Sessions that gap at all
80.4%
3,121 of 3,882 sessions
Nifty 50 Gap Up / Gap Down — The Full Record
Every overnight gap in the Nifty 50, counted and split by size, with what actually happened next. Measured on the close-to-open move from03 Jan 2011 to 25 Sept 2026 —3,882 sessions with a prior close.
Source: NSE daily index data. A gap is the open versus the previous close; sessions under 0.1% are treated as flat.
Gap up
54.1%
2,099 sessions · mean +0.46%
Gap down
26.3%
1,022 sessions · mean -0.55%
Flat (under 0.1%)
19.6%
761 sessions
The headline: Nifty gaps up about twice as often as it gaps down
Across 3,882 sessions from 03 Jan 2011 to25 Sept 2026, the index opened away from the previous close on80.4% of them. It gappedup 54.1% of all sessions anddown 26.3% — a ratio of roughly 2.1:1. The asymmetry is not only in frequency. The average gap up is+0.46% while the average gap down is-0.55%, so downside gaps are meaningfully larger than upside ones.
The single largest gap up was+4.86% on03 Feb 2026; the largest gap down was-9.14% on23 Mar 2020. Note that the biggest gap down is more than twice the size of the biggest gap up, which is the clearest single statement about asymmetry risk in this data.
A gap up in the 0.10-0.25% band closed beyond the prior close only 53.5% of the time, against 62.6% for a gap down in the same band. As the gap widens the hold rate climbs: 85.3% for gap ups of 1.00-2.00% and 100.0% for gap downs of 2.00-3.00%.
Gap frequency by size — up versus down
Share of all sessions that produced a gap of each size, split by direction. Small gaps dominate by volume; large gaps are rare but overwhelmingly downside-skewed in the tail.
What happens after the gap — fill versus hold
A gap is 'filled' when the session trades back through the prior close, and 'held' when it closes on the far side of it. Small gaps get filled and then abandoned; large gaps are almost never filled.
Gap size distribution
| Gap band | Gap up | % of sessions | % of gap ups | Gap down | % of sessions | % of gap downs |
|---|---|---|---|---|---|---|
| 0.10-0.25% | 664 | 17.1% | 31.6% | 340 | 8.8% | 33.3% |
| 0.25-0.50% | 823 | 21.2% | 39.2% | 321 | 8.3% | 31.4% |
| 0.50-1.00% | 469 | 12.1% | 22.3% | 241 | 6.2% | 23.6% |
| 1.00-2.00% | 116 | 3.0% | 5.5% | 86 | 2.2% | 8.4% |
| 2.00-3.00% | 19 | 0.5% | 0.9% | 21 | 0.5% | 2.1% |
| 3.00+% | 8 | 0.2% | 0.4% | 13 | 0.3% | 1.3% |
| All gaps | 2,099 | 54.1% | 100.0% | 1,022 | 26.3% | 100.0% |
The two most common bands — 0.25-0.50% gap ups and0.10-0.25% gap ups — together account for 38.3% of all sessions. Anything above 2% is genuinely rare: only 61 such sessions in 3,882.
Year by year — how often the index gapped
Count of gap-up and gap-down sessions per calendar year. The step change around 2011 is a data-vendor change in how daily opens were recorded, not a change in market behaviour; see the data integrity note.
Year by year
| Year | Sessions | Gap ups | Gap downs | Flat | Mean |gap| | Worst move | Year return |
|---|---|---|---|---|---|---|---|
| 2006 | 247 | 24 | 16 | 206 | 0.064% | −1.40% | +39.8% |
| 2007 | 249 | 28 | 18 | 203 | 0.066% | −0.93% | +54.8% |
| 2008 | 246 | 37 | 21 | 188 | 0.089% | −1.92% | -51.8% |
| 2009 | 242 | 25 | 20 | 197 | 0.064% | −0.75% | +75.5% |
| 2010 | 251 | 50 | 13 | 188 | 0.116% | −1.46% | +18.0% |
| 2011 | 247 | 117 | 90 | 40 | 0.537% | −3.33% | -25.1% |
| 2012 | 247 | 120 | 83 | 44 | 0.398% | −1.71% | +27.3% |
| 2013 | 248 | 120 | 92 | 36 | 0.396% | −2.48% | +6.2% |
| 2014 | 243 | 128 | 53 | 62 | 0.289% | −2.06% | +31.0% |
| 2015 | 247 | 138 | 57 | 52 | 0.372% | −2.94% | -3.9% |
| 2016 | 246 | 139 | 57 | 50 | 0.381% | −5.57% | +3.1% |
| 2017 | 248 | 149 | 35 | 64 | 0.263% | −1.76% | +28.3% |
| 2018 | 246 | 143 | 51 | 52 | 0.356% | −3.48% | +3.1% |
| 2019 | 244 | 141 | 45 | 58 | 0.279% | −2.38% | +11.8% |
| 2020 | 250 | 167 | 64 | 19 | 0.801% | −9.14% | +14.6% |
| 2021 | 248 | 160 | 51 | 37 | 0.424% | −2.13% | +24.0% |
| 2022 | 248 | 130 | 90 | 28 | 0.567% | −3.01% | +4.1% |
| 2023 | 245 | 134 | 48 | 63 | 0.270% | −1.65% | +19.9% |
| 2024 | 246 | 130 | 63 | 53 | 0.318% | −3.58% | +8.8% |
| 2025 | 248 | 108 | 67 | 73 | 0.291% | −5.00% | +10.5% |
| 2026 (to Sept 2026) | 181 | 75 | 76 | 30 | 0.482% | −4.86% | -11.6% |
2009 was the calmest year by average gap size (0.064%) and 2020 the most volatile (0.801%). Note the mean absolute gap is not directly comparable across the 2011 boundary.
