Leversens
← Back to algorithm

Momentum 12-1

Academic momentum, not technical analysis. A time-tested factor used as a timing filter on top of quality selection.

The formula
Momt = (1 + rt-252:t) / (1 + rt-21:t) − 1

12-month cumulative return, excluding the most recent month.

Why exclude the last month?

Jegadeesh and Titman (1993) showed that stocks exhibit short-term reversal over the most recent month—winners pause, losers bounce. Including that window in a momentum signal contaminates the strategy with mean-reversion noise. The "12-1" construction is the academic standard.

Why combine with quality?

Asness, Frazzini and Moskowitz (2013) demonstrated that quality and momentum are negatively correlated at the factor level (correlation ≈ −0.45). When quality underperforms, momentum often does well, and vice versa. Combining them produces a portfolio whose drawdowns are dampened.

Practically, momentum solves the "dead money" problem of pure quality strategies: a stock can be cheap and fundamentally strong yet drift sideways for years. The momentum filter avoids buying quality stocks that the market is actively rejecting.

Implementation in this strategy

  1. Compute Momt for every quality-eligible stock on every rebalance date.
  2. Drop stocks with negative momentum (clear downtrend).
  3. Keep only stocks with momentum above the cross-sectional median (relative strength).
  4. Apply quality weights to the remaining set.

Common misconception

This is not chasing the latest hot stocks. It's a slow-moving filter that excludes clearly broken names from the quality candidate set. Turnover is moderate (monthly), and the average holding period is several months.