The Momentum Pullback screen: momentum leaders back near their 20-day average

Momentum Pullback is the seventh screen on QM Screener. It takes the Momentum Continuation list and keeps the names that made a strong run in the last few weeks and have since come back to their short moving averages. This page covers what it tests, how it relates to the screen it is cut from, what the backtest behind it showed and where it shows up in the product.

What it is and why it exists

Momentum Continuation lists liquid US stocks in the top of the market's three-month relative-strength ranking that are holding above their 50-day moving average. On a strong day that list runs to a few hundred names, and they sit in very different places on their charts: some are far above their averages after a fast run, some have gone sideways for weeks, and some ran recently and have pulled back to their 20-day line.

Momentum Pullback keeps that last group. It is a pullback cut of Momentum Continuation, chosen by backtest rather than taken from a published method. We tested a set of narrower pools drawn from the Momentum Continuation list over twelve years of daily data, and this one had the highest volatility-matched lift of the pools that were not parabolic. It was also the one whose lift held up when SPY was below its own 50-day average. The figures are in the backtest section below.

The rule in plain words

A name has to pass every Momentum Continuation rule first, unchanged:

a three-month relative-strength percentile strictly above 85,
an average daily range (ADR%, the mean of each session's high over its low over the last 20 sessions, minus one, as a percent) strictly above 2.50%,
average dollar volume strictly above $20 million over 50 days and $10 million over 20 days,
a close above its 50-day simple moving average, and
an industry outside pharmaceuticals and biotechnology, which that screen leaves out by default.

Then it has to pass two more tests, both measured at the close:

It ran. At some point in the last 20 sessions, today included, it closed at least 15% above its 50-day simple moving average.
It came back. Today's close is within one ADR% of its 20-day exponential moving average (EMA20), above or below, or it sits inside the band the 10-, 20- and 50-day EMAs make. The ADR% used here is the stock's own figure on the row. It is not the version capped at 12% that the consolidation score uses.

Both inputs are fields on every row, so any verdict can be checked against the numbers beside it:

vs EMA20 (dist_ema20_pct) is the signed distance from the close to the EMA20, in percent.
20d max ext (ext_max_20d_pct) is the highest close-over-50-day extension of the last 20 sessions, each day against its own 50-day average. It needs 69 bars of history, and a name without them fails rather than being guessed at.

How it relates to Momentum Continuation and the setup tags

Every Momentum Pullback name is also a Momentum Continuation name. On October 2, 2026, Momentum Continuation passed 227 names and Momentum Pullback passed 81 of them. The list opens on the same sort, relative strength from the top, and carries the same controls: relative strength, consolidation score and setup tag.

Of those 81, 77 were tagged BASE or TIGHT. That overlap is expected. A close sitting on its 20-day EMA or inside the EMA band is the shape the consolidation score rewards, and BASE and TIGHT are drawn from that score. The setup tags look only at where the stock is now. The run-up requirement looks back over the last 20 sessions, and it is what separates this screen from a plain BASE and TIGHT filter on Momentum Continuation: a name that has drifted sideways for two months can carry a BASE tag and still fail here, because it never closed 15% above its 50-day average in that window.

What the backtest showed

The study ran from January 2, 2014 to October 7, 2026 over 3,380 tickers: the stocks listed today with a current market capitalization of at least $300 million. It used end-of-day prices and no trading costs. It measures the quality of the candidate pool. It does not measure returns anyone could have traded. Each day's Momentum Pullback names were compared with the full Momentum Continuation list that day and with the liquid universe as a whole.

Within 20 sessions of the signal close	Momentum Pullback	Momentum Continuation	Liquid universe
Reached +20%	21.8%	17.8%	8.8%
Fell 20%	18.3%	14.6%	6.9%

Two more figures carry most of the weight. Matched for volatility, so that a pool of faster-moving stocks gets no credit just for moving more, the Momentum Pullback pool's lift was 1.06 (95% confidence interval 1.02 to 1.10), and it was the only pool that was not parabolic whose interval cleared 1 on every upside measure. With SPY below its 50-day average, its pooled lift was 1.90, against 1.29 for the full Momentum Continuation pool. The pool held about 32 names a day on average over the whole period and 72 a day over 2025 and 2026.

Read those figures with the caveats that come with them:

Survivorship. The study used today's survivors only, which inflates every absolute rate; comparing one pool with another, as these figures do, partly cancels that.
In-sample. The rule was chosen on the same data it is reported on.
End-of-day, no costs. Signals are taken at the close, and nothing is deducted for slippage or commissions.
No strategy was tested. No entries, exits or position sizes were tested. The study counts how often names in the pool moved by a set amount within a set window.
Wider outcomes in both directions. The pool fell 20% more often too, as the table shows.

Where to find it

The screen picker. Open the dashboard and pick Momentum Pullback. It is the last entry, after the six screens that each follow a published method.
Near misses. The Near misses tab lists the Momentum Continuation names that missed only the pullback test, strongest first. On this screen the near-miss list keeps its relative-strength control, because those names span the whole strength range above the gate.
The details panel. Beside the chart, the panel shows vs EMA20 and 20d max ext for the charted name and says how many screens it passes. It counts in families, so a name on both Momentum Pullback and Momentum Continuation counts once and the heading still reads "of 6".
Market breadth. The Breadth card counts Momentum Pullback names per session like any other screen. Its history was replayed from saved prices back to July 2025.
The daily digest. The email lists the screen with the others. Its lead ranks names by how many screen families they pass first, so being on both Momentum Pullback and Momentum Continuation is not a second family vote, though that name's line in the email still lists both screens.

A free account sees the top 5 names of each screen, this one included, each ranked by that screen's own sort (relative strength on this one). The full list, the near misses and the exports are $10 a month or $84 a year, each starting with a 7-day free trial. Start there, or sign in if you already have an account.

For assistants and scripts

The screen's key in the HTTP API and the MCP server is pullback: the MCP screen_results tool takes it as screen, and the API serves the same document as pullback_results.json. Each row carries dist_ema20_pct and ext_max_20d_pct beside the usual fields. The docs cover authentication and the error format.

QM Screener is an independent product and is not affiliated with, endorsed by, or sponsored by any trader, author, publisher or firm.

Frequently asked questions

Is Momentum Pullback a published trading method?

No. The other screens here each follow a publicly documented method. This one is a cut of the Momentum Continuation list chosen by backtest: of the pools we tested, it had the highest volatility-matched lift among the ones that were not parabolic, and it was the one that held up best with SPY below its own 50-day average. That makes it a choice made on past data, with the limits any such choice has.

Why does a name here also count as a Momentum Continuation name?

Because it is one. Every name on this list has passed every Momentum Continuation rule first, unchanged, and then one more test, so the two are one screen family. The details panel counts screen families, so its heading still goes up to 6 and a name on both counts once there. The daily digest's lead ranks by families first, so passing both is not a second family vote, though the name's line in the email still lists both screens.

Is a name on this list lower risk than the rest?

Nothing in the backtest says so. Over 20 sessions the Momentum Pullback pool fell 20% more often than the full Momentum Continuation pool (18.3% against 14.6%) as well as rising 20% more often. It is a pool with wider outcomes in both directions.

Why are its near misses different from Momentum Continuation's?

On Momentum Continuation a near miss passed every rule except relative strength. The other five screens have no near-miss list. On Momentum Pullback a near miss passed all of Momentum Continuation, relative strength included, and missed only the pullback test. Its near misses therefore span the whole strength range above the gate, which is why the near-miss list on this screen keeps its relative-strength control.