Guide · Market structure · Updated 2026-09-23

Trade for trade is a settlement rule, not a verdict, and a moved stock keeps about a fifth of its volume

The move is announced in a circular after the close and is almost always read as an accusation. The exchange's circulars say the opposite in plain words: shortlisting is on account of market surveillance and is not to be construed as an adverse action against the company. What changes is how every trade in the security settles, and that has a price in liquidity that the exchange's own files can measure. It is not a corner of the market: of the 3,270 symbols that traded in the rolling series EQ between 3 January 2022 and 18 September 2026, 1,270, or 38.8 per cent, also spent time in trade for trade, as the guide to every bhavcopy field first measured. The surveillance frameworks behind most of those moves are set out in the ASM and GSM guide; this page measures what the move does.

Gross settlement, one trade at a time

In series EQ the clearing corporation settles each client's net position. Purchases and sales by the same client code in the same security on the same day offset, and only the residual is delivered on T+1. That offset is what makes intraday trading possible: a client who buys a thousand shares at 10:05 and sells them at 14:40 has, for settlement purposes, done nothing except make or lose a price difference.

Trade for trade withdraws the offset. The exchange's Surveillance and Investigation Consolidated Circular, NSE/SURV/74008 of 30 April 2026, states that securities in the segment trade under series BE, settle trade by trade, and that no netting off is allowed. The clearing corporation runs them as a separate settlement type, trade for trade surveillance, beside the normal segment. The same two trades now create two obligations: pay the full value of the purchase and take delivery, and deliver the sold shares from the demat account. In a T+1 settlement the securities pay-in falls due at 10.30 in the morning and the pay-out arrives by 3.30 in the afternoon, by the clearing corporation's own schedule, so shares bought in a session cannot be delivered against a sale made in the same session.

What happens to the resulting shortage is the sharpest difference between the two settlements, and most descriptions skip it. In the normal segment the clearing corporation holds a buy-in auction for short deliveries on T+1 and completes it on T+2. In trade for trade there is no auction: its shortage rules close the position out directly at the higher of the highest price the security traded at from the trade day to T+1 and 20 per cent above the official close on T+1. The price is deliberately punitive. It turns an unbacked sale in a surveillance series into a known loss before the order is placed.

So nothing prohibits intraday trading by name. It becomes a short sale with a penal close-out attached, which is why many Indian brokers refuse intraday orders in these series, and why the exchange's circular NSE/SURV/63563 of 23 August 2024 directed members to show clients a cautionary pop-up, driven by the series in the daily regulatory indicator file, stating that the security is under gross settlement.

The dash in the delivery column is a consequence, not a gap

The same fact explains a defect every user of the bhavcopy meets. The full bhavcopy's delivery columns match, row for row, the exchange's security-wise delivery position file, whose header reads Compulsory Rolling Settlement and whose records carry settlement type N. The quantity it reports is what remains for delivery after netting inside each client code. Trade for trade has no netting, so deliverable quantity equals traded quantity by construction, and the file carries no record for it. In the cache used here all 279,597 BE and BZ rows show a dash in both delivery columns, 247,702 in BE and 31,895 in BZ. Reading the dash as zero turns a security in which every share was delivered into one in which none was.

So a delivery percentage cannot be compared across a move, because it does not exist on one side of it. What can be measured is the delivered share in EQ before a move and after a return, and both say something about who leaves.

Seven published routes into the same two series

Trade for trade is a destination that several separate rules can send a security to. Each is written down, each has its own trigger and its own extra conditions, and each has a way out.

The oldest route, the periodic trade-for-trade review, runs on two clocks. Every fortnight the exchanges identify securities to move in, and three tests must all hold: a price to earnings multiple at or below zero or at or above an upper limit; a fortnightly price rise at least 25 percentage points above the sector or broad index's own move, with a floor of 10 per cent; and a market capitalisation of ₹500 crore or less. Every quarter the review runs in both directions on four alternative tests, one of which needs no price move at all: fewer than 500 non-promoter shareholders. Companies above ₹500 crore with institutional holding above 20 per cent, and companies that paid a dividend or issued bonus shares in two of the last three years, are dropped from the list; securities with derivatives, which carry dynamic bands, are outside the review altogether.

