Big gaps don't sustain: Indian stocks that open far from the last close mostly hand it back
Every trading morning a few dozen Indian stocks open at a price nowhere near where they closed the previous day. Overnight news, a foreign listing that moved while India slept, an order imbalance in the pre-open auction — something repriced them before the first tick. Anyone looking at that list wants one thing answered: does the jump keep going, or does it give the money back?
Strota put the question to its own data, and the answer is almost entirely about size. Small gaps drift a little further the way they opened; large ones reverse, and the bigger the gap the more reliably they reverse. Stocks that opened at least five percent below the previous close finished above their opening price 79% of the time, surrendering an average of 1.85% by the bell. Taken blind, gaps are a coin toss: half close beyond the open in the direction they gapped, for an average move of 0.06%.
The first of three studies swept every gap day in India's derivatives-eligible stocks across six months — 11,212 of them — and sorted those days by how far the open sat from the prior close. Each measurement runs forward of the 09:15 open, so nothing in the result leans on information that was unavailable at the moment of entry. A round-trip cost of 0.20% served as the line a result has to clear before it means anything.
What held up was a narrow band. Gap-ups of two to three percent, a sample of 533 days, continued 54% of the time for an average of 0.27%. Gap-downs of one to two percent were the pick of the lot: 1607 days, continuation 57%, average 0.40%. Past three percent the sign flips both ways. The 109 gap-ups beyond five percent lost 0.87%, and gap-downs of two to three percent were already negative at 0.25%. At the violent end sit the 84 gap-downs beyond five percent, at 1.85% against the way they opened.
A second study went after a more appealing idea. Indian gaps often echo whatever New York did while the market here was shut, and the prior American session is settled by roughly 01:30 in the morning, long before anyone here can act on it. If Indian traders under-react to that, riding the gap ought to pay. Across 51,481 stock-days over a year, the correlation between the overnight American move and the Indian open-to-close move came in at 0.05, and 0.07 on days with a real gap. That is noise: the overnight information is already inside the opening price.
Riding gaps the overseas tape agreed with fared no better. Gaps of two percent or more faded whether they pointed the same way as the American close, at 0.29% against the position, or the opposite way, at 0.28%. Where the overseas link does earn its keep is price discovery: for the five Indian stocks with liquid American listings, the Indian open absorbs roughly a third of the foreign move before the first trade prints, and when it overshoots, the stock drifts back toward that price during the day. Real, but small, and five names is a curiosity.
The third study asked something far cheaper to check: is the stock's sector index gapping the same way? That is knowable at the open, and it carries real information. Most gaps — 86% of the 14,689 observations across a year — move with their sector. The minority that do not behave very differently.
Gaps that fight their sector fade hard. In all four buckets at two percent or less, a gap the sector did not share ended the session roughly half a percent underwater: 0.52% for the smallest gap-ups, 0.51% for gap-ups of one to two percent, 0.26% for the smallest gap-downs. Comparable gaps that moved with their sector finished close to flat. A move nobody else in the sector is making looks like an idiosyncratic opening-auction imbalance, and imbalances revert.
One continuation setup survived the filter. Gap-downs of one to two percent moving with their sector closed beyond the open 53% of the time for an average of 0.26%, and 49% of them cleared the cost line. Gap-ups of two to three percent with their sector managed 0.03%, which does not cover the cost of trading them. Anything of three percent or more faded regardless of what the sector did; size simply overwhelmed the filter.
Read together, the studies produce a rule better at saying no than yes. Gaps of three percent or more reliably went the wrong way, large gap-downs worst at 1.2%; so did gaps of two percent or less that their sector was not sharing. Under one percent, a gap says nothing. The single ride setup is real, and thin.
None of the caveats here are cosmetic. The entry price is the official open on a daily bar, while a real fill lands seconds or minutes later, and the continuation results — the whole span from 0.12% to 0.40% — sit close enough to the 0.20% cost line that a little slippage erases them. Only the fade side, from about half a percent out to 1.85%, carries margin worth the name.
Regime is the sharper worry. The year behind the second and third studies was net mean-reverting for gaps, continuation rates mostly under half, which makes the ride buckets the likeliest artifacts of their window: the best of them printed 0.40% over six months and 0.26% over the year. Sample sizes on the against-sector side are thin as well — 182 to 974 observations in the small-gap buckets, as few as 33 in the large ones, which is why the single cell that inverts the pattern should be read as nothing. The sector mapping is rough too: 91 of 211 symbols had no clean sector index and were parked in the broad market, which blunts the signal rather than inventing it.
And the feature that might matter most is still missing. Whether a gap opens on genuine early participation or on almost no volume cannot be tested here, because the available minute-by-minute history reaches back only about eight days. Whatever that is worth will have to be established live, session by session, rather than claimed from the past.
More money stories
- Volume was the only thread in our biggest movers, and it was unknowable at entry
- Buying weakness worked, buying strength lost: results from a broad signal sweep
- An 80% win rate collapsed to 31% the first time we tested it forward
- Crypto's momentum formula loses money on Indian stocks. A six-month version doesn't.
- Two of eight sessions made the entire profit — and the short side lost money
- Low-volatility stocks, not momentum, produced a book positive in every completed year