CRTLAB Answers · measured, not opinion

How often do gaps fill on the Nasdaq, S&P 500 and Dow?

Name: How Often Do Gaps Fill? 6,984 Index Gaps Measured
Creator: CRTLAB
License: https://creativecommons.org/licenses/by/4.0/

Answered with real trades on 3 markets of CRTLAB's own price history, 2017–2026. The rules were fixed before any result was seen, and every one of them is on this page.

The answer

58.4% of opening gaps on the Nasdaq 100, S&P 500 and Dow filled the same day, across 6,984 gaps since 2017 — 82.0% of small ones (under 0.25%) but only 22.8% of gaps over 1%. A level the same distance on the other side was reached 55.5% of the time, so gaps fill a little more than chance, not much.

Gaps measured6,984

Filled same day58.4%

Mirror-level control55.5%

Filled by 10:3039.2%

Measured 2 November 2026 · data to 19 October 2026Replay a gap open on NAS100

By gap size

Gap size as a percentage of the previous 16:00 price. Small gaps sit close to price, so they fill — and so does a level that close on the other side.

Gap	Gaps	By 10:30	By 12:00	By 16:00	Mirror level (16:00)
Under 0.25%	2,694	68.2%	76.2%	82.0%	77.5%
0.25–0.5%	1,878	32.9%	45.0%	56.9%	55.4%
0.5–1%	1,565	16.2%	27.0%	38.8%	38.0%
Over 1%	847	3.5%	11.3%	22.8%	18.2%

Gap up or gap down

Direction	Gaps	By 10:30	By 12:00	By 16:00	Mirror level
Gap up	3,910	37.6%	47.6%	56.4%	52.9%
Gap down	3,074	41.2%	50.7%	60.9%	58.8%

Every index

Index	Gaps	By 10:30	By 12:00	By 16:00	Mirror level
Nasdaq 100	2,343	41.2%	50.4%	58.7%	55.7%
S&P 500	2,325	36.6%	47.2%	57.5%	54.3%
Dow Jones (US30)	2,316	39.8%	49.3%	59.0%	56.4%

What this means for your trading

"Gaps always fill" is not true the same day: big gaps mostly do not.
Gaps beat the mirror level by about 2.9 points — a lean, not a magnet. A gap-fill trade needs more than the gap.
Most fills that happen, happen early: by 10:30 is already more than half of the day's fills.

Test it yourself

NAS100 on the 5-minute chart. Pick a gap day, replay from 09:30, and decide whether this one fills. The chart opens in your browser with nothing to install — 10,000 candles free, no card. Replay a gap open on NAS100.

How we measured it

CRTLAB's own 5-minute price history for the Nasdaq 100, S&P 500 and Dow futures, 3 January 2017 to 19 October 2026. The gap is the 09:30 New York open against the previous trading day's 16:00 price (the close of the 15:55 candle) — the cash-session gap, measured on the futures. Gaps under 0.02% are not counted.

A gap fills when price trades back to the previous 16:00 price, counted by 10:30, by 12:00 and by 16:00 the same day.

THE CONTROL is a mirror level the same distance from the 09:30 open on the far side, walked the same way. Every rule was fixed before any result was seen.

Frequently asked questions

How often do gaps fill?

Do all gaps fill?

No. Same day, 58.4% filled; of gaps over 1%, only 22.8%.

Do gap downs fill more than gap ups?

Gap downs filled 60.9% of the time and gap ups 56.4% the same day.

