What 10 years of NIFTY data says about gap-up days

6 min read · data: 437-stock universe · descriptive study, not advice

Every trading morning the same question shows up in every chat group: “gap-up hai, ab kya hoga?” Folklore has two confident answers — “gaps always fill” and “strong gaps never look back” — and they cannot both be true. So instead of picking a side, we measured it.

The setup

We took every NIFTY 50 session since June 2016 where the index opened more than 0.5% above the previous close — 412 instances in roughly 2,470 trading days, or about one session in six. For each one we recorded three things: did the index close above its open (gap held), did it trade back to the previous close intraday (gap filled), and what the market regime was on the day before.

What the data says

The honest numbers, 2016–2026:

• 57% (236 of 412) closed above their open — the gap held into the close.
• 23% (95 of 412) filled the gap the same session — “gaps always fill” is simply false on a same-day basis.
• 20% (81 of 412) did something in between: faded part of the gap but neither held the open nor tagged the previous close.

So the folklore splits the difference. Gaps held more often than they filled, but a 57/23/20 split is far from a sure thing in either direction.

Context changed the base rate more than the gap size did

The single biggest modifier was not the size of the gap — it was the state of the market before the gap. When the index was already in an established uptrend (above its rising 50-day average), gap-ups held into the close in roughly two out of three instances. When the same gap appeared inside a downtrend, the hold rate dropped to nearly a coin flip, and same-day fills almost doubled. The gap was the headline; the regime was the story.

Gap size mattered less than expected: very large gaps (>1.2%) filled same-day slightly less often than small ones — the opposite of the “too far, must revert” intuition.

The limitation

This is an index-level study of 412 instances across one decade and a handful of regime states — slice it thinner and the sample sizes stop being honest. It says nothing about any individual stock, nothing about tomorrow’s gap, and nothing about what anyone should do. Base rates describe the past; they do not predict the future. What they can do is replace folklore with a measured prior — which is the entire point of this site.