CAS, One Month In: What Actually Happened, and What SEBI Is Now Reviewing

In late July we published a plain-language explainer on SEBI's Closing Auction Session, before it went live. That piece was about the mechanics — what would change on August 3.

This one is about what actually happened in the first month.

Most of what follows was written in mid-August, after two weeks of live sessions. We've added the August 27 monthly expiry and the September 4 regulatory development at the end, rather than rewriting the earlier observations with hindsight.

The short version: the mechanism does what it was designed to do, and a thin auction pool behaves the way a thin auction pool behaves. That combination produced some genuinely large closing moves. SEBI has now said it will review the settlement-price methodology — which is the system working as it should, not a sign that something is broken beyond repair.

What happened on day one

On Monday, August 3, the first day of CAS, the Nifty 50 moved more than 200 points during the closing auction itself — not during the trading day, during the ~20-minute window after continuous trading in F&O stocks had already stopped.

The tell that this was mechanism, not news: Nifty August futures moved only about 30 points in the same stretch. The cash index leapt; the futures — which kept trading continuously right through — barely shrugged. Those two numbers are supposed to track each other closely. When they don't, the gap is usually telling you something about how a price is being formed, not about what the market thinks anything is worth.

The magnitude moderated over the following four sessions. But it didn't go away.

The divergence nobody had priced in

Here's the statistic that got market-structure people's attention.

On debut, the Sensex and the Nifty closed 0.9 percentage points apart. Two benchmark indices, same country, same session, same large-cap names doing most of the heavy lifting — closing nearly a full percentage point apart.

Per Business Standard's analysis of data since January 2010, that was the second-largest closing divergence between the two indices on record — behind only the 105 basis-point spread of March 26, 2020, in the thick of the COVID crash.

That comparison is worth sitting with. A routine Monday in August produced an index divergence rivalling the most disorderly week in modern Indian market history. Not because anything happened. Because the plumbing changed.

It also wasn't a one-day artifact. The divergence persisted into a fourth consecutive session, and widened again on August 13, a Sensex expiry day, when Nifty closed red and Sensex closed green.

At the individual stock level, the median price difference between NSE and BSE hit 0.86% on Monday, narrowing to 0.58% on Tuesday. Those gaps were large enough and persistent enough to open genuine arbitrage for high-frequency participants — which is, notably, the exact category of edge the reform was meant to close.

Why the two exchanges diverged is still contested, and worth being honest about rather than confident. SEBI's framework nominally covers NSE, BSE and MSEI alike. Some analysts quoted in the press — including Ventura's head of research, who called the divergence "structural, not accidental" — attribute it to the two exchanges not discovering the close the same way over overlapping windows. Zerodha's own post-launch thread suggested it may instead reflect differences in index constituent composition. Both explanations are circulating; the divergence itself is not in dispute.

Why a thin pool moves so much

The auction concentrates an entire day's closing interest into one match. That is the design. It only produces a better price than a 30-minute average if enough orders show up.

In the first week, they didn't.

Goldman Sachs put it directly: weak participation meant even limited order flow could produce disproportionately large changes in the auction-determined closing price. Proprietary trading desks and high-frequency firms — normally the ones providing depth — scaled back or stayed out while they worked out how the new session behaved.

So the first sessions ran with the mechanism that concentrates all closing pressure into one price, minus the participants who would normally absorb it. A small, one-sided cluster of orders could set the close for a large-cap stock, and through it, the index.

This should thin out as participation normalises. It is a genuine bootstrapping problem, not a permanent flaw. But "it will settle down" is not much comfort if your position settled last Thursday.

The part that actually affects your P&L

Now the mechanic that matters most for anyone carrying F&O positions into a close.

