Gap fill statistics for $SPY, $QQQ and $IWM say "most gaps fill" is only half true: small gaps usually close the same day, big ones usually don't. Across 10,129 opening gaps of 0.25% or more from January 2, 2004 to the November 18, 2026 close, a 0.25%–0.5% gap traded back to the prior close the same session 57% to 71% of the time. A 1%–2% gap did so only 29% to 40% of the time. Give it five sessions and fill rates climb to roughly 74% to 82% overall, with down gaps filling more reliably than up gaps in all three ETFs.
That is the whole story in one line: gap size, not folklore, decides the odds. The tables below show exactly how much.
What counts as a gap, and what counts as a fill
The rules here are deliberately simple, so anyone can rebuild them from a daily chart:
- Gap: the session's open versus the prior session's close. An open 0.6% above yesterday's close is a 0.6% up gap.
- Fill: price trades back to the prior close. For an up gap, that means the low reaches or breaks the prior close. For a down gap, the high reaches or clears it.
- Same-day fill: it happens during the gap session itself.
- Five-session fill: it happens on the gap day or any of the next four sessions.
- Buckets: 0.25%–0.5%, 0.5%–1%, 1%–2% and over 2%, each split into up and down gaps. Gaps under 0.25% are ignored as noise.
StockCharts ChartSchool separates a full gap from a partial gap: a full gap up opens above the prior day's high, while a partial gap up opens above the prior close but not above the prior high. The definition used here counts both, and the split turns out to matter (more on that below). ChartSchool's own gap analysis primer also warns that the adage that all gaps eventually get filled may not hold, especially for breakaway and runaway gaps.
Why the test starts in 2004
Open prices are the weak link in free historical data. In the early years, opens often look stale. For $IWM, the open equaled the day's high or low on 12% to 37% of sessions each year from 2000 to 2003, then 0% to 6% in every year from 2004 on. $SPY shows the same artifact at 22% to 40% in 1993–1995. So the test uses a common window for all three ETFs: January 2, 2004 through the November 18, 2026 close, 5,757 sessions each.
Data are daily bars from Yahoo Finance historical data (open, high, low and close, split-adjusted, price-only). Ex-dividend sessions were dropped as gap events, because the open falls mechanically by the payout; that removed 92 $SPY, 89 $QQQ and 91 $IWM sessions and moved no headline rate by more than 1.1 points. As a cross-check, the same rules run on Nasdaq.com data for the last 10 years matched the Yahoo results within 0.3 points on every headline rate.
The gap fill table: same session versus five sessions
Each cell shows the share of gaps that filled the same day, then the share that filled within five sessions. The last column counts events for $SPY / $QQQ / $IWM.
| Gap | $SPY same / 5-day | $QQQ same / 5-day | $IWM same / 5-day | Events |
|---|---|---|---|---|
| Up 0.25%–0.5% | 57.4% / 81.3% | 64.2% / 82.9% | 71.1% / 89.0% | 894 / 876 / 845 |
| Up 0.5%–1% | 41.5% / 69.3% | 46.5% / 71.6% | 50.1% / 77.4% | 610 / 746 / 757 |
| Up 1%–2% | 29.6% / 55.6% | 29.0% / 56.0% | 34.6% / 61.8% | 196 / 300 / 335 |
| Up over 2% | 35.3% / 62.7% | 22.4% / 50.7% | 32.1% / 57.7% | 51 / 67 / 78 |
| Down 0.25%–0.5% | 65.4% / 89.8% | 68.7% / 86.8% | 70.2% / 89.1% | 607 / 613 / 607 |
| Down 0.5%–1% | 43.4% / 76.8% | 50.5% / 82.1% | 51.8% / 82.9% | 440 / 543 / 550 |
| Down 1%–2% | 39.9% / 73.7% | 36.7% / 69.5% | 40.2% / 72.0% | 213 / 283 / 296 |
| Down over 2% | 38.7% / 66.1% | 22.2% / 58.3% | 27.3% / 61.4% | 62 / 72 / 88 |
| All up gaps | 48.1% / 73.7% | 50.8% / 73.5% | 55.6% / 78.9% | 1,751 / 1,989 / 2,015 |
| All down gaps | 52.7% / 81.8% | 53.9% / 80.5% | 55.4% / 82.0% | 1,322 / 1,511 / 1,541 |
Three patterns jump out.
- Size rules everything. Every ETF's same-day fill rate falls from the 0.25%–0.5% bucket to the 1%–2% bucket, on both sides. A small gap sits well inside a normal day's range: the median high-low range was 0.95% for $SPY, 1.27% for $QQQ and 1.47% for $IWM. Price only has to wander a little to touch the prior close.
- Down gaps fill more than up gaps over five sessions. That held in all three ETFs: 81.8% versus 73.7% for $SPY, 80.5% versus 73.5% for $QQQ, 82.0% versus 78.9% for $IWM. One plausible reason: in a market that drifted higher for most of two decades, climbing back over a down-gap level has been easier than falling back through an up-gap level.
