What Is the 12-1 Momentum Signal?

Why the most recent month gets dropped — and why that omission is the signal's most important design decision.

The 12-1 momentum signal is one of the most replicated and debated constructs in quantitative equity research. At its core it is simple: rank stocks by their cumulative return over the past twelve months, but exclude the most recent month from the calculation. That single omission — dropping month t-1 entirely — is not a rounding convention or a data quirk. It is a deliberate correction for a well-documented short-term effect that runs in the opposite direction to medium-term momentum. Understanding why that month gets dropped is the clearest path to understanding the signal itself. Practitioners who treat the 12-1 momentum signal as a black box and implement it naively — including the most recent month, using raw prices instead of corporate-action-adjusted returns, or ignoring rebalance timing — tend to underperform the academic benchmarks considerably.

The signal sits at the center of factor investing, and it feeds directly into products like the Quant Galore Momentum Index, which operationalizes these principles at scale. The mechanics described below apply whether you are constructing a standalone momentum screen or incorporating the signal into a broader multi-factor model.

Defining the 12-1 momentum signal precisely

The formation window runs from month t-12 to month t-2, inclusive — eleven calendar months of return data, with one month of deliberate silence at the near end. The return used is the compounded cumulative return across those eleven months, not the sum of monthly returns. Compounding matters: the difference between arithmetic and geometric accumulation widens as the holding period lengthens and return dispersion rises.

Formally, if P denotes the adjusted closing price and the current month is t, the formation-period return is:

Momentum_12-1 = (P_t-2 / P_t-13) − 1

The price at the start of month t-12, which is the price at the end of month t-13, anchors the calculation. The price at the end of month t-2 closes it. Month t-1 is excluded — the stock's return from the end of month t-2 to the end of month t-1 is not in the window. The resulting cross-sectional rankings tell you which stocks have been the strongest and weakest performers over the medium term, without contamination from the most recent four weeks.

The short-term reversal effect and why it contaminates the signal

The exclusion of the most recent month is a response to a separate and well-established empirical phenomenon: short-term reversal. Stocks that have been strong performers over the trailing four weeks tend to underperform over the following four weeks, and stocks that have been weak performers tend to mean-revert upward. This is the opposite of what momentum predicts. If you include month t-1 in the formation window, you are blending a medium-term continuation signal with a short-term reversal effect, and the two partly cancel each other.

The mechanism behind short-term reversal is microstructural rather than behavioral. Several forces contribute. Market makers widen bid-ask spreads in response to order imbalance, and the subsequent normalization of spreads creates a return reversal. Temporary liquidity demand — a large institutional investor selling a block position — can move price away from fundamental value; once the selling pressure clears, price rebounds. Thin-float stocks are particularly susceptible to this kind of short-term noise. At the portfolio level, if you are ranking stocks on trailing twelve-month returns without the skip month, stocks that happened to have a strong month t-1 due to liquidity effects rather than fundamental momentum will appear at the top of your ranking. You buy them, the short-term reversal kicks in, and your portfolio immediately faces a headwind from the very stocks you have overweighted.

The foundational work on medium-term momentum by Jegadeesh and Titman documented this contamination and recommended the skip-month design specifically to remove it. The skip month is not an afterthought — it was part of the original empirical design.

How to compute the signal in practice

The inputs require care. Prices must be adjusted for all corporate actions: splits, reverse splits, rights offerings, and special dividends. Using unadjusted prices in the denominator creates phantom return calculations that rank stocks incorrectly — a stock that split two-for-one midway through the window looks like a 50 percent decline in price if you ignore the adjustment. Dividends are typically included; total return data is preferable to price-only data, though the difference matters more over longer windows and for high-yielding sectors.

Point-in-time pricing matters for realistic backtesting. End-of-month prices are the standard, but the relevant question is which price was actually available to a portfolio manager at the rebalance date. Prices that appear in a database but were not yet published or were subject to revision introduce look-ahead bias.

New listings without full twelve-month histories create another gotcha. A stock listed six months ago has no valid P_t-13. The standard practice is to require a minimum formation-period history — often ten or eleven months — before including a name in the cross-section. Failing to screen for this inflates the apparent universe with names whose ranking is meaningless or misleading.

For building a momentum strategy from scratch, the implementation checklist should include: adjusted returns, minimum history requirements, exclusion of micro-cap names with negligible liquidity, and a clear rebalance schedule that does not create inadvertent look-ahead by using prices from after the ranking date.

Why the skip month matters more in high-turnover implementations

The magnitude of the skip-month effect is not uniform across all portfolios. It is largest for strategies that rebalance at monthly frequency and hold positions for one to three months — the holding periods where short-term reversal is still active and has not yet mean-reverted further. A strategy that holds positions for twelve months and rebalances annually will see the skip-month effect diluted because the short-term noise in the initial selection is small relative to twelve months of subsequent return.