Walk through the session as it exists today for an F&O stock:

Time (IST)	What's happening
3:00–3:15	Normal continuous trading. Exchange computes the reference price.
3:15–3:20	Transition. Trading in the stock has stopped. No order entry.
3:20–3:25	Order Entry I — market and limit orders accepted.
3:25–3:30	Order Entry II — limit orders only. Closes at a random moment between 3:28 and 3:30.
3:30–3:35	Orders matched at the single equilibrium price. That becomes the official close.
to 3:40	F&O contracts keep trading continuously the whole time.

Read the last two rows together, because that's the whole story:

The underlying stops. The derivative doesn't.

For roughly 20 minutes, the cash price on your screen is frozen at its 3:15 level while the options and futures written on it trade live. On expiry day, the closing price being formed inside that frozen window is the number your contract settles against — it decides ITM or OTM.

One trader in Zerodha's own comment thread described a stock quoting 8338–8340 at 3:15 that printed 8585 when the auction cleared — roughly a 3.5% move, arriving as a single print, in an instrument that had stopped trading. If you were short a strike in between, you found out at 3:35.

And because every large index constituent is an F&O stock sitting in its own auction, the cash index itself effectively freezes too — then re-prints when the auctions clear.

That's the blind window. Not a bug in any one broker. A structural consequence of stopping the underlying while its derivatives keep trading.

The number the exchange publishes and you probably can't see

Here's the genuinely useful, under-reported part.

The exchange is not silent during those 20 minutes. Per the operational circular, exchanges disseminate continuously throughout the order-entry phases:

• the indicative equilibrium price for each stock
• cumulative buy and sell quantities
• the imbalance quantity and which side it's on
• an indicative index value

So the window is only dark on your end. The exchange is broadcasting a live running estimate of where the close is heading — including a finished indicative index number that nobody has to reconstruct.

The gap is distribution. Most retail platforms simply aren't surfacing these fields yet. Zerodha has acknowledged the data doesn't currently appear in Kite's market depth and has said it's being worked on. Traders on multiple platforms reported the indicative prices weren't visible anywhere in their apps during the first week.

This is worth being precise about, because it changes what the complaint even is. The common framing — "retail is blindfolded while institutions see the flow" — is only half right. The data is public and it's being published to everyone. It just hasn't reached most retail front-ends yet. That's a software backlog, not a structural privilege. Which also means it's fixable, and probably will be.

If you trade positions into the close, it's a reasonable thing to ask your broker for by name: indicative equilibrium price and imbalance quantity during CAS.

What the impact estimates look like

Two numbers worth knowing, both from market participants rather than regulators:

• Zerodha estimated a 1–5% reduction in industry-wide brokerage revenues.
• Jefferies projected a 10–20% decline in expiry-day contracts, which could pull overall options volumes down 5–10%.

Both were estimates made in the first fortnight, about a mechanism that was barely two weeks old at the time. Treat them as informed guesses at direction, not forecasts. But the direction is consistent: less expiry-day activity, because the final settlement print is less predictable than it was.

Is this getting rolled back?

No — and that was SEBI's consistent position through August.

Senior officials, including board member KVR Murty, characterised the early sessions as "teething issues" and urged brokers to move faster on their technology upgrades. The mechanism stays; the adaptation was expected from brokers, platforms and traders.

There's also a second, smaller change already scheduled: the pre-open auction gets its own randomised cutoff from September 7, 2026. Same logic, other end of the day.

---

The first monthly expiry — August 27

Added after the original piece. This is the session that moved the conversation on.

August 27 was the first monthly derivatives expiry under CAS, and it produced the largest closing-auction move so far.

The Sensex traded above 77,100 for most of the day. During the auction window the indicative price fell as far as 74,988 — a swing of roughly 2,100 points, about 2.7% — before recovering to close at 76,933.6, down 0.70% on the day.

Because index options settle against that close, near-the-money puts repriced sharply in the process:

Sensex put	Move during the window
76,400	+446%
76,300	+370%
76,500	+350%
76,600	+335%
76,900	₹4.70 → ₹200 → expired at zero

Bankex saw a larger relative move, falling 1.67% against Bank Nifty's 0.5%, and its 65,000 put went from about ₹6 to roughly ₹1,000 in minutes.