→RelatedAll answers

Does the ICT Silver Bullet work?No edge on its own. Across 46,392 Silver Bullet trades on 13 markets since 2017, the setup averaged +0.02R a trade in the AM window — breakeven before spread and commission.What time is the high or low of the day made?Between 08:00 and 12:00 New York. Those four hours printed the day's high or low 27.8% of the time across 27,187 trading days — about 2.2× what the same hours would get if the day's moves came in a random order (12.5%).Is the ICT Judas swing real?Not on average. The first move after midnight New York pointed the OPPOSITE way to the day's move on only 43.0% of 26,125 days — it continued more often than it reversed, in every one of 13 markets. After a large early move, the day went the same way 74.0% of the time.Does the opening range breakout work?Slightly, before costs. Trading the first break of the 09:30–09:45 New York range — stop at the other side, held to 16:00 — averaged +0.09R ± 0.01 a trade over 26,894 trades, and was positive on 13 of 13 markets. The win rate is low; the edge comes from the days that run.Which day of the week makes the weekly high or low?Monday and Friday — 55.6% of weekly highs and 58.3% of weekly lows across 4,767 weeks. But the same weeks with their days shuffled into a random order give 58.5% and 57.7%: the Monday/Friday pattern is the arithmetic of the week's two ends, not market behaviour.How often is the previous week's high or low taken?87.6% of weeks took last week's high or low, across 4,005 weeks since 2017: the high 52.5%, the low 44.8%, both 9.8%. When it was taken, it was most often on Monday — 51.7% of the weeks that took the high.Does price return to the midnight open?About as often as it reaches any level the same distance away. From 08:30 New York, price came back to the midnight open on 58.2% of 25,401 days by 16:00 — and reached a mirror level, the same distance on the other side, on 58.3%. The midnight open is not a magnet; it is reached because it is close.Do ICT macros work?No more than the time around them. Across 214,928 macro windows on 13 markets since 2017, a macro moved 0.98× as much as the same-length windows just before and after it, and out-moved both neighbours about a third of the time — what chance gives.Does yesterday's daily candle predict today's?Barely. After an up day, the next day closed up 50.5% of the time; after a down day, 53.8% — against 52.1% for every day, across 26,801 days on 13 markets. If anything, a down day is followed by a slightly better chance of an up day, not a down one.Does London sweep the Asian range and reverse?Usually not. After London (02:00–05:00 New York) took one side of the Asian range, price went on to the other side first on 41.1% of 9,037 days, and ran a full Asian range further the same way first on 44.7%. Both levels are the same distance away, so a reversal pattern would show the first number well ahead; it is behind, on 11 of 13 markets.What time does gold make its high or low of the day?Most often between 08:00 and 12:00 New York. Those four hours printed gold's high or low of the day 31.3% of the time across 2,429 trading days since 2017, 2.5× what the same hours get when each day's hours are shuffled into a random order (12.6%). The strongest single hour was 09:00–10:00.Does the 8:30 news candle show the direction of the day?It is big, but it does not tell you the day's direction. The 08:30 New York candle averaged 1.8× the range of the candles around it (3.7× on jobs-report Fridays), yet the 08:30–09:00 move pointed the same way as the rest of the US day on only 49.4% of 26,239 days, against 49.9% for the same test an hour earlier. A coin gives 50%.Does the London open set the direction for the day?No. The first hour of London (02:00–03:00 New York) pointed the same way as the rest of the London and New York morning on 50.1% of 26,246 days — a coin flip, and the same as the hour before London opens (50.1%). Following it averaged 0.00R ± 0.01 a trade.Does the opening range breakout work on gold?On gold, slightly, before costs. Trading the first break of the 09:30–09:45 New York range — stop at the other side, held to 16:00 — averaged +0.12R ± 0.04 a trade over 2,394 trades since 2017, number 4 of 13 markets tested. With a 2R target it averaged +0.09R ± 0.03.What time does EUR/USD make its high or low of the day?Most often between 08:00 and 12:00 New York. Those four hours printed EUR/USD's high or low of the day 29.5% of the time across 2,478 trading days since 2017, 2.4× what the same hours get when each day's hours are shuffled into a random order (12.5%). The strongest single hour was 10:00–11:00.Previous day high & low statisticsFree tool Average daily rangeFree tool

The average says what usually happens. Replay tells you if your version beats it.

Replay real market history candle by candle with the future hidden, take the setup by your own rules, and every trade is scored for you — on 38 markets and every timeframe.

Replay a gap open on NAS100

10,000 candles freeNo card to startData & charts

Databento

TradingView