WhitmanTrading

Gap Fill: The Claim That Needs a Number

A gap fill is price trading back to the level where an opening gap began, closing the untraded space. Whether gaps fill is an empirical question with a different answer for every instrument and timeframe, and the useful form of it always includes a deadline.

How it works

A candlestick series containing several opening gaps, with the largest opening gap marked. The headline on the chart reads: Price returning to where the gap started.
Price returning to where the gap started. Illustrative chart - not real market data.

A market closes at one price and opens at another. The space between the previous close and the new open never traded — the market was shut. That space is the gap.

A gap fill is price returning through that space to the previous close. The untraded band gets traded, and the chart no longer shows a discontinuity.

The folk claim is that gaps always fill. It is repeated everywhere, it is not a law of anything, and the useful version of it needs two things the folk version leaves out: a definition of “fill” and a deadline.

A candlestick chart of the site's shared price history. The headline on the chart reads: The fill rate is an instrument-specific number, not a law.
The fill rate is an instrument-specific number, not a law. Illustrative chart - not real market data.

The two things every fill statistic hides

A 72-bar window of the shared price history, with the entry price and a lower level drawn as horizontal lines. The headline on the chart reads: A partial fill is the common outcome, not the full one.
A partial fill is the common outcome, not the full one. Illustrative chart - not real market data.

First: what counts as filled. Price trading one tick into the gap is not the same as price reaching the previous close. Most quoted fill rates count any entry into the gap band, which makes the number much larger and much less useful — a gap “filled” by a fifth is a losing trade for anyone who faded it expecting the full return.

Second: filled by when. Given enough time, most prices revisit most levels. “Gaps fill eventually” is close to a statement about how markets move rather than about gaps, and it is unusable because no position can be held for an undefined period.

A long-horizon candlestick view of the same price series. The headline on the chart reads: And the question is always filled by when.
And the question is always filled by when. Illustrative chart - not real market data.

A usable version of the claim looks like this: on this instrument, over this period, what share of opening gaps larger than X% were fully filled within the same session? That produces a number you can size a position against. Everything vaguer than that is folklore.

The two kinds of gap behave differently

A strongly rising stretch of the long price series. The headline on the chart reads: A gap on real news is a repricing, not an error.
A gap on real news is a repricing, not an error. Illustrative chart - not real market data.

A gap on genuine news is a repricing. Earnings, a trial result, an acquisition — the market has new information and the old price is simply wrong now. There is no reason for price to return to a level that reflected the world before the news, and treating that gap as an error to be corrected is misreading what happened.

A flat, quiet stretch of the long price series. The headline on the chart reads: A gap on nothing is thin overnight trading.
A gap on nothing is thin overnight trading. Illustrative chart - not real market data.

A gap on no news is thin trading. A small number of participants moved price in a session with almost no volume, and the ordinary population disagrees when it arrives. That gap has a much better claim to filling, because nothing changed except who was awake.

Those two look identical on a chart and behave completely differently, which is the single most important distinction on this page and the one most fill statistics ignore by pooling them together.

A candlestick chart with a volume histogram beneath it, with the volume histogram emphasised. The headline on the chart reads: Participation at the open decides which one it was.
Participation at the open decides which one it was. Illustrative chart - not real market data.

Volume at the open is the best available separator. Heavy participation supporting the new price suggests a repricing that will hold; light participation suggests the overnight move has not been ratified by the wider market.

A flat but volatile stretch of the long price series. The headline on the chart reads: The larger the gap the less often it closes quickly.
The larger the gap the less often it closes quickly. Illustrative chart - not real market data.

Size matters and it runs the opposite way to intuition. Small gaps close often — they are usually noise. Large gaps close far less often within a useful timeframe, because a large gap almost always means something actually happened.

In practice: the trade nobody discusses honestly

A declining stretch of the long price series, with the entry price and the level at which a stop would trigger drawn as horizontal lines. The headline on the chart reads: The stop on a gap fade is the whole problem.
The stop on a gap fade is the whole problem. Illustrative chart - not real market data.