Why it hit so hard: this is the twenty-minute gap described earlier, on the day it matters most. F&O stocks stop trading at 3:15, so the index is effectively frozen — but index options keep trading to 3:40, and on expiry day the auction print decides which contracts finish in the money. A position can move a long way while the underlying you would normally hedge against isn't quoting.

That's a real structural consideration for anyone carrying expiry-day positions, and it's worth planning around rather than being surprised by.

On the day itself, SEBI chairman Tuhin Kanta Pandey said there would be no changes to CAS.

What changed on September 4

Eight days later, that position moved.

SEBI announced it will review the methodology used to determine settlement prices for derivative contracts on expiry, after monitoring the first month and gathering feedback from exchanges, brokers, traders, mutual funds and FPIs. A consultation paper is expected in about a week.

Note what is and isn't under review. The closing auction itself is not being rolled back. What's being examined is how derivatives settle against it — which is precisely the gap this article has been describing: the index freezes at 3:15, the contracts written on it trade until 3:40, and the settlement print is formed in between.

It's worth reading this as the process working. A structural reform went live, the first month surfaced problems that weren't visible in advance, participants fed that back, and the regulator opened a formal review inside five weeks. That is a faster correction loop than most market-structure changes get.

Nothing changes until the consultation paper is published and any resulting rules take effect. Until then the mechanics below still apply exactly as they do today.

How to think about it from here

Not advice — just what the mechanics imply.

Assume 3:15 is the real close for F&O stocks. Anything you wanted to do at a known, continuously-quoted price in the underlying needs to be done before then. After 3:15 you're not trading a price, you're submitting into an auction.

Treat expiry-day close as an auction outcome, not an extrapolation. The old 30-minute VWAP could be estimated by watching it form. A single-price match cannot be, and the last continuously-traded price is a weaker guide than it used to be — especially in names where the auction pool is thin.

Watch the synthetic, not the stale spot. During the freeze the cash number on your screen is a 3:15 fossil. The options on it are live. The synthetic forward — ATM strike plus call minus put — keeps updating right through the auction, because its inputs never stopped trading. It's not a prediction of the equilibrium price, and it shouldn't be read as one. But it is a live read of where the options market is pricing that underlying, at a moment when the cash tape has nothing to say.

Thin names deserve more caution than liquid ones. The whole failure mode here is small order flow moving a concentrated match. That risk scales inversely with how crowded the auction is.

Where FNODATA stands on this

Two pieces of this are already live in the product, and it's worth being exact about what is and isn't built.

Our market calendar now models the closing auction as its own state: continuous trading to 15:15, an auction window through 15:35, and derivatives running to a 15:40 close — instead of the old assumption that everything stopped at 15:30. The feed stays live through the auction, because options are still trading.

And the synthetic forward described above — ATM + (CE − PE), with a sanity guard that discards the value if a stale quote pushes it implausibly far from spot — is computed from the live chain and already used in the app's strike logic.

What is not built yet is the UI treatment: visibly flagging the frozen spot and promoting the market-implied number during the auction window. That's specced and in progress, not shipped. We'd rather say that plainly than let a blog post imply a button that doesn't exist.

What we cannot give you today is the exchange's indicative equilibrium price or indicative index value. Our data comes through your own broker connection, and no broker we integrate with currently exposes those fields. We've raised it with them. When a broker ships it, we can surface it.

---

This article is general market-structure education, not investment advice, and FNODATA is not a SEBI-registered investment adviser. It is not a recommendation to buy, sell, or hold any security or derivative contract. Figures cited are from published reports between August and early September 2026, some of them early estimates by market participants. SEBI's settlement-price review is at consultation stage and nothing has changed yet — confirm exact timings and mechanics against the exchanges' official circulars and your broker.