Does NQ fill the gap?
NQ, one-minute bars, 2008-01-02 to 2026-08-14. The gap is the 9:30 open against yesterday's 4:00 close. A quarter of the way back toward the close by 3:59 is the widely quoted "gap fill" stat; beside it, the same distance away from the close, the whole gap each way, the 5:00 close as the reference instead, and a price picked at random from yesterday's session. Gap ups and gap downs apart.
What the file says
The stat is real, and it is the first minute, not the gap. A quarter of the gap was filled on 85.5% of gap ups and 89.2% of gap downs. A quarter of the gap in the other direction, on the same days, 87.0% and 88.2%. A random price from yesterday's session was retraced a quarter of the way on 82.0% and 87.9%. The median gap is 18.00 points, so the quarter sits 4.50 points from the open, and the 9:30 bar alone has a median range of 6.50: 43.0% of the fills happened inside that one bar, 66.5% by 9:34. The whole gap, the number a reader hears in "gap fill": filled 60.7% and 64.7%, the same distance the other way 58.6% and 61.4%.
Every gap day, three references
Toward the close and away from it, a quarter of the gap and the whole gap. The white rows are the 4:00 close, the one the question is asked about.
| reference | gap | days | a quarter back | a quarter away | the whole gap back | the whole gap away |
|---|---|---|---|---|---|---|
| yesterday's 4:00 close | up | 2,675 | 2,286 · 85.5% | 2,326 · 87.0% | 1,623 · 60.7% | 1,568 · 58.6% |
| yesterday's 4:00 close | down | 2,102 | 1,874 · 89.2% | 1,853 · 88.2% | 1,361 · 64.7% | 1,291 · 61.4% |
| yesterday's 5:00 close | up | 2,652 | 2,286 · 86.2% | 2,333 · 88.0% | 1,659 · 62.6% | 1,599 · 60.3% |
| yesterday's 5:00 close | down | 2,121 | 1,906 · 89.9% | 1,879 · 88.6% | 1,395 · 65.8% | 1,322 · 62.3% |
| a random price from yesterday's session | up | 2,682 | 2,199 · 82.0% | 2,226 · 83.0% | 1,382 · 51.5% | 1,326 · 49.4% |
| a random price from yesterday's session | down | 2,097 | 1,843 · 87.9% | 1,793 · 85.5% | 1,149 · 54.8% | 1,112 · 53.0% |
Of the 4,160 days that came a quarter of the way back, 1,788 (43.0%) did it inside the 9:30 bar, 2,765 (66.5%) by 9:34 and 3,642 (87.5%) by 9:59; the median first touch was 9:31. By year, a quarter back on gap ups ran from 80.8% (2021) to 90.8% (2008), on gap downs from 82.9% (2025) to 97.9% (2020); no year of nineteen is below 80% on either side.
The same, on the last five years only
2021-09-10 to 2026-08-14: the window a shared version of this stat was cut on.
| reference | gap | days | a quarter back | a quarter away | the whole gap back | the whole gap away |
|---|---|---|---|---|---|---|
| yesterday's 4:00 close | up | 688 | 595 · 86.5% | 594 · 86.3% | 409 · 59.4% | 381 · 55.4% |
| yesterday's 4:00 close | down | 580 | 503 · 86.7% | 504 · 86.9% | 353 · 60.9% | 344 · 59.3% |
| yesterday's 5:00 close | up | 684 | 590 · 86.3% | 594 · 86.8% | 406 · 59.4% | 390 · 57.0% |
| yesterday's 5:00 close | down | 587 | 512 · 87.2% | 516 · 87.9% | 367 · 62.5% | 350 · 59.6% |
| a random price from yesterday's session | up | 710 | 574 · 80.8% | 588 · 82.8% | 366 · 51.5% | 350 · 49.3% |
| a random price from yesterday's session | down | 560 | 488 · 87.1% | 469 · 83.8% | 295 · 52.7% | 292 · 52.1% |
By year, a quarter of the gap back
| year | gap ups | a quarter back | share | gap downs | a quarter back | share |
|---|---|---|---|---|---|---|
| 2008 | 131 | 119 | 90.8% | 125 | 107 | 85.6% |
| 2009 | 138 | 119 | 86.2% | 116 | 99 | 85.3% |
| 2010 | 146 | 124 | 84.9% | 111 | 101 | 91.0% |
| 2011 | 124 | 104 | 83.9% | 129 | 111 | 86.0% |
| 2012 | 143 | 127 | 88.8% | 111 | 100 | 90.1% |
| 2013 | 157 | 130 | 82.8% | 99 | 94 | 94.9% |
| 2014 | 151 | 127 | 84.1% | 106 | 98 | 92.5% |
| 2015 | 141 | 123 | 87.2% | 114 | 107 | 93.9% |
| 2016 | 138 | 117 | 84.8% | 120 | 108 | 90.0% |
| 2017 | 155 | 133 | 85.8% | 102 | 92 | 90.2% |
| 2018 | 146 | 125 | 85.6% | 111 | 101 | 91.0% |
| 2019 | 153 | 128 | 83.7% | 105 | 90 | 85.7% |
| 2020 | 162 | 136 | 84.0% | 97 | 95 | 97.9% |
| 2021 | 151 | 122 | 80.8% | 106 | 95 | 89.6% |
| 2022 | 115 | 102 | 88.7% | 141 | 117 | 83.0% |
| 2023 | 134 | 115 | 85.8% | 123 | 110 | 89.4% |
| 2024 | 151 | 130 | 86.1% | 108 | 95 | 88.0% |
| 2025 | 152 | 131 | 86.2% | 105 | 87 | 82.9% |
| 2026 | 87 | 74 | 85.1% | 73 | 67 | 91.8% |
What it does not say, and how it was done
What this does not measure. Any trade: a quarter of the gap being touched says nothing about a fill at the open, a stop, or a cost, and on the median day the level is inside the first minute's range. Roll days are in: the file is unadjusted continuous, so four days a year carry the contract spread inside the gap; they were not removed because the shared version did not say it removed them, and the numbers reproduce with them in. Set aside and counted: 23, 27 and 21 zero gaps on the three references. The five-year window a shared version of this stat was cut on, 2021-09-10 to 2026-08-14, is the second table, where a quarter back reads 86.5% and 86.7% against the 87% and 87% that were quoted.
How it was measured. The measurement, its references and eight predictions were written down before it ran (SETUP 06, September 14, 2026; seven of the eight held). Script nq_gap_fill-2026-09-14.py v0.1 on the one-minute NQ file: 6,180,656 lines, 4,801 dates with a session, 4,800 measured, 2008-01-02 to 2026-08-14. The random draw is one close from yesterday's session a day, seed 3. The script's own print is the record for this page, nq_gap_fill.txt; every number here is read from it.
Raw market data: FirstRate Data (firstratedata.com). A measurement on a file, not advice.