Drawdown Quality as a Momentum Filter – A Screening Method

Two stocks are up 30% over the past year. One pulled back 8% to its 50-day moving average, held support for three sessions, and resumed its trend in a clean V-shape. The other gained that 30% through a sequence of 15-20% drops, messy recoveries, and whipsaws that shook out every trailing stop along the way. Standard momentum screens rank them identically. Your portfolio will not experience them identically.

Drawdown quality as a momentum filter addresses this gap. Instead of ranking stocks only by trailing return, you examine how they pulled back and recovered. Research from Choi (2021) found that stocks sorted by the structure of their maximum drawdown, not just the magnitude of their return, produced stronger risk-adjusted performance than classic momentum. I use a version of this filter in my own screening, and it catches problems that raw 12-month return rankings miss entirely.

What Drawdown Quality Means

Drawdown quality measures the character of a stock’s worst pullback during a lookback period. Two dimensions matter: the shape of the decline and the speed of the recovery.

A high-quality drawdown is a controlled pullback. The stock drops to a recognizable support level, holds it without breaking lower on expanding volume, and recovers within a few sessions. The decline is orderly. You can draw a trendline on it. Support holds where you expect it to hold.

A low-quality drawdown is a grinding, choppy mess. The stock drops, bounces, drops again past the prior low, bounces less, drops again. There is no clean support. The decline takes weeks, sometimes months, and the recovery is weak and stuttering. Even if the stock eventually rallies, the drawdown period showed disorganized selling with no clear institutional accumulation.

The simplest way to think about it: high-quality drawdowns look like a stock pausing within a trend. Low-quality drawdowns look like a stock losing its trend and trying to rebuild one from scratch.

Why Raw-Return Momentum Fails

Classic momentum sorts stocks by 12-month return (sometimes excluding the most recent month). The logic is sound at the portfolio level. Stocks that went up tend to keep going up. But the strategy has a well-documented problem: momentum crashes.

When the market reverses hard, the stocks with the highest trailing returns tend to fall the most. This is not random bad luck. Many high-return stocks got there through volatile, leveraged-feeling price action. They are high-beta names that rode the trend harder in both directions.

I ran into this myself screening for momentum candidates in late March 2026. A raw 6-month return screen surfaced names that were technically “up” but had spent weeks grinding through lower highs and lower lows. Their returns were positive only because of a strong move months earlier. The recent price structure was deteriorating.

Drawdown quality catches this. A stock that is up 25% but experienced a choppy 18% drawdown with three failed recovery attempts is a fundamentally different animal than a stock up 25% with a clean 7% pullback. The first one is fragile. The second one is trending.

The common mistake is assuming that a stock with a large drawdown is automatically disqualified. It is not about the size of the drawdown alone. It is about how the stock behaved during the drawdown. A sharp, fast 10% drop to support that recovered in four days shows buyers stepping in. A slow 10% decline over six weeks with no clear floor shows sellers in control and buyers waiting.

Clean Pullbacks vs. Choppy Ones

Here is what separates a clean pullback from a choppy one, with real recent examples.

AAPL in early April 2026 offers a clean drawdown structure. The stock hit $262.16 on April 6, dropped to $245.70 on April 7 (about 6.3% intraday), then recovered to $260.49 by April 9. Three trading days from low to recovery. The decline was sharp but brief. The low was a single spike, not a grinding series of lower lows. Volume expanded on the recovery day ($258.90 close on April 8, $260.49 on April 9). This is what institutional re-accumulation looks like.

Contrast that with MSFT over the same period. From its March 6 high of $413.05, MSFT declined through a sequence of lower highs and lower lows: $405 on March 10, $395 on March 13, $381 on March 20, $372 on March 24, $365 on March 26, and finally $356.77 on March 27. That is three weeks of grinding decline. The bounces along the way (March 16 to $399.95, March 31 to $370.17) were weaker each time. Each failed recovery created a lower high. By the time MSFT bottomed, it was down 13.6% and the market structure had shifted from uptrend to downtrend.

Both stocks recovered eventually. But the quality of their drawdowns was completely different. AAPL’s pullback was a pause. MSFT’s was a breakdown.

What to look for in a clean pullback:

• Single leg down, not multiple legs with bounces in between.
• Decline stops at a recognizable level (moving average, prior consolidation, Fibonacci retracement).
• Volume dries up near the low, then picks up on the reversal.
• The recovery candle is decisive, not a slow drift upward.

What signals a low-quality drawdown:

• Multiple lower lows separated by weak bounces.
• No clear support level. The stock just slides.
• Volume stays elevated through the decline (distribution, not shakeout).
• Recovery attempts fail at lower and lower levels.

Recovery Speed as a Signal

Recovery speed is the second half of drawdown quality, and I find it even more telling than the pullback structure itself.

The logic is straightforward. When a stock drops and snaps back fast, it means buyers were waiting. Demand was latent. The pullback cleared weak hands, and the next wave of buying came in quickly. When a stock drops and spends weeks crawling back, it means the buying pressure that created the trend has weakened. The stock may still recover, but the character of the move has changed.

You can measure recovery speed simply: count the trading days from the drawdown low to the point where the stock reclaims 75% or 100% of the drawdown.

Higher values mean faster recovery relative to the size of the drop. A stock that fell 8% and reclaimed 75% of it in 3 days scores 2.67. A stock that fell 8% and took 15 days to reclaim 75% scores 0.53. The first stock is showing you that the trend is intact. The second is showing you indecision at best.