- $IWM fills the most. Its wider daily range means even mid-sized gaps get revisited. In the 0.25%–0.5% up bucket, $IWM filled 71.1% the same day versus 57.4% for $SPY.
One quirk: $SPY's over-2% up gaps filled more often than its 1%–2% up gaps. With only 51 events, that bucket is too thin to read much into.
Full gaps hold, partial gaps fade
Splitting the same 2004–2026 events by ChartSchool's full-versus-partial definition sharpens the picture. Full gaps, which open beyond the prior day's high or low, filled the same day 42.1% (up) and 46.2% (down) of the time in $SPY. Partial gaps, which open inside the prior day's range, filled 55.8% and 60.3% of the time. $QQQ (42.8% versus 59.5% for up gaps) and $IWM (48.7% versus 61.8%) show the same gap of roughly 13 to 17 points.
Gaps of 1% or more that did not fill the same day also tended to keep going. In $SPY, 73.7% of those unfilled up gaps closed above the open, and 62.7% of unfilled down gaps closed below it. $QQQ ran 79.6% and 64.7%; $IWM 73.9% and 66.4%. Part of that is built into the selection, since a day that never touched the prior close rarely collapses back through its own open, but it is a fair warning against fading a big gap that is holding.
Is gap fill getting harder?
Slightly. Over the last 10 years alone (November 21, 2016 to the November 18, 2026 close), same-day fill rates for all gaps of 0.25% or more were 43.2% up and 47.9% down for $SPY, 47.4% and 48.2% for $QQQ, and 52.1% and 51.4% for $IWM. That is 3 to 6 points lower than the full 2004–2026 figures. Five-session rates slipped less, by 0.5 to 3.3 points, and still ranged from 70.4% to 80.7%. A fill rate is also not an edge by itself; a fade that wins often but loses big when the gap runs needs the payoff math covered in the risk-reward ratio and win rate breakdown.
The November 18, 2026 open: three gaps, three fills
Here is how the rules score the latest completed session, using the November 17 close as the reference:
| ETF | Nov 17 close | Nov 18 open | Gap | Nov 18 range | Filled same day? |
|---|---|---|---|---|---|
| $SPY | $732.44 | $733.37 | +0.13% | $726.80–$735.06 | Yes (below 0.25% threshold) |
| $QQQ | $656.95 | $661.03 | +0.62% | $653.05–$662.80 | Yes |
| $IWM | $269.05 | $268.15 | -0.33% | $267.89–$272.36 | Yes |
Worked example for $QQQ: $661.03 divided by $656.95 is 1.0062, a 0.62% up gap. The session low of $653.05 was below $656.95, so the gap filled. Historically, $QQQ up gaps of 0.5%–1% filled the same day 46.5% of the time, so this one landed on the slightly less common side. $IWM's 0.33% down gap sat in the bucket that filled 70.2% of the time, and it did, with the high reaching $272.36. $SPY opened just 0.13% higher, too small to count as a gap here, yet still traded down through the prior close. $SPY, $QQQ and $IWM closed at $730.20, $659.51 and $270.12.
Traders who use gap levels for exits often pair them with a volatility stop such as an ATR trailing stop, since a big gap that does not fill has more often kept running than reversed, as the numbers above show. More rules-based studies live on the Analysis page and the Traders On Trend homepage.
Limits, in one breath: these are price-only daily bars from one data source, with no costs, slippage or taxes, and a daily bar cannot show whether the fill came in the first five minutes or the last.
Bull/Bear Verdict
Bull Case: Where gap fill statistics help: they put a real number on small gaps. A 0.25%–0.5% gap in $SPY, $QQQ or $IWM touched the prior close the same day 57% to 71% of the time, and within five sessions 81% to 90% of the time. For traders sizing a fade or choosing a target, the prior close has been a frequently revisited level at that scale.
Bear Case: Where they mislead: "gaps always fill" falls apart once gaps get big. Up gaps of 1%–2% filled the same day only 29% to 35% of the time, full gaps filled less often than partial gaps, and unfilled big gaps mostly closed further in the gap's direction. Recent years show lower same-day fill rates, too.
FAQ
What percentage of gaps fill the same day?
From 2004 to the November 18, 2026 close, gaps of 0.25% or more filled the same session 48.1% (up) and 52.7% (down) of the time in $SPY, 50.8% and 53.9% in $QQQ, and 55.6% and 55.4% in $IWM. The rate depends heavily on size: small gaps fill far more often than large ones.
Do down gaps fill more often than up gaps?
Over five sessions, yes, in all three ETFs. Down gaps filled within five sessions 80.5% to 82.0% of the time, versus 73.5% to 78.9% for up gaps. Same-day rates were closer, within 5 points of each other.
Is a full gap less likely to fill than a partial gap?
Historically, yes. Gaps that opened beyond the prior day's high or low filled the same day 42% to 49% of the time across the three ETFs, versus 56% to 62% for gaps that opened inside the prior day's range.
Did the November 18, 2026 gaps fill?
Yes. $QQQ opened 0.62% higher and $IWM 0.33% lower, and both traded back to their November 17 closes during the session. $SPY's 0.13% opening move was below the 0.25% threshold but also traded through the prior close.