that existing models already struggle with.
A three-factor global model (market, global value, global momentum) prices the strategies; value loads negatively and momentum positively on global funding-liquidity risk.
This is the original source; no separate OOS replication is reported.
Why it might work
Common global risks, notably funding-liquidity risk (Brunnermeier–Pedersen), whose role grew over the sample; value does well and momentum poorly in worsening-liquidity states (the 50/50 combo is roughly liquidity-immune after 1998).
The robust negative value-momentum correlation suggests they capture complementary mispricings/underreaction-overreaction, challenging US-equity-only behavioral, institutional, and rational stories.
Limitations and risks
Implementation costs and shorting constraints differ sharply across asset classes; some value proxies (e.g., the 5-year-yield-change bond measure) are weak (Sharpe ~0.18) and sensitive to definition.
Both legs carry crash risk; the combination mitigates but does not eliminate it.
Key references
Asness, C., Moskowitz, T. & Pedersen, L. (2013) — Value and Momentum Everywhere — Journal of Finance
Fama, E. & French, K. (1992) — The Cross-Section of Expected Stock Returns — Journal of Finance
Jegadeesh, N. & Titman, S. (1993) — Returns to Buying Winners and Selling Losers — Journal of Finance
Provenance: verified/generated from the paper's full text.
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):
