What are circuit limits on the NSE and BSE?

Start with what the exchange is actually deciding, because every other fact on this page follows from it. A circuit limit is not a safety feature bolted onto price discovery. It is a deliberate suspension of price discovery. The exchange has looked at a class of situations, disorderly falls, engineered ramps, fat-fingered orders, cascading margin calls, and concluded that in those situations the price the market would produce is worse than no price at all. So it stops producing one. That is the trade being made on your behalf, and it is a real trade with a real cost, not a free protection.

The case for it is genuinely strong. A market in free fall stops being an auction and becomes a stampede, where prices reflect who is being forced to sell rather than what anything is worth. Forced sellers beget margin calls, margin calls beget forced sellers, and a mechanical loop can take a price a long way from any level a calm participant would accept. A pause breaks the loop. It gives the exchange time to look for a rogue algorithm, gives clearing members time to establish who is actually solvent, and gives everyone else time to notice that the news does not justify the tape. Every major exchange in the world has some version of this because the alternative has been tried and it was worse.

But be precise about what you are buying. The pause does not reduce the selling pressure; it defers it. It does not make anyone's position smaller; it makes it untradeable. And for the person holding the position, the honest consequence is stark and rarely stated plainly. A locked circuit is not a bad price. It is no price. Your order is not filled badly, it is not filled at all. Your position cannot be exited at any level, good or terrible, because there is no level on offer. Liquidity here is not merely thin; below the band it is legally absent, refused by the exchange before it can reach the book. Everything else in this guide is an elaboration of that one sentence.

The first mechanism is the one most people mean by circuit. Every cash-segment security is assigned a daily price band, a percentage of the previous close beyond which it may not trade that session, applied either way, so there is a ceiling and a floor and they are symmetrical. The tiers are 2, 5, 10 and 20 percent. Orders priced outside the band are not filled badly or filled late; they are rejected. The band is recomputed every session against the last close, which sounds like a footnote and is in fact the most consequential detail in this entire guide, for reasons the last two sections will make painfully concrete.

Which stock gets which band is not a rule you can derive from first principles. The exchange assigns bands security by security, and the assignment reflects liquidity and surveillance history rather than size alone, which is why two companies of comparable market capitalisation can carry different bands. The number is published, so the honest instruction is to look it up rather than infer it. What you can derive is the shape of the exception, and the exception is where the mistakes live.

Securities on which derivative products are available carry no daily price band at all. Not a wide one: none. They are governed instead by an operating range, discussed further down, and by the market-wide breaker. Then comes the trap. A stock that is merely a constituent of an index, but has no derivatives of its own, is still subject to an ordinary price band. So the phrase index stock tells you nothing about whether a stock can lock. Having its own listed derivatives is what removes the band, and being in an index is not the same thing. This distinction accounts for a large share of the surprised traders on any given day.

Two smaller facts round the picture out, because they show the band is a general instrument and not a special measure. Securities in the Limited Physical Market carry whatever band they carry in the normal market, and the auction market runs a 20 percent band of its own. Bands are not an emergency lever the exchange pulls in a crisis. They are the ordinary furniture of the cash segment, sitting there every single day, doing nothing at all until the one day they do everything.

Now the mechanic that matters, because it is the one almost every explanation skips. What does locked at the lower circuit actually mean at the level of the order book? It does not mean the book is thin. Thin is a book with a bad price in it. A lock is a book with one side missing entirely, and the difference between those two states is the difference between a painful fill and no fill at all.

Follow the two halves. Sellers who want out place orders at the lowest price the exchange will accept, which is the band itself, so the queue is largest exactly at the floor. That queue is not a wall of stubbornness; it is a wall of people who would take less if they were allowed to, and are not. Meanwhile the buy side is empty at every price, and it is empty for two entirely different reasons stacked on top of each other. Below the band, no order can exist, because the exchange rejects it before it reaches the book. At the band itself, any genuine bid would match the queue instantly, so the fact that none arrives is the market telling you plainly that it does not think this price is right. One half of the emptiness is a legal fact and the other half is an economic one, and neither is something you can trade around.

This is why liquidity is the correct lens for circuits and price is not. In ordinary bad conditions liquidity thins and you pay for it in slippage, which is unpleasant and survivable and, crucially, still a transaction. At a lock there is nothing to pay. Your order is valid, correctly priced, live in the system and completely inert. The stop-loss execution guide works through the other ways a released order can fail, gaps and thin books and rejected limits, but they all share one feature that a circuit does not: in every one of them, a trade eventually happens. A lock is the only failure where the answer to what price did I get is that there was no price to get.

