The last twenty minutes of an expiry day changed. Here is what we measured.
On 3 August 2026, NIFTY stopped trading continuously at 15:15 and started closing by auction. The move into the close more than doubled. Then the options market repriced it inside a week. We measured every auction session since, against an 81-session control.
The index stops moving before the day ends
Before CAS, the closing price was a volume-weighted average of the last half hour. Now the closing price of every F&O-eligible stock — and therefore the closing value of NIFTY itself — is set by a single call auction.
The consequence is a discrete jump that cannot be hedged through. Whatever an option is worth at 15:13 is a bet on an auction that has not happened yet.
The move into the close roughly doubled
For every session we took the index at 15:13 — two minutes before the freeze — and compared it to that day's official close. If anything the pre-CAS figure is overstated: back then the official close was a smoothed half-hour average, and we are comparing against the last tick.
| Window | Sessions | Mean | Median | Largest | Over 50 pts |
|---|---|---|---|---|---|
| Before CAS · Apr–Jul | 81 | 19.1 | 14.6 | 84.4 | 6.2% |
| With CAS · Aug–Sep | 25 | 49.8 | 33.9 | 199.0 | 32.0% |
| With CAS, excluding first week | 20 | 39.6 | 31.2 | 82.4 | 25.0% |
One pattern we cannot yet explain
18 of the 25 auction jumps were upward — 72%, mean +26 points — over a stretch in which NIFTY fell 678 points. So it is not simply a rising market. It weakens once the first week is excluded (13 of 20), and 25 sessions is nowhere near enough to call it real. We are still collecting.
The options market repriced it in about a week
The obvious reaction to a bigger jump is to buy cheap out-of-the-money options just before the auction and wait. That depended on them staying cheap. They did not.
| Era | Typical premium | What it reflects |
|---|---|---|
| Before CAS | ₹0.05 – ₹0.30 | Seventeen minutes to a smoothed average settlement — almost no chance of finishing in the money. |
| With CAS | ₹4 – ₹48 | A discrete auction that has moved the index by 30 to 199 points. Priced accordingly. |
That is a twenty- to two-hundred-fold repricing, and it happened within days of the first auction. The premium moved with the payoff. A rule that would have been worthless before CAS is now possible — which is not the same thing as profitable.
Only expiry day has room in it
A natural next question is whether the auction jump is worth anything on ordinary days, not just expiry. To test it we priced the same simple defined-risk structure every session — one that can pay at most 50 points — and asked how much of that 50 the premium already consumed at 15:13.
With one to six days left, NIFTY is near-certain to finish more than 50 points from where it sits now, so a structure like this is already worth almost its maximum. A 30-point auction jump cannot move something that is already at its ceiling. Only on expiry day — when there are seventeen minutes left and the outcome is genuinely open — is there room for the auction to matter.
Not a day-of-week effect
It is tempting to read this as “Tuesdays are different”. They are not. NIFTY's weekly expiry is Tuesday, so in this window day-of-week and days-to-expiry are the same variable wearing two names. Nothing in the data separates them, and a rule built on the weekday would be building on the expiry.
Four things worth knowing if you trade the close
- Settlement is the auction, not the average. If you hold an index option into expiry, what settles it is the single auction price — not the last half hour of trading.
- An intraday product can square off before the auction. A position held as MIS may be closed out by your broker's intraday cutoff, which now falls before the price that decides the day. If you intend to be there for the close, the product type matters more than it used to.
- Spreads at 15:13 are not what they were. The options that used to cost a few paise now carry real premium, and real bid-ask. On a four-legged structure you cross that spread eight times round trip — enough, in our own testing, to turn a positive result negative on its own.
- Between 15:15 and 15:35 you are trading against an indicative price. Options stay open while the index is frozen, and they reprice sharply as the auction's indicative equilibrium moves. By 15:29 they sit within a rupee or so of their settlement value — the uncertainty is largely gone by then.
How this was measured
- Data. One-minute NIFTY and NIFTY option bars from Tradetron's own historical market-data store — the same data our backtesting engine runs on.
- Window. 25 auction sessions from 3 August to 4 September 2026, against 81 control sessions from April to July.
- Settlement. Expiry values are taken at intrinsic against the auction close. We checked this against real traded prices: by 15:30 every leg we looked at traded within about a rupee of intrinsic.
- The control is conservative. Before CAS the official close was a half-hour VWAP; using the last tick instead makes the pre-CAS move look larger, which works against the finding rather than for it.
- Sample size is the real limit. CAS is five weeks old. Five expiry days is not enough to conclude anything about a trading rule, and we have deliberately not published one. These are observations about market structure, and we will revisit them with more data.
What we are not saying
This page describes what changed in how NSE closes the day, and what we measured in the prices that resulted. It does not recommend a strategy, and nothing here is a prediction or investment advice. Historical and simulated results do not indicate future returns.