Seasonality — which months gap up more
Share of all gapped sessions in each calendar month that were gap ups, and the average absolute gap size by month.
Seasonal pattern
The gap-up bias is remarkably stable across the calendar. Every month favours gap ups, ranging from 63.9% in Mar to 72.5% in Jul. That narrow spread means there is no strong seasonal edge in which direction the index gaps — the direction is set by overnight news, not the month.
Seasonality does appear in size. March carries the largest average absolute gap ( 0.468%), which lines up with fiscal-year-end repositioning and the annual budget period, while the quieter post-summer months run smaller.
Following the gap: filled versus held
| Gap band | Filled (up) | Held (up) | Filled (down) | Held (down) | Next day (up) | Next day (down) |
|---|---|---|---|---|---|---|
| 0.10-0.25% | 82.2% | 53.5% | 72.6% | 62.6% | +0.02% | +0.01% |
| 0.25-0.50% | 60.8% | 70.2% | 58.9% | 76.0% | +0.10% | -0.07% |
| 0.50-1.00% | 41.4% | 80.4% | 39.0% | 78.0% | +0.12% | -0.01% |
| 1.00-2.00% | 27.6% | 85.3% | 24.4% | 88.4% | +0.40% | -0.05% |
| 2.00-3.00% | 15.8% | 94.7% | 0.0% | 100.0% | -0.46% | +0.21% |
| 3.00+% | 25.0% | 100.0% | 30.8% | 84.6% | -0.94% | +0.78% |
"Filled" means the session traded back through the previous close at some point. "Held" means it finished beyond that level. The two together describe the day's shape: a gap that gets filled and then closes beyond it was reclaimed and re-bought.
The pattern is close to monotonic in gap size. A 0.10-0.25% gap up is filled 82.2% of the time but only holds 53.5%; a 1.00-2.00% gap up holds 85.3% of the time. Large gaps are not reliable to fade for this reason — once the index has moved a full percent overnight, sellers tend to be absorbed rather than overwhelm the open.
Next-day returns carry the same shape. A 1.00-2.00% gap up is followed by a mean next-session move of +0.40%, while the same band gap down is followed by -0.05% — continued but weaker in the gap-up case, and slightly negative in the gap-down case. The strongest continuation appears in the rarest bands, where sample sizes fall into the tens and the figures should be treated as indicative rather than precise.
Why the pre-2011 data is excluded
Median absolute overnight move and share of sessions classed flat, before and after 2011. Before 2011 the feed's daily open was copied from the prior close, so almost every session looks flat — an artefact of the data, not of the market.