Fading a gap means betting against the direction of the most recent information, and the stop is the unsolved part. There is no structure above a gap — the level the position needs to fail past is untraded space with no prior highs, no consolidation, nothing to anchor to.

So the stop is arbitrary, which means the position size is arbitrary, which means the risk on the trade is a guess. That is a more serious objection than any fill-rate statistic.

A candlestick chart of the site's shared price history, annotated with the round-trip cost. The headline on the chart reads: And fading it costs 2% of a bar per attempt.
And fading it costs 2% of a bar per attempt. Illustrative chart - not real market data.

Each attempt costs 2% of a typical bar’s range on this site’s shared history — and gap fades are taken at the open, when spreads are widest, so the real cost is higher than the baseline.

A 72-bar candlestick section of the shared price history. The headline on the chart reads: There are no orders inside the gap to come back for.
There are no orders inside the gap to come back for. Illustrative chart - not real market data.

And there is nothing inside the gap. No resting orders, no prior activity, no memory. The order book has never seen those prices in this context. Whatever pulls price back, it is not something waiting there.

What a gap fill is not

It is not a law. It is a tendency with a different magnitude for every instrument, timeframe and gap size, and the tendency is much weaker on the gaps that matter most.

It is not a reason on its own. “There is a gap below” is not a thesis. It is an observation about the chart’s shape.

It is not the same as an imbalance. A gap happens between sessions when the market was closed. An imbalance happens during trading, when the market was open and moved too fast. Different causes, different behaviour, frequently conflated.

And it is not symmetrical. Gaps down and gaps up do not fill at the same rate on most instruments, because declines and advances are not driven by the same mechanics.

When it fails

A calmly advancing stretch of the long price series. The headline on the chart reads: An unfilled gap can stay unfilled for years.
An unfilled gap can stay unfilled for years. Illustrative chart - not real market data.

The expensive failure is waiting. A gap on real news may never fill, and a position held for the fill is capital committed indefinitely to a level with no mechanism behind it.

A sideways, range-bound candlestick series. The headline on the chart reads: And in a range every gap fills, which proves nothing.
And in a range every gap fills, which proves nothing. Illustrative chart - not real market data.

The second failure is where the belief comes from. In a range, every gap fills — price is oscillating and revisits everything. Anyone who learns gap fading during a range learns that it works almost always, and then meets a trending market with a strategy calibrated on the wrong regime.

A third is selection in the examples. Filled gaps are visible on a chart because price passed back through them. Unfilled gaps sit far above or below current price and are not in the screenshot.

A fourth is fading into the news. The gaps offering the largest apparent opportunity are the ones on the biggest news, which are exactly the ones least likely to reverse.

And a fifth is quoting a statistic you have not computed. The fill rate for your instrument, your timeframe and your definition of filled is a number you can produce in an afternoon, and it will not match the one in the article.

The original data

The corpus of 24,971 videos measured for this site covers gap trading widely and gap-fill rates almost never — the claim is repeated far more often than it is counted, which is the pattern this whole site is built to interrupt.

A candlestick chart of the site's shared price history, cut short at the decision bar. The headline on the chart reads: Gapped up three percent on earnings. Fade it?
Gapped up three percent on earnings. Fade it? Illustrative chart - not real market data.

This page deliberately does not give you a fill rate, because a number from someone else’s instrument would be worse than no number. What it gives you is the specification for computing your own: your instrument, gaps above a stated size, full fill rather than partial, closed within a stated window, split by whether news was scheduled. That takes an afternoon of data work and settles a question you would otherwise take on faith for years.

Gap trading is the wider subject, including how gaps are classified. Premarket and after-hours is where the gap is actually created. And imbalance is the intraday event this is most often confused with.

What I actually do

I traded gap fades for a year on the strength of a statistic I had read and never checked. When I finally counted it on the instrument I was actually trading, the number was nothing like the one I had been repeating, and the difference was entirely in what counted as a fill.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.