The P/E limit is not a fixed number. It is 30 while the benchmark index's own P/E is between 15 and 20, and moves by the rounded distance the benchmark sits outside that range, never below 25. Computed from the exchange's daily index files, the benchmark's P/E ran from 18.92 to 25.37 across 2022 to 2026, which put the limit anywhere from 30 to 35; on 18 September 2026, at 19.74, it was 30. A page quoting one P/E threshold for this test is quoting one day of it.

The other routes are younger or narrower. ESM, introduced from 5 June 2023 for main board companies below ₹500 crore by circular NSE/SURV/56948 of 2 June 2023, puts every security it shortlists straight into trade for trade at Stage I. It was extended to main board companies below ₹1,000 crore after the joint surveillance meeting of 9 August 2024, to public sector companies from 23 September 2024 and to the SME platform after the meeting of 4 October 2024, and revised with effect from 28 July 2025. GSM reaches trade for trade only at Stage II; long term ASM only at Stage IV. A company in insolvency moves to gross settlement two sessions after disclosing it. The Z category follows SEBI's circular of 22 January 2020 on listing non-compliance, now carried in its listing master circular of 11 November 2024. And the ten-session rule for small listings, set by SEBI circular CIR/MRD/DP/02/2012 of 20 January 2012, is visible in the files: of the 392 symbols whose first row in the cache was in BE, 125 moved to EQ after exactly ten sessions.

Most of today's segment came in through ESM

The familiar picture, a periodic review of small, loss-making or richly valued companies whose price ran, describes the minority of the segment today. The regulatory indicator file that drives the brokers' alert carries one column per surveillance framework. On the file of 18 September 2026, 242 securities stood in BE. Of them, 171 carried an ESM stage, 25 an insolvency or inter-creditor flag, 8 long term ASM Stage IV and 3 a GSM stage of II or higher, which leaves 35 carrying none of the flags that impose trade for trade: the share the periodic review and the unflagged routes account for. Another 36 stood in BZ.

The band files show when it happened. The band file of 1 June 2023 listed 100 securities in BE; the file of 9 June listed 198. The largest single batch of moves in the whole cache, 70 on 7 June 2023, falls exactly two sessions after ESM took effect, the T+2 its circular specifies. The segment peaked at 393 securities in September 2024, after the extension to companies below ₹1,000 crore, and held 242 on 18 September 2026. Of the 1,843 moves in since the indicator file became available on 23 January 2024 whose governing file already shows the new series, 1,062, or 57.6 per cent, carried an ESM stage on their first session in BE; for 18 more the file still showed EQ.

The last column matters as much as the counts. The two large groups lost about the same share of their volume: an ESM move, which also brings 100 per cent margin, kept 21.7 per cent of earlier quantity, and a move carrying no framework flag, which brings no margin change of its own, kept 18.0 per cent. The smaller groups are noisier; the 18 insolvency moves kept the most. What the large groups share is the settlement, and the settlement is what the volume responds to.

What the move does to volume, measured

A move here is a change of series between two consecutive sessions for the same symbol in the full bhavcopy, 1,217 daily files holding 1,164 distinct sessions once holiday copies are removed. Each move is checked against the band file, a second exchange file that lists every security's series for the next session whether it traded or not: the two agree on 6,190 of the 6,191 moves where both can be read. Each move then needs 20 sessions in the old series before it and 20 in the new series after it, listed throughout, which leaves 2,268 of 3,077 pinned moves in and 2,271 of 3,124 moves back. Every security is compared only with its own history.

The detector was also checked by hand against what the exchange published, including the largest batch in each direction and the one move on which the two files disagree.

