On September 3, hours after the Sensex weekly expiry closed, SEBI issued a press release confirming what four weeks of Closing Auction Session (CAS) data had already been pointing toward: the regulator is reviewing how derivative settlement prices are determined under the new mechanism. A consultation paper is expected within a week.
The timing is instructive. CAS went live on August 3, replacing the old 30-minute VWAP-based close with a discrete call auction - continuous trading in F&O stocks stops at 3:15pm, a reference price is built over the next few minutes, and a single clearing price is set sometime between 3:20 and 3:30pm. The stated goal was a cleaner, harder-to-game closing print. On two of the four weekly Sensex expiries since launch, the auction window itself has produced the opposite.
Two Thursdays, Same Signature
On August 27, the Sensex's low for the CAS window came from an illiquid Indicative Equilibrium Price (IEP) pinned at 74,964 - the exact -3% floor of that session's band. The index swung 2,171 points during the auction window before settling.
On September 3, the pattern repeated. The low came from an illiquid IEP at 74,268 - again, precisely the -3% floor. This time the swing was wider: 2,658 points.
Two consecutive Sensex expiries. Two auctions pinned to the exact mathematical edge of the permitted band. Both producing four-figure index swings inside a ten-minute window.
Size isn't the variable. Liquidity is. The obvious explanation - that large flows cause large swings - doesn't survive contact with the month's own data. On August 31, the MSCI India rebalance ran through the same mechanism: roughly Rs 39,000 crore of CAS value, an order of magnitude above a normal session. That auction cleared in an orderly fashion.
Today's expiry auction did the opposite on a fraction of the size. Sensex CAS turnover came to roughly Rs 126 crore - thinner than an ordinary non-expiry day - and the index moved 2,000 points getting there.
The contrast is the finding: a large, genuinely two-sided flow settles calmly. A thin, one-directional flow on an ordinary expiry Thursday does not. The problem isn't volume. It's what has to trade against that volume - Index derivative open interest running into the Rs 30,000-60,000 crore range, settling off a cash auction that, in a normal week, clears barely a tenth of that in matched value.
The Gap Derivatives Are Trading Across
The mechanical issue underneath both sessions is timing, not size. Index derivatives keep trading in continuous mode through the CAS window - until 3:40pm. For roughly thirteen minutes, the number displayed as the "price" - the IEP - is not a transacted price. It is a live projection, recalculated every time an order is added, modified, or withdrawn, of where the book would clear if it closed right now.
Nothing has actually traded until the final match. Derivatives are pricing and hedging against that non-transacted number in real time, and on expiry day, ultimately settling against wherever it lands.
That same gap is the mechanism behind SEBI's own Copthall/Mansi order of August 19 - large orders placed to move the IEP, then withdrawn before the match, timed against options positions that stood to gain from where the closing print landed.
What A Mature Closing Auction Looks Like Elsewhere
NYSE doesn't carry this gap, because its design doesn't create one. Continuous trading in the underlying runs all the way to 4:00pm - the actual close. From 3:50pm, the exchange additionally publishes an order-imbalance feed, but that sits on top of a still-live, still-trading market. There is no window where the underlying goes dark.
The last trade before the bell is a real trade
India's CAS made a different, deliberate choice: halt continuous trading, collect orders, clear once. That design solved a real problem - a single well-timed order distorting a thin VWAP window under the old system. What it gave up was a live reference for the minutes in between. Two Thursdays of data now show what that costs on an ordinary expiry week.
What The Data Points Toward
None of this requires rebuilding CAS. It targets the specific gap.
Shrink the CAS window to five minutes, down from the current thirteen. Less time for an indicative price to be built, tested, and unwound before consequence.
Accept limit orders only during that window - no market orders. A market order carries no price discipline; in a thin auction it is the fastest route to the band edge.
Disallow cancellations once an order is submitted. This does the most work of any single change: it removes the incentive behind the Copthall/Mansi pattern entirely. An order that cannot be withdrawn is a commitment, not a signal.
Stop displaying the live IEP during the collection window. Show order interest and imbalance direction if that's needed, but not a number formatted to look like a price before an actual trade has printed.
Nothing downstream - algos, news feeds, trading terminals - should be treating a projection as a transacted price.
Tighten the band from 3% to 1% until CAS liquidity in the affected names improves. A wider band exists to let an auction discover its true clearing price. On a Rs 126 crore expiry close, it has instead let a thin order walk the print to the mathematical edge of what's permitted.
SEBI has said a consultation paper is coming within the week. The data above is what four weeks of Sensex Thursdays and Nifty Tuesdays have already produced - eight times, in the same shape.
The first month of CAS has shown that India's closing auction can efficiently discover a closing price if volume is good. The next stage is ensuring that derivative settlement remains equally efficient when that price is still being discovered.
About The Author: Shai Coelho is the Founder of Vtrender.
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.