The second mechanism has almost nothing in common with the first except a nickname. The index-based market-wide circuit breaker does not cap a stock. It brings about a coordinated trading halt in all equity and equity derivative markets nationwide, and each word in that phrase is load-bearing. Coordinated means both exchanges stop together, so the market cannot migrate to the other venue. Nationwide means everything, including the stock that was behaving perfectly well. And equity derivative markets means the futures and options market shuts in the same breath, which has a consequence for hedging that the closing section will come back to, because it is the least appreciated fact on this page.

It applies at three stages of index movement, either way, at 10, 15 and 20 percent. Note the symmetry: the breaker fires on the way up too, which surprises people who think of it purely as a panic brake. It is triggered by a move in either the BSE Sensex or the Nifty 50, whichever is breached earlier, so a trader watching only one index is watching only half the trigger. The system has been in force since July 2001 under a SEBI circular of June 2001, and the current shape comes from a SEBI circular of September 2013 that partially modified it, given effect by the exchange in October 2013. It is a settled, published, mature mechanism, not a discretionary decision made in the moment by somebody in a control room.

The part most sources handle badly is the duration, because it is not a number. The halt length is a function of two variables: which level was breached, and what time of day it was breached. Early in the session the market can afford a long pause and still reopen; late in the session a breach either costs almost nothing or ends the day. This matrix is the authority core of the whole topic, so here it is in full, including the column that is nearly always dropped.

Here is something that ought to be better known: the market-wide breaker is not a surprise. The exchange computes the circuit breaker limits for the 10, 15 and 20 percent levels on a daily basis from the previous day's closing index level, rounds them to the nearest tick size, and publishes them. The trigger for today is a specific number of index points, and it was knowable before the opening bell. To make that concrete rather than abstract, take one real published example. On 16 July 2026 the Nifty 50 closed at 24,072.75. The trigger distances the exchange published for the following session were 2,407.25 points at the 10 percent level, 3,610.90 at 15 percent and 4,814.55 at 20 percent. Those are not estimates and there is no judgement in them. They are the previous close multiplied by a fraction and rounded, recomputed fresh every single day.

So the mystique is misplaced. The breaker is arithmetic, published in advance, and you can read today's numbers this morning. What is genuinely hard to internalise is not the level but the clock, because the halt is a step function and steps have edges. This is where the matrix stops being a table and starts being a cliff you can fall off.

Sit with the two cliffs for a moment, because they are the practical content of the matrix. A ten percent fall that lands at 12:59 pm buys a 45 minute halt; the same fall two minutes later buys 15. A fifteen percent fall at 1:59 pm halts the market for 45 minutes and then lets everybody trade again; at 2:01 pm the day is simply over, positions frozen where they stand until tomorrow. Nothing about the market changed in those two minutes. Nothing about your position changed. The only thing that changed was the clock, and the clock decided whether you get another chance to act today. This is the honest reason to distrust any source that quotes one halt length: it is quoting one row of a seven-row table and presenting it as the rule.

The third piece of the picture is the one where precision pays, because the popular account of it is subtly wrong in a way that matters. The usual telling is that derivatives get a special dynamic circuit that flexes instead of freezing. The more accurate statement, and the one the exchange actually makes, is blunter: there are no day minimum or maximum price ranges applicable in the derivatives segment. Not a flexible band. No band. The daily circuit simply is not part of that rulebook.

What exists instead is an operating range, and its purpose is different in kind. It is set at 10 percent of the base price for index futures and for futures on individual securities. For index and stock options a flat percentage would be meaningless, since a 10 percent move in the underlying is emphatically not a 10 percent move in the option, so the exchange computes a contract-specific price range based on the option's delta and updates it daily. And the stated reason for all of it is not to contain panic. It is to prevent erroneous order entry. An order priced beyond the operating range does not vanish; it reaches the exchange as a price freeze, which is a check to be resolved rather than a wall that holds.

Hold those two purposes apart and a great deal of confusion dissolves at once. A price band exists to stop the market going somewhere; it is a judgement about prices. An operating range exists to stop a keyboard sending something absurd; it is a judgement about orders. That is why the derivatives market keeps trading through days when small stocks are frozen solid, and it is not because derivatives have a gentler circuit. It is because they have no circuit of that kind at all, and the mechanism they do have was never designed to stop a genuine move. The range relaxes in the direction of sustained genuine pressure rather than holding as a wall, which is exactly what you would expect of a device whose job is refusing typos rather than refusing prices.

