What the Closing Auction (CAS) has done to Your Close

Two months of NSE Closing Auction Session data, and what it means for a position carried into 3:30.

Since 3 August 2026, the closing price of every stock carrying a derivative contract is not an average any more. It is an auction print.

Continuous trading in those stocks stops at 3:15 pm. Orders collect for twenty minutes. At about 3:35 the exchange matches them at a single price, and that price becomes the official close — the one your portfolio is marked at, the one indices are built from, and the one stock derivatives settle against at expiry.

We have now measured two full months of it. This is what the mechanism actually does.

What happens between 3:15 and 3:40

The last twenty-five minutes of the day now run differently for a stock in the F&O segment than for one outside it.

Two details matter for anyone holding into the close.

The reference price is not the last traded price. It is a fifteen-minute VWAP, so the auction starts from where the stock was trading between 3:00 and 3:15, not from the 3:14:59 tick.

And the 3% band is measured from that reference price, not from the previous close. A stock that has already moved during the day carries its band with it.

How far the auction moves the close

NSE publishes both numbers every day: the reference price the auction inherits, and the final price it produces. The gap between them is the mechanism's own footprint, and nobody had been measuring it.

We did, for every CAS-eligible stock on every session since launch. On an ordinary September session, across about 210 stocks:

Put the other way round: on a normal day, roughly three-quarters of stocks close more than 0.1% away from where they were trading at 3:15. Around 43% move more than 0.25%. About 29 stocks a session move more than 0.5%, and two move more than 1%.

That is not a large number, and it should not be. An auction that routinely threw the close half a percent away from the market would be a broken auction. But it is not zero either, and it is not noise. It is a systematic, daily, measurable difference between the price you watched at 3:15 and the price you are marked at.

There is also a direction to it. In September the auction closed a stock above its reference price 54.8% of the time, with an average signed move of +0.039%. In August the figure was 60.0% and +0.138%. The upward tilt is real in both months and it has more than halved.

Then there are rebalancing days

On 31 August the MSCI August 2026 index review took effect at the close. The auction cleared Rs 39,718 crore, about forty-two times the previous session and 22% of all NSE cash turnover that day.

What happened to prices in that auction is the more interesting part.

On 31 August, 60 of about 210 stocks closed at the 3% price band. More than a quarter of the CAS universe was price-limited in a single auction. The band did what it is there to do, and it was binding for a large part of the market.

It is worth being clear about what that means and does not mean. It does not mean the auction failed. It means the mechanism absorbed a very large one-directional index flow and the band constrained where a lot of it could print. On an ordinary session the same band binds three times in ten thousand.

The largest single move we recorded on an ordinary September session was POLICYBZR on 24 September, from a reference price of Rs 1,244.50 to a close of Rs 1,207.20. That is exactly −3.00% — the band, on a normal Thursday.

Read our earlier report - https://vtrender.com/posts/closing-auction-session-cas-the-cash-close-is-only-part-of-the-change

The twenty-five minutes when futures are the only continuous market

This is the part that matters most for an Orderflow trader, and it follows directly from the timings.

From 3:15 to 3:40, the cash market in an F&O stock is not trading continuously. It is in a call auction. Orders go in, nothing matches until the equilibrium is struck, and no continuous tape exists.

The futures contract on that same stock keeps trading throughout, right through to 3:40.

So for those twenty-five minutes the futures book is the only place a continuous price is being made on that underlying. Whatever the market thinks the auction will produce is being expressed there, in a live order book, five minutes after the cash market has gone quiet and five minutes after the auction price is known.

That is a structural window, not an opinion. It did not exist before 3 August. And it is visible on a futures Orderflow chart in the ordinary way — the same delta, absorption and imbalance reading you would apply at any other hour, in a window where the cash market cannot contradict it tick by tick.

We are not going to tell you what to do with it. Two months of end-of-day data cannot say anything about intraday behaviour in that window, and we do not have intraday measurements to publish. What the data does establish is that the window is real, that the auction it lead

Expiry settlement now flows from the auction

A stock future or stock option settles at expiry against the closing price of the underlying in the cash market. That closing price is now the auction's equilibrium price.

So a position carried through expiry settles at whatever the auction discovers on that day, not at a thirty-minute average of the tape. The auction on 29 September, the September monthly expiry, moved the average stock 0.872% from its 3:15 reference price and moved 64 stocks by more than 1%.

Index derivatives inherit this too, one step removed. The Nifty's closing value is built from the closing prices of its constituents, and most of those constituents now close through the auction.

If you carry single-stock positions into expiry, this is the single most practical change the new mechanism has made.

What two months say, and what they do not

The auction is settling down. Measured on ordinary sessions, the average move fell from 0.323% in August to 0.269% in September. The share of stocks closing at the 3% band fell from 0.17% to 0.03%. The upward bias more than halved.

A caution on that first number, because it is the one most likely to be quoted. About half of the August-to-September fall is August's own first week. The mechanism was new on 3 August and the auction moved the close 0.670% that day, decaying to 0.285% by 6 August and staying there. Measured from 10 August onward, the fall to September is 7.5% rather than 16.9%. Both are in our report. The smaller one is the durable improvement; the larger one includes a settling-in that by definition happens once.

The futures-to-cash relationship tells the same story from a different direction. The dispersion of the day-to-day change in the basis — the gap between a stock's futures close and its cash close — ran at 0.279% in July before the auction existed, widened to 0.502% in its first month, and has come back to 0.369% on ordinary September sessions. On the three index-event sessions it ran at 1.056%.

Two measures, computed independently, both about 16% tighter on ordinary days. That is the finding.

What two months cannot tell us: whether this continues, whether the remaining gap to July is permanent, or whether anything here is caused by the auction rather than merely coincident with it. We measure it again next month.

What to do with this

Three practical things follow from two months of measurement.

1. The price at 3:15 is not the close. On a normal day the difference is a couple of rupees on a thousand-rupee stock. Know that it exists rather than being surprised by it.

2. Watch the index calendar. Rebalancing days are when the auction stops being a formality. MSCI in August, FTSE and the Nifty semi-annual in September. Those are published in advance.

3. The 3:15 to 3:40 window is now a futures-only window. If you read Orderflow, that is twenty-five minutes where the futures book is the only continuous market on the underlying.

We publish the full study monthly. It covers the auction, the futures-to-cash basis, cash and derivatives turnover, and where single-stock participation is concentrating.

For a companion piece on reading the auction screen itself, see Reading the NSE CAS screen, left to right.

Vtrender Market Structure Desk Vtrender Technologies Private Limited

Data: NSE Closing Auction Session archives and NSE UDiFF Common Bhavcopy Final, 3 August to 29 September 2026.

Market-structure research and charting context only. Nothing here is a buy or sell recommendation, and nothing here says what any price will do next.