Look at META through late March 2026. From its $672.19 high on March 4, it declined to $520.26 on March 27. That is a 22.6% drawdown. It bounced to $592.55 on April 1, reclaiming about 47% of the loss in 3 trading days. Then it stalled, pulling back to $574.46 on April 2. The initial recovery was fast but incomplete, and it immediately lost momentum. That partial, stalling recovery is a warning sign. Compare it to AAPL’s 6.3% drawdown where 95% of the loss was recovered in 3 days. The difference in recovery quality is more important than the difference in drawdown size.

Teams that screen only on drawdown magnitude (e.g., “exclude anything with a max drawdown over 15%”) miss this signal. A 15% drawdown recovered in 4 days is bullish. A 10% drawdown still not recovered after 20 days is not.

How Drawdown Quality Filters Work in Practice

Choi’s 2021 research sorted stocks into portfolios based on the quality of their maximum drawdown, measured by the depth-to-recovery ratio and the smoothness of the decline path. The finding: high-drawdown-quality stocks (clean pullbacks, fast recoveries) outperformed low-quality ones, even when both groups had similar trailing returns. The effect persisted after controlling for standard momentum, volatility, and size factors.

This is not a single indicator you add to a chart. It is a filter that sits on top of your existing momentum screen. Here is how I structure it:

Step 1: Start with a standard momentum universe. Rank stocks by 12-month return (excluding the most recent month) or 6-month return. Take the top quintile or whatever your usual cutoff is.

Step 2: For each stock in that universe, calculate the maximum drawdown over the lookback period. Note both the depth and the duration.

Step 3: Measure recovery speed. From the max drawdown trough, count how many trading days it took to reclaim 75% of the loss. If the stock has not reclaimed 75%, flag it.

Step 4: Score the pullback structure. This is the qualitative part. Count the number of distinct lower lows during the drawdown. A single-leg pullback (one move down, one move up) scores better than a multi-leg decline. You can approximate this programmatically by counting swing lows using the momentum indicator or a zigzag filter.

Step 5: Combine. Rank the momentum universe by a composite of recovery speed and pullback structure. Discard the bottom third. What remains is your drawdown-quality-filtered momentum list.

The common mistake at this stage is overcomplicating the scoring. I tried a weighted composite of six sub-factors once and it overfitted badly. Two factors (recovery speed and number of swing lows during the drawdown) capture most of the signal. Adding a third for volume behavior during recovery can help, but beyond that, you are likely fitting noise.

What This Filter Does Not Tell You

Drawdown quality is a filter, not a timing signal. It tells you which momentum stocks are higher quality. It does not tell you when to buy them.

A stock can have excellent drawdown quality and still be extended. If AAPL shows a clean pullback structure but is sitting 15% above its nearest support, the quality of the prior drawdown does not protect you from the next one. You still need an entry framework. The Ulcer Index measures cumulative drawdown pain over time and can complement this filter by helping you assess whether a stock’s overall drawdown behavior, not just the worst episode, matches what your risk tolerance allows.

It also does not work well in market regime shifts. When the entire market breaks down, even high-quality momentum names get sold. The filter improves stock selection within normal conditions. In crashes, everything correlates toward one. Choi’s research showed the outperformance was concentrated in normal and recovering markets, not during acute stress.

And this filter has a lookback dependency. A stock that had a messy drawdown four months ago may have completely changed character since. I re-run the filter monthly and use a rolling 6-month window. Using 12 months introduces too much stale information.

Building the Screen Step by Step

If you want to implement this in a screener or spreadsheet, here is the minimum viable version.

Pull daily closing prices for your momentum universe over the past 6 months. For each stock, find the maximum peak-to-trough drawdown. Record the peak date, trough date, and depth.

Count the number of swing lows during the drawdown period. A swing low is a day where the low is lower than both the prior day’s low and the next day’s low. One swing low means a clean V. Three or more means choppy, grinding action.

Calculate recovery speed as described earlier. If the stock has not recovered 75% of the drawdown by the screen date, assign it a penalty score or exclude it.

Rank by: (1) fewer swing lows is better, (2) faster recovery is better. Equal weight both. Take the top half of your momentum universe after this filter.

The result is a momentum list that skews toward stocks in genuine trends rather than stocks with volatile, directionless returns. In my experience, this filter removes about a third of the names from a standard momentum screen. The ones it removes are usually the ones that cause the most pain when momentum reverses.

When Drawdown Quality Matters Most

This filter adds the most value in two market environments.

First, during late-cycle momentum. When a bull market is aging, momentum screens fill up with names that have risen sharply but on deteriorating internals. Their drawdowns get sloppier. Recovery times lengthen. Raw returns still look good, but the underlying structure is weakening. Drawdown quality catches the rot before the return numbers do.

Second, after corrections. When the market bounces off a selloff, some stocks recover cleanly and others limp. Running a drawdown quality filter on a post-correction momentum screen helps you identify which stocks have genuine buying support versus which ones are just drifting up with the tape.

Where it matters less: in strong, broad rallies where almost everything trends cleanly. In that environment, most momentum stocks have good drawdown quality by default, and the filter does not differentiate much. Use it when you need to separate signal from noise, not when the signal is already clean.

Educational content only. Not investment advice. Trading involves risk. You are responsible for your decisions.