Above both mechanisms sits a surveillance layer, and it is genuinely a layer rather than a third kind of circuit. It does not introduce a new wall; it moves the walls that already exist, and it moves them on specific named stocks with very short notice. Ignoring it is how a trader ends up holding something whose rules are not the rules they looked up.

The Additional Surveillance Measure is applied to securities with surveillance concerns, and the shortlisting is not discretionary. It runs on objective parameters jointly decided by SEBI and the exchanges, and the list is worth knowing because it tells you exactly what the surveillance system is looking at: high-low variation, close-to-close price variation, client concentration, volume variation, delivery percentage, market capitalisation, the number of unique PANs and the price-to-earnings multiple. Read that list again as a description of a stock rather than a checklist. It is a portrait of a name being moved by a small number of accounts, on volume that is not turning into delivery, at a price with no earnings behind it. The stage-wise measures are published by circular and revised over time, so the operative list is the one on the exchange site today, not any summary of it, including this one.

The Graded Surveillance Measure is aimed at securities whose price is not commensurate with their financial health and fundamentals, judged on earnings, book value, fixed assets, net worth, the price-to-earnings multiple and market capitalisation, and reviewed on a quarterly basis. Unlike ASM it has a published ladder, and the ladder is worth reading closely, because it ends somewhere genuinely remarkable.

Look at what that ladder does. Stage I takes your leverage and halves your corridor. Stage II removes intraday netting, so every share must be delivered, and locks up half the trade value in cash. Stage III leaves you a single trading day a week, which is a stop-loss that can only be attempted on Mondays, and demands the entire trade value as a deposit. Then Stage IV adds the sentence that ought to end every argument about whether a circuit is a technicality: with no upward movement. The upper band is set to zero. The exchange has decided this security may fall and may not rise. It is a one-way circuit, and it is the tightest one in the market. This ladder is the machinery behind a great deal of what happens to a penny stock, and it explains why such a name can be nearly impossible to exit at any price you would recognise as fair.

Return now to the fact from the price-band section that looked like a footnote: the band is a percentage of the previous close. Put that together with what a lock does, and something unpleasant follows automatically. If a stock locks at its lower circuit, the locked price is the close. Tomorrow's band is computed from it. The floor is not a floor at all. It moves down by the same percentage every time you fail to escape through it.

Trace the arithmetic, because it is worse than intuition suggests and it involves no unusual event whatsoever. The prices below are illustrative and the steps between them are exact. A stock with a 5 percent band locks at 95 on day one. Day two's band is 5 percent of 95, so it opens at 90.25 and locks again. Then 85.74, then 81.45, then 77.38. Five ordinary five percent sessions, each of them individually unremarkable, compound to a fall of 22.6 percent. And look at the middle of that chart: three sessions that are flat lines, because a locked session has no price action to draw. The flatness is not an artistic decision. There was nothing to plot. Those are three complete trading days during which the honest answer to what could you have done is nothing at all.

Now put the stop back in. It was placed at 95.00, five percent below entry, which is a perfectly sensible level chosen for perfectly sensible reasons. It triggered at 13:40 on day one, at the exact moment the book locked, which is not a coincidence: the very move that reaches your stop is the move that empties the other side of the book. A stop placed within the band's reach can be triggered by precisely the event that makes it unfillable. The order was released, was valid, was correctly priced, and waited four sessions. It filled on day five at 77.60. The intended loss was 5.00 a share. The realised loss was 22.40 a share, four and a half times the risk that was actually authorised. Nothing was mis-executed. The stop did exactly what a stop can do, which is less than most people think it does.

So take the position seriously as an engineering problem. You are long, the stock is locked at its lower circuit or the market is halted, and you want out. Enumerate every route honestly and check each one against the mechanism rather than against your hope.

The hedge deserves a second look, because it is the sophisticated answer and it fails in a genuinely instructive way. Suppose you decide to buy a protective put instead of selling the stock. If your problem is an individual stock locked at its band, then that stock is band-eligible, and band-eligible means it has no derivatives listed on it, so the put you want has never existed. If your problem is instead a market-wide halt, the derivative market is shut in the same coordinated action that shut the cash market, so the venue is closed. The hedge fails in both cases, for opposite reasons, and the two reasons between them cover every case in which you would want it. That is not bad luck. It is the structure.