Fifteen largest gap ups
| Date | Gap | Close vs prev | Filled |
|---|---|---|---|
| 03 Feb 2026 | +4.86% | +2.55% | no |
| 07 Apr 2020 | +4.48% | +8.76% | no |
| 13 May 2020 | +4.22% | +2.03% | no |
| 17 Apr 2020 | +3.68% | +3.05% | no |
| 03 Jun 2024 | +3.58% | +3.25% | no |
| 27 Mar 2020 | +3.56% | +0.22% | yes |
| 08 Apr 2026 | +3.16% | +3.78% | no |
| 24 Mar 2020 | +3.13% | +2.51% | yes |
| 31 Mar 2020 | +3.00% | +3.82% | no |
| 01 Dec 2011 | +2.87% | +2.17% | no |
| 07 Oct 2011 | +2.79% | +2.88% | no |
| 28 Oct 2011 | +2.69% | +3.05% | no |
| 09 Apr 2020 | +2.56% | +4.15% | no |
| 01 Apr 2026 | +2.54% | +1.56% | no |
| 10 Mar 2022 | +2.52% | +1.53% | no |
Fifteen largest gap downs
| Date | Gap | Close vs prev | Filled |
|---|---|---|---|
| 23 Mar 2020 | -9.14% | -12.98% | no |
| 09 Nov 2016 | -5.57% | -1.31% | no |
| 13 Mar 2020 | -5.03% | +3.81% | yes |
| 07 Apr 2025 | -5.00% | -3.24% | no |
| 19 Mar 2020 | -4.79% | -2.42% | yes |
| 12 Mar 2020 | -4.00% | -8.30% | no |
| 16 Mar 2020 | -3.69% | -7.61% | no |
| 12 Jun 2020 | -3.61% | +0.72% | yes |
| 06 Feb 2018 | -3.48% | -1.58% | no |
| 09 Aug 2011 | -3.33% | -0.89% | yes |
| 04 May 2020 | -3.31% | -5.74% | no |
| 30 Mar 2020 | -3.17% | -4.38% | no |
| 24 Feb 2022 | -3.01% | -4.78% | no |
| 24 Aug 2015 | -2.94% | -5.92% | no |
| 24 Jun 2016 | -2.92% | -2.20% | no |
Data integrity notes
History starts in 02 Jan 2006, not 1995. The Nifty 50 was launched in November 1995, but the daily open series available for this study begins in 2006. Earlier index history exists but carries zero or missing OHLC values, so it cannot support gap analysis. "Entire history" here means the full2006–2026 record, 5,117 sessions.
Daily opens before 2011 are not usable, so the headline figures above start in 2011. In the 2006–2010 portion the reported open is copied from the previous close rather than being a genuine opening print: the median absolute move from previous close to open is0.027% in that era against0.280% from2011 onwards — a difference of roughly an order of magnitude. As a result 79.6% of pre-2011 sessions look "flat" versus 19.6% after. Counting them would understate every band and exaggerate the flat share, which is why the distribution, fill rates and monthly statistics are all reported on the 2011–present window. The year-by-year table is shown in full so the discontinuity is visible rather than hidden.
Weekend-dated bars were removed. 20 bars carrying Saturday or Sunday timestamps existed in the raw feed. NSE sessions run Monday to Friday, so these are artefacts rather than sessions and were dropped before any calculation. The final series contains no weekend-dated rows and averages 244 sessions per year.
Flat sessions are excluded, not discarded. 761 sessions moved less than0.1% between the previous close and the open. They are counted in every base rate but excluded from the bucket tables, so the bucket percentages are shares of all measured sessions and will not sum to 80.4%.
The largest single-bar moves in the raw feed were checked. The extremes resolve to 23 Mar 2020 at-9.14% and03 Feb 2026 at+4.86%, both consistent with known market events rather than bad data.
Methodology
- Gap percentage = (open − previous close) ÷ previous close × 100, measured on the Nifty 50 index level for consecutive trading sessions.
- Sessions with an absolute gap below 0.1% are classified flat. The threshold is fixed in the analysis code, not fitted to the data.
- Bands are 0.10–0.25, 0.25–0.50, 0.50–1.00, 1.00–2.00, 2.00–3.00 and 3.00+ percent, applied to the absolute gap and mirrored in both directions.
- Filled means the session's low (gap up) or high (gap down) reached the previous close. Held means the close finished beyond it.
- Next-day figures use the following session only, so the most recent session has no next-day value and is excluded from those means.
- All figures are computed at build time from a committed daily OHLC file. Nothing on this page is estimated or interpolated.
Data sources
- Nifty 50 Index — historical data and constituents — National Stock Exchange of India
- Nifty 50 index methodology and factsheet — National Stock Exchange of India
- SEBI guidance on advertisements and research — Securities and Exchange Board of India
- The daily OHLC file behind this page was retrieved through a commercial broker market-data API and cleaned as described in the integrity notes above: weekend-stamped bars removed, and daily opens before 2011 excluded from the headline statistics because that vendor did not record a genuine opening print. The public endpoints used to assemble it are documented in the repository's fetch script for anyone wishing to reproduce the figures.
Disclaimer
This page is an educational and informational statistical study of historical index behaviour. It is not investment advice, not a research recommendation, and not an offer or solicitation to buy or sell any security or derivative. Past index behaviour does not predict future performance. Gap statistics describe what happened; they do not constitute a trading strategy. The author is not a SEBI-registered investment adviser or research analyst. Consult a qualified, SEBI-registered professional before making any investment decision.