Traded quantity fell to a median 21.1 per cent of its earlier level, with half of all moves between 12.8 and 36.9 per cent. The number of trades fell further, to 16.3 per cent, so the trades that remained were larger: a median 119 shares a trade before and 167 after. What did not happen matters as much. The securities kept trading, on 99.8 per cent of sessions after a move against 99.9 before. Trade for trade thins a market. It does not close it.

The fall arrives in one session. Measured against its own average over the 20 sessions before, the median security traded 0.77 times that average on its last session in EQ and 0.26 times it on its first in trade for trade. Activity peaks 3 sessions before the switch, at 1.26 times the average, and is already easing in the sessions between a circular and its effective date, but the break comes on the switch session itself. A fading of interest does not look like that; a change in what trading is possible does. On the way back the step runs the other way, from 0.93 to 1.63, and then holds.

Two comparisons separate the settlement change from what would have happened anyway. On the same dates the median security that stayed in EQ throughout traded 0.93 times its earlier quantity, so the market was not falling. The fairer comparison is with securities that had just had the same kind of run. Moves in followed a median 20-session rise of 24.2 per cent, which is the trigger the criteria look for, and securities left in EQ after a rise at least that large over the same windows kept 64.1 per cent of their volume. Rallies fade. They do not fade to a fifth in a session.

The length of the window barely moves the answer, which runs from 0.211 to 0.240 across windows of 5 to 40 sessions. Securities that stayed in BE across the same return dates traded 0.89 times their earlier quantity, so the doubling on the way back belongs to the move, not to the calendar.

The delivered share is a ceiling, and almost nothing reaches it

If trade for trade removed only the volume that netted, a security would keep exactly the share that had been going to delivery. That share is measurable in EQ before the move, a median 49.9 per cent. The volume kept was a median 21.1 per cent. Sorted into fifths by delivered share, the relationship is plain and one-sided: the share kept rises steadily with the delivered share, a rank correlation of 0.67, and falls short of it in 88 per cent of moves. Across all moves the volume kept is a median 0.43 of the delivered share.

The shortfall has identifiable parts. The band narrows with most moves, 968 of them from 20 to 5 per cent and 600 from 10 to 5, which caps how far a price can travel in a session and so how much there is to trade. ESM adds full margin, which ends buying with borrowed money. Overnight positions held for a day or two count as delivery in EQ, and in trade for trade a failed purchase leaves their resale to the penal close-out. And some of the delivery that remains belonged to the same short-horizon traders who leave. Where the band did not change at all, a security still kept only 33.4 per cent of its volume.

The mirror holds on the way back. After a return to EQ the delivered quantity alone was a median 1.36 times all the volume traded in the last 20 sessions under compulsory delivery, so delivery trading itself grows once netting and the wider band come back, alongside the intraday volume that returns with them.

The band moves with the series, and it explains most of the range

All 2,268 measured moves in left the security on a band of 5 per cent or narrower. The median daily range, the high over the low, fell from 5.70 to 4.93 per cent. Split by band, the settlement change alone does little to it: where the band was already 5 per cent the range went from 5.35 to 5.19 per cent, and where it came down from 20 per cent, from 5.61 to 4.73. Volume is the reverse: it falls hard in every group. How bands are set, and why a band is not a circuit, is covered in price bands against circuits and circuit limits in India.

On the way out the routes part company, exactly as written. The periodic review sends a departing security to a 5 per cent band until the next review for upward revision, and 513 of the 514 measured exits since January 2024 that carried no framework flag, the periodic review's population, went to exactly that. ESM exits went back to the band the security held before it was moved in, in 847 of 892 checked cases. The ESM framework as revised from 28 July 2025 states that rule in writing, and the files show it applied before the revision as well: 506 of 534 before it, 341 of 358 after.

A position already held on the effective date

A move is announced ahead of time, and the circular carries everything a holder needs. The ESM list of 9 January 2025, circular NSE/SURV/66052, is typical. Published on a Thursday, it applied 100 per cent margin from Monday 13 January to all open positions as on Friday 10 January and to new positions from the 13th, and moved the named securities from EQ to BE with effect from the 13th. In the files, the main board security it named has its last EQ row on 10 January 2025 and its first BE row on 13 January 2025, and the band file dated 10 January already lists it in BE.