Meanwhile the clock does not stop and neither does the arithmetic. This is the part traders find hardest to accept, so it is worth saying without any softening. The halt pauses the price, not the position. Your holding is still marked at the circuit price, the loss is entirely real, and any margin obligation that follows from that mark is entirely real too. You can be required to fund a position you are not permitted to close. The exchange has suspended your ability to act; it has suspended nothing about your exposure. Those two things are not symmetrical, and the asymmetry runs against you.

Which leaves exactly one control standing, and it is the least glamorous one available. Not the stop, which is a request the market may decline. Not the hedge, which needs a venue that may be shut. Not your judgement in the moment, because a locked circuit does not accept judgement as an input. Only the size, and only because it was settled while the market was still open, at a moment when you still had the standing to decide anything at all. This is what makes circuits the cleanest proof of the whole argument for risk management: a stop-loss bounds your intent, not your loss, and the gap between those two is exactly as wide as the market decides to make it on the day it stops accepting your orders.

So the practical instruction is not to fear circuits, and certainly not to avoid every stock that carries a band. It is to size every position as though the exit may simply not be available, because on the days it matters most, it will not be. A position small enough that a five-session staircase is survivable is a position that respects how the plumbing actually works rather than how it usually behaves. Building that assumption into the size before the trade, instead of discovering it during one, is the whole of the discipline that the method we teach is built around. The circuit is not your enemy. Assuming it will never fire is.

• The index-based market-wide circuit breaker, and the exact halt matrix. NSE documents the three stages at 10, 15 and 20 percent either way, the coordinated nationwide halt across all equity and equity derivative markets, the trigger by the BSE Sensex or the Nifty 50 whichever is breached earlier, the halt duration and the 15 minute pre-open call auction by trigger time, and the daily computation of the trigger limits from the previous close rounded to the nearest tick size. Establishes the matrix and the arithmetic. Read as of 17 July 2026. nseindia.com
• Cash-segment price bands, and the exemption that catches people. NSE publishes the daily price bands of 2, 5 and 10 percent either way, 20 percent on all remaining scrips including debentures and preference shares, 20 percent on the auction market, the Limited Physical Market treatment, the absence of any price band on scrips with derivative products available together with the 10 percent operating range applied in their place, and the rule that scrips without derivatives of their own remain banded even when they are part of index derivatives. Establishes the band tiers and the index-constituent trap. Read as of 17 July 2026. nseindia.com
• The derivatives segment has no daily price range. NSE states that there are no day minimum or maximum price ranges applicable in the derivatives segment, and sets the operating ranges kept in order to prevent erroneous order entry: 10 percent of the base price for index futures and for futures on individual securities, and a contract-specific price range computed from the delta value and updated daily for index and stock options, with orders beyond the range reaching the exchange as a price freeze. Establishes the operating-range mechanism and its purpose. Read as of 17 July 2026. nseindia.com
• The GSM ladder, stage by stage. NSE publishes the Graded Surveillance Measure framework for securities whose price is not commensurate with financial health and fundamentals, the quarterly review, the shortlisting criteria, and the four-stage table of surveillance actions running from a 100 percent margin with a 5 percent band, through trade-for-trade with an Additional Surveillance Deposit of 50 and then 100 percent of trade value in cash, to trading permitted once a week and finally the same with no upward movement. Establishes the GSM table. Read as of 17 July 2026. nseindia.com
• The ASM shortlisting parameters. NSE sets out that the Additional Surveillance Measure applies to securities with surveillance concerns based on objective parameters, shortlisted on a criterion jointly decided by SEBI and the exchanges covering high-low variation, client concentration, close-to-close price variation, market capitalisation, volume variation, delivery percentage, the number of unique PANs and the price-to-earnings multiple. Establishes what surveillance is actually measuring. Read as of 17 July 2026. nseindia.com
• The regulatory basis. The market-wide circuit breaker has been in force since 2 July 2001 under SEBI Circular SMDRPD/Policy/Cir-37/2001 dated 28 June 2001, partially modified by SEBI Circular CIR/MRD/DP/25/2013 dated 3 September 2013 and given effect by the exchange in October 2013. The circular references are as cited by NSE on the circuit breakers page above.