Conversely, a monthly rebalancing strategy is buying and selling stocks at the exact moment when short-term reversal is strongest. Including month t-1 in that context means you are systematically buying names that have just experienced a noise-driven price spike and systematically selling names that have just been temporarily depressed. The skip month corrects for this in formation but does not change the rebalance timing — the two levers are related but distinct.

High-turnover indices and factor ETFs that track momentum also face implementation costs that amplify the effect. A stock at the top of a twelve-month ranking that also had a strong month t-1 is likely to attract more buyer interest at rebalance time, pushing its price up further just before you purchase it. Execution slippage concentrates in exactly those names. The skip month partially mitigates this by de-emphasizing the most recently run-up stocks.

Variants of the signal

The 12-1 specification is the most widely cited, but it is not the only useful configuration. The naming convention follows the same logic throughout: the first number is the total lookback in months, the second is the number of months skipped at the near end.

The 6-1 signal uses a five-month formation window — months t-6 to t-2 — and is more responsive to recent performance shifts. It captures turning points faster but is noisier and has higher turnover. It tends to perform better in trending markets and worse in choppy or mean-reverting regimes.

The 12-2 variant skips two months instead of one, creating additional insulation from short-term noise. Some practitioners prefer this when working with less liquid small-cap universes where microstructure effects extend beyond a single month.

Risk-adjusted or residual momentum replaces raw compounded returns with returns normalized by their volatility over the formation period, or with the residual returns from a factor model. The intuition is that a stock with a strong twelve-month return achieved through extreme volatility is a different kind of bet from one that delivered a similar cumulative return smoothly. Residual momentum also strips out market-beta and sector contributions, leaving only the idiosyncratic component of price performance — which some research suggests is more persistent.

The relationship between these variants and the broader landscape of the momentum factor is important context: they all measure different time-slices of the same underlying phenomenon but have meaningfully different risk and turnover profiles.

A worked example of the formation window

Consider a hypothetical stock evaluated at the end of June 2025. The relevant months and their role in the 12-1 calculation are as follows:

• July 2024 (t-12): Start of the formation window. The price at the end of June 2024 — the last trading day before July — is P_t-13, the denominator of the return calculation.
• July 2024 through May 2025 (t-12 to t-2): These eleven months of return are included. Each month's price change compounds into the formation-period return.
• June 2025 (t-1): Excluded entirely. The return from May 31, 2025 to June 30, 2025 does not enter the calculation. This is the skip month.
• End of June 2025 (t): Ranking date. Stocks are ranked cross-sectionally by their eleven-month compounded returns. Portfolios are constructed based on those ranks.

If this hypothetical stock had a cumulative price increase from end-June 2024 to end-May 2025 of 28 percent, that is its 12-1 momentum score, regardless of whether it fell 5 percent or rose 8 percent in June itself. A competing stock that returned 22 percent from July 2024 to May 2025 but surged 12 percent in June — perhaps on a single news event — would still rank below the first stock on the 12-1 measure, even though its twelve-month total return appears higher when June is included. That distinction in ranking is the exact correction the skip month is designed to produce.

Holding period, rebalance timing, and the full implementation picture

The formation window is only one half of the signal design. The holding period and rebalance schedule determine how much of the momentum premium the strategy can capture net of costs. Academic papers typically study one-month holding periods with monthly rebalancing, which maximizes gross exposure to the signal but also maximizes turnover. Practical implementations often extend the holding period to three or six months and stagger rebalancing across weeks to reduce market impact.

An important asymmetry exists between the formation period and the holding period: extending the formation window from six to twelve months tends to improve signal quality, but extending the holding period beyond six months begins to erode returns as the original momentum effect dissipates and, eventually, long-run reversal — a multi-year effect — begins to dominate. The 12-1 signal with a one-to-six-month holding period occupies the empirically productive middle ground.

Corporate actions during the holding period require the same diligence as during formation. A stock that is acquired at a premium, spun off a major division, or delisted midway through the holding period needs a clear handling rule. Most systematic implementations use the announced deal price for acquired names and treat the spinoff as a new security requiring a new formation history before it is eligible for ranking.

The skip month is ultimately an acknowledgment that markets are not perfectly efficient at all frequencies simultaneously. Medium-term momentum and short-term reversal are both real, both measurable, and both exploitable — but they operate on different timescales and in opposite directions. The 12-1 construction tries to hold both truths at once by keeping the window long enough to capture momentum and short enough, at the near end, to stay clear of reversal.