Novy-Marx shows that standard momentum is driven primarily by performance 7 to 12 months before formation (intermediate-horizon past returns), not by the most recent 2 to 6 months (recent past returns). "Echo" momentum from the intermediate horizon predicts returns; recent performance does not, and can even reverse. This challenges behavioral and risk models that treat momentum as a response to recent news.
What anomaly it documents
Predictor: intermediate past performance (returns from months t-12 to t-7).
Shape: the predictive power sits in the intermediate window, not the recent (t-6 to t-2) window.
OSAP predictor: IntMom (intermediate momentum).
How to construct it
Sorting variable: cumulative return from 12 to 7 months before formation.
Universe: NYSE/AMEX/Nasdaq common stocks.
Portfolio formation: decile sorts on intermediate past return.
Long / short: long intermediate winners, short intermediate losers.
Weighting: value-weighted.
Rebalancing: monthly.
Evidence and replication
Period
Notes
Source
IS (1927–2010)
intermediate-horizon returns drive momentum; recent returns do not
this paper
OOS (post-2012)
"echo" pattern robust across markets
post-publication
OSAP (IntMom)
replicates
Chen & Zimmermann 2022
Why it might work
Underreaction to persistent information: intermediate performance better proxies for slowly-diffusing fundamental news.
Recent-return contamination: the t-6..t-2 window mixes in short-term reversal and microstructure noise.
Limitations and risks
Still momentum risk: inherits momentum crashes and turnover.
Definition dependence: results hinge on the exact lookback windows.
Overlap: highly correlated with conventional momentum in practice.
Key references
Novy-Marx, R. (2012) — Is Momentum Really Momentum? — JFE — DOI: 10.1016/j.jfineco.2011.05.003
Jegadeesh, N. & Titman, S. (1993) — Returns to Buying Winners and Selling Losers — JF
Provenance: generated from the paper's abstract and metadata, not full text; sample periods and replication notes are indicative — verify against the source.
