Momentum: The 12-minus-1 Signal, the 2009 Crash and the Cost of Turnover

What you will learn

• How the standard cross-sectional momentum signal is defined and why the most recent month is skipped.
• Why momentum portfolios suffered an extreme loss in spring 2009 and what a "momentum crash" is.
• How turnover turns a strong paper result into an expensive strategy to hold.
• How a long-only momentum fund behaves in a crash.

Explanation

Cross-sectional momentum is the tendency of shares that outperformed others over the past six to twelve months to keep outperforming for the next few months, and of past losers to keep lagging. It was documented for US shares by Jegadeesh and Titman in 1993 and has since been found in most equity markets. It differs from the intraday momentum setups in the day-trading course and from indicators such as RSI: it ranks stocks against each other rather than judging one chart.

The standard signal is called 12-minus-1 (or 12-1, or "12_2"). At the end of each month you measure each share's total return from twelve months ago to one month ago, that is, over months t−12 to t−2. The most recent month (t−1) is skipped because returns over one month tend to reverse slightly, an effect called short-term reversal; including it weakens the signal. Shares are then ranked, typically into deciles, and the academic factor (UMD, up minus down, or WML, winners minus losers) buys the top decile and shorts the bottom decile, holding for one month before re-ranking. Many implementations hold longer with overlapping portfolios to reduce trading.

Momentum's average return is strong on paper, but its distribution has a dangerous left tail. Kent Daniel and Tobias Moskowitz ("Momentum Crashes", 2016) showed that the worst losses come in panic states: after a long market decline, when volatility is high and the market suddenly rebounds. The reason is mechanical. After a bear market the "loser" decile is full of beaten-down, high-beta companies, so a long-short momentum portfolio is short the stocks that bounce hardest. In the three months from March to May 2009, the past-loser decile of US shares rose by about 163 percent while the past-winner decile gained only about 8 percent. A one-for-one long-short portfolio lost most of its capital. A similar episode occurred in July and August 1932 (losers about +232 percent, winners about +32 percent).

The second structural problem is turnover. A value ranking changes slowly because book values and earnings change slowly. A twelve-month return ranking changes every month, so a sizeable share of the portfolio is replaced at every rebalance. Each replacement costs the bid-ask spread, commissions and market impact, and on the short side a stock-borrowing fee. Research results are reported before these costs, which is why momentum shows the largest gap between paper and realised returns of the major factors. Lesson 10 covers capacity in detail; here we estimate the order of magnitude.

Long-only momentum products avoid the short leg and so the worst of the crash mechanics, but after a bear market they hold defensive, low-beta previous winners and will lag a sharp rebound badly, even without an absolute loss.

Worked example

Hypothetical example: a share closed at $50 thirteen months ago, at $62 one month ago and at $65 today. No dividends were paid; fees, spreads and taxes are excluded.

• 12-minus-1 signal: $62 ÷ $50 − 1 = +24.0%. This is the number used for ranking.
• Plain twelve-month return: $65 ÷ $50 − 1 = +30.0%. The difference is the skipped month, $65 ÷ $62 − 1 = +4.8%, which is deliberately excluded.

Now a cost estimate for a hypothetical long-short decile strategy with 150% one-way turnover per year on each leg (a realistic order of magnitude for monthly re-ranking), an all-in cost of 0.25% per unit traded and a borrowing fee of 0.5% per year on the short leg:

• Trading cost: 2 legs × 150% × 0.25% = 0.75% per year.
• Borrowing fee: 0.5% per year on the short leg.
• Total implementation drag: about 1.25 percentage points per year before the manager's fee.

If the paper spread were 6 points, about one fifth would disappear in costs under these assumptions; in thin small-cap liquidity, impact costs can be several times higher.

Common mistake

Practice activity

Use the hypothetical March–May 2009 figures from the lesson: past losers +163%, past winners +8%. (a) Calculate the gross return of a one-for-one long-short portfolio that is long winners and short losers, ignoring costs and margin calls. (b) Calculate the return of a long-only fund fully invested in past winners. (c) If the broad market rose 35% over the same period, state by how much the long-only fund lagged the market.

Key takeaways

• The 12-minus-1 signal ranks shares by return from twelve months ago to one month ago, skipping the last month.
• Momentum crashes happen in sharp rebounds after bear markets, when the short side is full of high-beta losers; March–May 2009 is the modern example.
• Monthly re-ranking creates high turnover; trading costs and borrowing fees consume a meaningful part of the paper spread.
• Long-only momentum funds avoid the short-side disaster but can lag a rebound heavily.

Next step

The next lesson covers quality and profitability, including why quality tends to pair well with value in one portfolio.