Narasimhan Jegadeesh & Sheridan Titman
The 12-1 momentum factor: buy last year's winners, but skip last month
Momentum is the most replicated anomaly in the academic literature: what has outperformed over the past year tends, on average, to keep outperforming for months. The definition that survives replication measures that year from twelve months ago to one month ago — and the skipped month is not a detail. Returns over a few weeks tend to reverse, and a version that includes them systematically buys names that have just spiked and are about to give it back.
- Style
- Factor & anomaly
- Approach
- Mechanical
- Difficulty
- Intermediate
- Horizon
- Position (weeks to months)
- Holding period
- Three to twelve months
- Time needed
- An hour a month
- Markets
- Single stocks · Sector ETFs · Country ETFs
- Source
- Returns to Buying Winners and Selling Losers, Journal of Finance (1993) — Narasimhan Jegadeesh & Sheridan Titman
The rule set
- Measure the return from twelve months ago to one month ago — never the plain trailing year
- Require that 12-1 return to be clearly positive, not merely above zero
- Require the name to be outperforming the market over the same window
- Rebalance on a fixed schedule rather than holding until something breaks
- Leave early if the 12-1 measure turns negative before the rebalance date
What makes it distinctive
- The most replicated anomaly in finance — documented across decades, countries and asset classes
- Skipping the most recent month is what separates the factor from naive performance chasing
- Needs no fundamentals at all: price history is the entire input
When it works
Trending markets with clear leadership — a sector or theme that persists for quarters rather than weeks, so that the past year's ranking is still informative about the next few months.
When it fails
Momentum crashes are its signature risk: at sharp market bottoms the previous losers rally hardest, and the factor can lose in weeks what it made in years. It also turns over constantly, so costs and taxes matter more than for any other factor.
How a decision moves through it
Input
Daily bars for the candidate and for the market
Roughly two years of history per name, plus the same for a market benchmark. The benchmark is not decoration — half of the entry condition is a comparison against it.
Measure
The 12-1 return
Price one month ago divided by price twelve months ago, minus one — in bars, the close 21 sessions back against the close 250 sessions back. The most recent month is left out on purpose: short-horizon returns reverse, and including them buys the names that are already rolling over.
Decide
Strong on both an absolute and a relative measure
The 12-1 return must be above +10%, and relative strength against the market over the same year must be positive. The published factor ranks the whole universe and buys the top decile; a one-symbol-at-a-time backtest has no rank, so the rank becomes two thresholds — with the consequence that in a market where everything falls, the real top decile is always occupied but these thresholds go empty.
Act
Buy, and add nothing to the decision
There is no judgement stage. A name that clears both measures is bought; a name that misses either is not, however good its story.
Size & protect
Two exits: momentum gone, or the calendar
Sell early if the 12-1 measure falls to zero or below. Otherwise the position is closed and the list rebuilt after six months of bars — the holding period the original study reports. The factor is rebuilt on a schedule, never held until something breaks.
The measurement window is odd on purpose
Ask most people to measure a stock's momentum and they will quote the trailing twelve-month return. The factor that has survived three decades of replication measures something slightly different: the return from twelve months ago to one month ago. The most recent month is thrown away.
That is because price behaviour has two speeds. Over the past year, winners tend to keep winning — continuation. Over the past few weeks, movers tend to snap back — reversal. The 12-1 window keeps the first effect and cuts out the second.
| Window | What it captures | What buying it gets you |
|---|---|---|
| Last 12 months, including last month | Continuation plus a recent spike | Names that just jumped and tend to give some back |
| 12 months ago to 1 month ago | Continuation only | The documented momentum factor |
| Last month alone | Mostly reversal | The opposite trade — recent movers snapping back |
Nothing else about the system is exotic. One return calculation, one comparison against the market, a schedule. The entire edge, such as it is, lives in measuring the right window and acting on it without editing.
It is Dual Momentum's relative leg, run across the whole market
If you know Dual Momentum (GEM), you already know this factor's mechanism. GEM's first step compares the 12-month return of two index sleeves and holds the stronger — relative momentum between two assets. The 12-1 factor is the same comparison made cross-sectional: rank every stock in the universe on the same measure and own the top of the list. The academic literature came first; GEM is the two-asset, retail-sized application of it.
The inheritance runs both ways. GEM adds an absolute gate — is the winner beating cash? — precisely because pure relative momentum has no exit and rides bear markets down. The cross-sectional factor has the same blind spot at stock level, which is why the shipped version pairs the ranking measure with an absolute threshold and a hard exit rather than trusting the ranking alone.
What the shipped version changes, and why it matters
The published factor is a portfolio construction: rank the market, hold the top decile, always. The version in this library tests one symbol at a time, and a single symbol has no rank — so the rank becomes two absolute thresholds: a 12-1 return above +10%, and relative strength against the market above zero.
The substitution changes behaviour in exactly one regime. In a market where everything is falling, the true top decile is still occupied — someone is always least bad — while the thresholds simply go empty. The threshold version therefore sits out broad declines that the academic factor rides down, which makes it more conservative than the paper it cites.
Five ways into this system
- The 12-1 calculation: one window, two thresholds, two exitsOne return calculation with a deliberate hole in it, two conditions that must both pass, and an exit that runs on the calendar.7 min read
- A cohort bet, not a conviction bet — and turnover is the real costThe factor's claim is about the average of a list, not about any name on it. Sizing follows directly: many positions, equal weights, and a hard look at what the turnover costs.6 min read
- Anywhere with a price history: stocks, sectors, countriesThe factor has been documented in almost every liquid market with a price history. What it demands is not insight but coverage: a measure computed across everything, on schedule.5 min read
- The momentum crash: the factor's worst days are all the same dayOne rare, violent, structural failure; one slow leak; and the question mark that hangs over every anomaly with a paper trail.7 min read
- 12-1 momentum for beginners: what it means to buy the winnersBuy what has done well over the past year — with one strange-looking adjustment that turns out to be the whole idea.6 min read
The ideas behind it
This system assumes you already know these. Each one is explained from scratch in Investing 101.
Compare with
These are documented methods described for study. Nothing here is investment advice, a recommendation, or a claim about future returns — every system on this page has losing periods, and the pages say where.
Reading about a system is not having one.
Plutux is where you write your own rules down, test them against real data, and keep the record your memory would otherwise rewrite. Join the waitlist for early access.