Shares in the demat account are untouched. Ownership does not change, and a holder can sell in any session in which the security trades, settling on T+1 from shares already held. At GSM Stage III and IV that means one session a week, and a suspended company's shares trade in BZ only on the first session of each week, for six months, under SEBI's listing master circular of 11 November 2024 as applied in the exchange's circular NSE/CML/65586.

A purchase that has not settled is the exposed case. Shares bought on the last EQ session arrive on T+1, the first trade-for-trade session. Selling them that day is a sale in trade for trade, due for delivery the next morning. If the original purchase was short delivered, its auction settles on T+2, the same day the resale must be delivered, and a resale that is not delivered is a shortage in a segment that closes out at 20 per cent or more above the close. How the auction side works is set out in short delivery and the auction.

Intraday and leveraged positions have to be closed while they still net. The last EQ session is the last on which a same-day round trip settles as a difference. ESM, long term ASM Stage IV and the insolvency framework also demand 100 per cent margin, and the ESM circulars apply it to positions already open on the session before the effective date.

Buyers pay more at the GSM stages. From GSM Stage II a buyer deposits half the trade value on top of the price, and the whole of it from Stage III, which taxes entering rather than leaving.

The market to exit into is smaller from the first session. The median is a fifth of the earlier quantity, and a quarter of moves kept less than 12.8 per cent. A position sized to EQ volume is now being sold into a market about five times thinner, inside a 5 per cent band.

The way back is scheduled

Every route has an exit written into it, and the files show them being used. Of the 3,107 moves in, 339 were still open on 18 September 2026. Treating open spells as unfinished rather than short, with a product-limit survival estimate, half of all spells had ended by 64 sessions, 14 per cent were still open after 125 sessions and 6 per cent after 250. The exits cluster on the review cycles: 2,862 of the 3,124 pinned returns took effect on a Monday or a Tuesday, 865 of the 918 measured ESM exits on a Monday after the weekly stage review, and 788 of the 2,768 completed spells ended between 90 and 100 calendar days after they began, the signature of a 90-day minimum stay.

The written exits are in the table of routes above, and two details of the periodic review decide their timing. A security moved in at a fortnightly review is not considered for moving out at the immediately following quarterly review, and one moved out at a quarterly review is not considered at the next fortnightly one. ESM, beyond its 90 days, holds a Stage II security for at least a month.

Reversible does not mean rare. Of the 1,150 symbols that moved in, 736 did so more than once, and one did so 13 times in under five years.

What the files cannot settle

Selection is the first limit. A security moves into trade for trade because its price and activity ran, so the window before a move is not ordinary trading. The rally-matched comparison narrows the problem without removing it, because securities that rose as much are not identical to securities that also failed a valuation or concentration test. The one-session step at the switch is the strongest evidence that settlement rather than fading interest drives the fall, and it remains an observation, not an experiment.

The routes of moves before 23 January 2024 cannot be read, because the indicator file is not archived earlier, and a security with no framework flag may have come through the periodic review or through a route the file does not mark. Delivery cannot be observed on the trade-for-trade side at all. Moves are keyed on the symbol as the file keys it, so a rename inside a window drops that move. Every figure is a median across moves, gross of every cost, and none of it says what any security will do next.

What the segment is for

Trade for trade is friction applied by rule to the part of the market where a price can be moved by a few accounts. It removes the two cheapest ways to trade a security, the same-day round trip and borrowed money, and leaves the one that requires owning the shares. The measurement shows what that costs a holder, most of the volume in one session with the band narrowed the same day, and nothing about the company. The habits that follow are mechanical: read the series before the order and the circular before the headline, size a position in a small company for the liquidity it would have after a move, and treat a dash in a delivery column as a fact about settlement. Reading an exchange's files as evidence rather than as tables is a method, and it is the method this curriculum teaches.