Does the Sunday gap fill on Monday?
ES, one-minute bars, every Monday from 2008-01-02 to 2026-08-14. The gap is Monday's 9:30 open against Friday's 4:00 PM close: the part of the weekend gap still open at the bell. A fill is Monday trading back to that close by 3:59 PM; the mirror is price going the same distance the other way; the base rate is a random price from Friday's session, one draw a Monday. Up gaps and down gaps are counted apart.
What the file says
A distance, not a magnet. Monday came all the way back to Friday's close on 490 of 923 Mondays, 53.1%, and went the same distance in the other direction on 497, 53.8%. Down gaps filled 57.0% of the time, up gaps 49.7%. A random Friday price was touched less often, 44.2%, but a random price sits farther from Monday's open on average than Friday's close does, so the fair comparison is the same distance the other way.
Every Monday, three references
Counts and shares, from the record. The 4:00 close is the one the question is asked about.
| reference | gap | Mondays | retrace 25% | extension 25% | full fill | full extension |
|---|---|---|---|---|---|---|
| Friday's 4:00 PM close | up | 497 | 386 · 77.7% | 426 · 85.7% | 247 · 49.7% | 252 · 50.7% |
| Friday's 4:00 PM close | down | 426 | 368 · 86.4% | 361 · 84.7% | 243 · 57.0% | 245 · 57.5% |
| Friday's 4:00 PM close | both | 923 | 754 · 81.7% | 787 · 85.3% | 490 · 53.1% | 497 · 53.8% |
| Friday's 5:00 PM close | up | 516 | 410 · 79.5% | 445 · 86.2% | 268 · 51.9% | 268 · 51.9% |
| Friday's 5:00 PM close | down | 409 | 356 · 87.0% | 343 · 83.9% | 224 · 54.8% | 237 · 57.9% |
| Friday's 5:00 PM close | both | 925 | 766 · 82.8% | 788 · 85.2% | 492 · 53.2% | 505 · 54.6% |
| a random price from Friday's session | up | 535 | 406 · 75.9% | 438 · 81.9% | 231 · 43.2% | 232 · 43.4% |
| a random price from Friday's session | down | 393 | 317 · 80.7% | 313 · 79.6% | 179 · 45.5% | 187 · 47.6% |
| a random price from Friday's session | both | 928 | 723 · 77.9% | 751 · 80.9% | 410 · 44.2% | 419 · 45.2% |
Of the 754 Mondays that came a quarter of the way back, 258 (34.2%) did it inside the 9:30 bar, 411 (54.5%) by the 9:34 bar and 610 (80.9%) by 9:59; the median first touch was 9:33. The median gap was 7.25 points against a median 9:30-bar range of 2.00 points, 3.6 times the bar, so a quarter-fill (1.81 points at the median) is about the noise of one bar and the full fill is not.
By year, on Friday's 4:00 close
Up gaps and down gaps apart.
| year | up gaps | retraced 25% | filled | down gaps | retraced 25% | filled |
|---|---|---|---|---|---|---|
| 2008 | 25 | 23 · 92.0% | 18 · 72.0% | 26 | 20 · 76.9% | 16 · 61.5% |
| 2009 | 25 | 16 · 64.0% | 11 · 44.0% | 24 | 16 · 66.7% | 12 · 50.0% |
| 2010 | 34 | 30 · 88.2% | 21 · 61.8% | 14 | 13 · 92.9% | 10 · 71.4% |
| 2011 | 23 | 16 · 69.6% | 12 · 52.2% | 26 | 18 · 69.2% | 8 · 30.8% |
| 2012 | 20 | 16 · 80.0% | 9 · 45.0% | 30 | 29 · 96.7% | 18 · 60.0% |
| 2013 | 29 | 24 · 82.8% | 17 · 58.6% | 21 | 20 · 95.2% | 10 · 47.6% |
| 2014 | 24 | 17 · 70.8% | 11 · 45.8% | 27 | 24 · 88.9% | 18 · 66.7% |
| 2015 | 22 | 15 · 68.2% | 9 · 40.9% | 27 | 23 · 85.2% | 15 · 55.6% |
| 2016 | 27 | 21 · 77.8% | 11 · 40.7% | 21 | 18 · 85.7% | 12 · 57.1% |
| 2017 | 27 | 19 · 70.4% | 13 · 48.1% | 21 | 19 · 90.5% | 12 · 57.1% |
| 2018 | 29 | 23 · 79.3% | 12 · 41.4% | 21 | 18 · 85.7% | 12 · 57.1% |
| 2019 | 29 | 22 · 75.9% | 15 · 51.7% | 21 | 16 · 76.2% | 8 · 38.1% |
| 2020 | 28 | 18 · 64.3% | 8 · 28.6% | 22 | 20 · 90.9% | 11 · 50.0% |
| 2021 | 28 | 22 · 78.6% | 14 · 50.0% | 21 | 18 · 85.7% | 11 · 52.4% |
| 2022 | 21 | 17 · 81.0% | 13 · 61.9% | 29 | 24 · 82.8% | 16 · 55.2% |
| 2023 | 23 | 20 · 87.0% | 12 · 52.2% | 25 | 23 · 92.0% | 20 · 80.0% |
| 2024 | 34 | 31 · 91.2% | 22 · 64.7% | 17 | 17 · 100.0% | 10 · 58.8% |
| 2025 | 33 | 25 · 75.8% | 12 · 36.4% | 18 | 17 · 94.4% | 12 · 66.7% |
| 2026 | 16 | 11 · 68.8% | 7 · 43.8% | 15 | 15 · 100.0% | 12 · 80.0% |
By year, on Friday's 4:00 close, the full fill on up gaps ran from 28.6% (2020) to 72.0% (2008) and on down gaps from 30.8% (2011) to 80.0% (2023 and 2026). No year is out of character with the whole.
What it does not say, and how it was done
What this does not measure. The part of the weekend gap that closed between Sunday's 6:00 PM open and Monday's 9:30 bell; this is the gap still open at the bell, which is the one a day trader sees. Set aside and counted: 27 Mondays after a Friday holiday, 10 Tuesdays after a Monday holiday, and 10 Mondays whose 9:30 open was exactly Friday's close (8 on the 5:00 close, 5 on the random price). Mondays on which the exchange ran a shortened session are in, as measured. Friday's 5:00 PM close, the settlement side of the same question, gives the same picture: filled 53.2%, the same distance the other way 54.6%.
How it was measured. The measurement, its references and six predictions were written down before it ran (SETUP 08, September 16, 2026; four of the six held). Script es_sunday_gap-2026-09-16.py v0.1 on the one-minute ES file: 6,514,391 lines, 0 unparsed, 4,801 dates with a 9:30 to 4:59 session, 2008-01-02 to 2026-08-14. The random draw is one price from each Friday's session, seed 3. The script's own print is the record for this page, es_sunday_gap.txt; every number here is read from it.
Raw market data: FirstRate Data (firstratedata.com). A measurement on a file, not advice.