Guide · Derivatives market structure · Updated 2026-09-23

Weekly expiries were concentrated, not abolished, and a test that runs across the change is testing two markets

Two easy readings of this change are both wrong. The first is that weekly options were banned: two remain, one per exchange, and they now carry nearly all of the index option premium traded in the country. The second is that one exchange moved to Tuesday while the other stayed on Thursday. Neither stayed anywhere. The BSE weekly expired on Friday until 3 January 2025, on Tuesday from 7 January 2025 and on Thursday from 4 September 2025; the NSE weekly expired on Thursday until 28 August 2025 and on Tuesday from 2 September 2025, each date as the exchanges' own files record it. A page that names an expiry day without a date is describing one of four calendars.

Four calendars in ten months, and the one in force now

The position in force has four parts. The Nifty 50 weekly expires on Tuesday and the Sensex weekly on Thursday. Every other equity derivative, index futures and single stocks included, runs at least a month and expires in the last week of the month on its exchange's day (circular 2025/76, paras 3.1 to 3.3). An exchange needs the regulator's prior approval to change that day (para 3.4). And an expiry that falls on a trading holiday moves to the previous session, which is why the NSE file of 18 September 2026 lists one October weekly on Monday 19 October and the November monthly on Monday 23 November.

Each change also re-dated contracts already trading: NSE its non-benchmark monthlies at the close of 1 January 2025 and its long dated options at the close of 31 July 2025 (NSE/FAOP/65336 and 68747), BSE at the close of 31 December 2024 and 24 June 2025. A re-dated contract sits in the archive under an expiry date it never reached, so an expiry calendar built from old files contains dates that did not happen: 22 on the three largest indices in this panel, removed before anything is counted.

What the consolidation was written to change

The circular of 1 October 2024 names its target. Paragraph 5.5.1 records that exchanges offered short tenure index options expiring on every day of the week, that expiry day trading at low premiums is largely speculative, and that the consultation paper had found "hyperactive trading in index options on expiry day", with holding periods in minutes. Paragraph 5.5.2 answers by allowing each exchange one benchmark index with weekly contracts, and paragraph 6 counts four of the package's six measures as aimed at expiry day.

The lever was availability. Before the cut a contract expiring that afternoon existed on almost every session; the cut removed it on three weekdays in five, leaving the appetite for it untouched. The May 2025 circular states the problem that followed: spacing expiries lowers concentration risk, but too many expiry days could revive the hyperactivity (para 2). Its answer, Tuesday or Thursday for everything, is the pair the March 2025 consultation paper chose as optimal spacing that avoids the first and the last day of the week (para 4.1), after one exchange had proposed Monday (para 3.1).

The nearest weekly contract, counted from the calendar

The plainest measure needs no trading data: for each session, count the sessions until the nearest weekly contract expires, on the actual expiry dates, holidays included.

Before the cut a weekly contract expired the same day on 98.9 per cent of sessions across the two exchanges and on 80.6 per cent on NSE alone. Now the figures are 42.1 and 21.1 per cent, and the mean wait on NSE has gone from 0.19 to 1.90 sessions. For the Nifty 50 nothing changed on this count, 1.90 sessions on average before and 1.90 now; only the weekday moved, and with it where the weekend falls in the final week, which the theta guide works through. Nifty Bank is where a contract changed character: its nearest contract averaged 1.90 sessions to expiry and never more than 4, and now averages 10.10 and reaches 23.

The trading moved to the two contracts that were left

Where the demand went can be read from both exchanges' daily derivatives files, every session from 8 July 2024, when NSE's archive moved to the current file layout, to 18 September 2026: 547 sessions each. Two traps sit in them. NSE reports traded volume in contracts and BSE in units, so the unit is detected from each file. And the value column on an option row is notional, strike plus average premium times quantity, so premium is recovered exactly as that value less strike times quantity.

Nifty Bank carried 39.4 per cent of index option premium before the cut, more than the benchmark; now it carries 5.9. The Nifty 50 went from 30.7 to 61.4 per cent and the Sensex from 8.4 to 31.9, which is why BSE's share of the two exchanges' total rose from 12.4 to 32.0 per cent. The indices left with monthly contracts kept almost nothing: Financial Services, Midcap Select and Bankex fell together from 21.4 per cent to 0.8. Premium per session fell from 66,463 crore to 57,101 in the first six weeks and stands at 76,229 crore. The same day share of premium went from 43.1 to 31.7 per cent, and contracts with eight days or more to run from 7.1 to 9.7.

The nearest contract still takes nine rupees in ten, which hides the shift

Ask the usual question, what share of trading sits in the nearest expiry, and the change vanishes: on NSE the nearest expiry of each index took 90.0 per cent of premium before and 89.0 per cent now, and stock options, monthly throughout and the control here, moved from 88.0 to 88.1. Traders crowd the nearest contract of whatever they trade. What the rule changed is what that contract is, a week away for the two benchmarks and up to a month for every other index, and a statistic keyed on "nearest" cannot see it.

The most quoted expiry figure has a similar blind spot. The regulator's study of August 2026 reports that 70 per cent of index option turnover in FY25 occurred on the expiry day itself and 59 per cent in FY26 (Chart 17, para 6.13), without saying in that chart whether turnover means notional or premium, though its other charts use premium for options. Recomputed from the public files, notional shares reproduce every published point to within 2.2 points. Weighted by premium, the money that changes hands, the same day share is 34.7 and 29.3 per cent. The premium arithmetic is not the cause: the same files reproduce the study's own premium averages for expiry and other days in October 2025 to March 2026 (its Table 10) to within 3.5 per cent.

Neither weighting is wrong. Notional counts exposure written, and an option expiring within hours at a strike far from the index writes a great deal of it for almost no premium; premium counts what traders paid. Read as a statement about money, "70 per cent on expiry day" overstates the concentration about twofold.

A weekly lives on the steep end of the curve, a monthly spends three weeks on the flat

Time value at the money grows with the square root of the time left, so it drains fastest at the end. The files show the curve as traded: for every contract and session, the closing call and put at the strike nearest the index, summed into the at the money straddle and expressed as a percentage of the index.

Every weekly series decayed close to the square root of sessions, whatever its weekday. With one session left a Nifty 50 weekly held 0.468 of its five session value when it expired on Thursday and 0.487 since the move to Tuesday, against 0.447 for the square root: the final session removes about half of a weekly's premium, and a weekly seller's whole trade lives on the steepest stretch. The weekend barely shows. Across the Friday to Monday step of a Tuesday expiry the straddle lost 0.188 of its five session value, where the square root of sessions, counting the weekend as nothing, predicts 0.185.

A monthly has the same shape stretched over four weeks. Sold twenty sessions out, a Nifty Bank monthly still held 0.764 of that value with ten sessions left and 0.528 with five, so three weeks of holding bought 0.472 of the premium and the final week the rest, 0.266 of it in the last session.

What fewer weekly expiries means for a seller

Steep weeks a year. The Nifty Bank weekly series offered its final five sessions 53 times in the twelve months before the cut; the index now offers that stretch 12 times a year. The steep part of the curve still exists on every index, but on all but two it arrives once a month.

Premium passing through. At today's volatility a five session Nifty Bank straddle cost 1.70 per cent of the index and a twenty session one 3.37. At the 53 expiries of the weekly year the first sums to 90.2 per cent of the index, and at the 12 of the monthly year the second to 40.4, a ratio of 2.23, near the square root of 53 over 12, 2.10. That premium is not income: at a fair price it equals the expected cost of the moves the seller absorbs. What a seller keeps, if anything, is the gap between the volatility priced and the volatility that follows, which the variance risk premium article measures for the thirty day index, positive on most days and very large and negative on a few. Under a random walk with a constant gap, weeklies collect about twice as much of it in a year for the same yearly spread of outcomes, because that spread does not depend on how the year is sliced. That is arithmetic about why sellers favoured weeklies, not a measurement of any gap, and its normal distribution understates the jumps that end such books.

The price of expiry day. Since 20 November 2024 a short option carries an extra extreme loss margin of 2 per cent of notional on the day it expires (circular 2024/132, para 5.6), on top of the 2 per cent base for index options within 10 per cent of the index (master circular of 30 December 2024, chapter 5, para 2.2.3), so the charge doubles on expiry morning with no change in the position. On 18 September 2026 a Nifty 50 lot of 65 units at 23,346.40 had a notional of 15,17,516 rupees, an add-on of 30,350 rupees; a Sensex lot of 20 units at 74,294.96, one of 29,718. From 1 February 2025 a weekly short hedged with a later expiry is margined as two positions on expiry day (para 5.2), which the spread margin guide computes. And with two weeklies left, the other side of every weekly sale now sits in the same two afternoons, where pin risk lives.

A backtest across the change is two backtests averaged by accident

A result pooled over market states is a weighted average of the states, weighted by how often each occurred in the sample; the regime article sets out the identity. The expiry calendar is such a state, with weights set by circular. Reading the actual expiry dates out of the exchange's files from January 2022 shows which markets a four year backtest contains.

A Nifty Bank weekly strategy tested over those years draws 87.2 per cent of its 172 expiries from weekly series that no longer exist, under three different weekday patterns, and the Nifty 50, which kept its weekly throughout, draws 77.6 per cent of its 246 from the Thursday calendar. The pooled figure describes the old market, weighted by how many years of it happened to be downloaded. The files also correct the circulars: a move of Nifty Bank to Friday from July 2023 was announced (NSE/FAOP/56967) and never took effect, and the files show Thursday expiries through 31 August 2023.

Two implementation choices hide this. A backtest trading "the nearest expiry" switched instruments silently after 13 November 2024, its Nifty Bank contract going from 1.90 sessions to expiry on average to 10.10, with a different decay, premium and weekend count. One trading "the contract with five sessions left" survived in form but lost three quarters of its trades, and those it kept open on a different weekday. Neither throws an error.

The fix is to treat each calendar as its own market. Split the test at every date below that touches the trade, run each segment with its own lot size, margin and tax, report the segments and never the pooled figure, and judge the present on the current segment alone. It is short, 55 Tuesday cycles on the Nifty 50 to 18 September 2026 and far fewer on any other index, so its interval is wide, which the sample size guide quantifies. A wide, honest interval on the market that exists is worth more than a narrow one on a market that does not.

What these measurements cannot settle

Timing is not cause. The weekly cut came with a larger contract, a higher tax, upfront premium and new limits within months; the regulator's study calls its own before and after comparisons association, not causation, and so are these. The old calendar is measured briefly: the current file layout begins on 8 July 2024. The straddle is a closing print at the nearest strike, which describes the curve's shape, not a tradable price. The settlement basis moved again: since 3 August 2026 the closing auction sets the index close options settle against, and the regulator's paper of 12 September 2026 proposes blending it with the last thirty minutes of trading or returning to the thirty minute average for at least a year. Nothing here is a return: no figure measures what a strategy earned.

Read this way the calendar is a map of which market a record was made in, and the first question for any result is whether its contract still exists. The expiry day guide covers the last session, the settlement guide the close, and the participation article the rest of the package.