Stop-loss execution mechanics: what happens between trigger and fill

The single most consequential fact about a stop-loss is that, until it fires, it is not in the order book at all. Nothing of yours rests on the bid or the offer. What exists is a condition held by the exchange's matching engine, watching one number: the last traded price. No one can see it, it competes with nothing, and it has no queue position. When a trade prints at or through your trigger, the engine releases a new order into the book, and only from that instant does your instruction compete for a fill against whatever liquidity happens to be sitting there. The trigger governs the timing of your entry into the queue. It has no say whatsoever over the price of your exit.

That distinction sounds pedantic until you notice that it is the source of every stop-loss complaint a retail trader ever files. The stop did not fail to work; it worked exactly as specified, and the specification is narrower than almost everyone assumes. What gets released depends on which stop you placed, and this fork decides your failure mode before the trade is even open. A stop-loss market order (SL-M) releases a plain market order: it takes the best available price, then the next, sweeping the book until it is filled. The exit is effectively certain, but the price is whatever the book offers, which in a fast move can be a long way from your trigger. A stop-loss limit order (SL-L) releases a limit order at a price you nominate: it will not trade worse than that price, which caps your damage, but if the market goes through your limit and does not come back, it sits unfilled, leaving you holding the position with no protection at all. One risks a bad price. The other risks no exit.

Look at the shape of that asymmetry, because it is the whole decision in one picture. SL-M has exactly one terminal state and it is always reached; the only open question is the price. SL-L has three, and two of them leave you worse off than you planned in ways a market order cannot. Neither order type refuses both failures, and no third option exists that does. So the choice is not which stop is better, a question with no answer, but which of the two failures you would rather own, on this instrument, in this market condition. The rest of this page is a tour of the conditions that make that choice bite.

The cleanest proof that a trigger is not a guarantee is the overnight gap, and it is worth being exact about the mechanism rather than waving at it. A stop can only act on prices that actually trade. A chart draws price as a continuous line, but the market underneath it is not continuous: it closes overnight, over weekends, and it faces scheduled shocks that land while it is shut, from results to the union budget to a policy decision. When news arrives in that window, the instrument does not travel down through the intervening prices. It simply opens somewhere else, and every price in between was never offered, never bid, and never traded. A trigger sitting in that skipped region has nothing to fire against. It is not slow, and it is not broken. There is no event for it to detect. The wider context of why prices jump this way is covered in the guide on gap ups and gap downs.

Work it through with numbers, because the arithmetic is the argument. A position closes at ₹100 with a protective trigger at ₹98, so the loss you signed up for is ₹2.00 per share. Call that 1R, the unit every other number on this page is measured in. Overnight the news is bad and the stock opens at ₹94. No trade ever occurred at ₹99, ₹98 or ₹97, so the trigger could not fire inside the gap; it activates at the open, at the first price that exists. A SL-M then fills near ₹94, four rupees below your level, and that overshoot is pure gap loss: a realised ₹6.00 against an intended ₹2.00, three times the plan. A SL-L set at ₹97 fails differently and arguably worse: price never returned to ₹97, so the limit order rests unfilled, and you are still long into a falling market with no working stop. Note which of those is the good outcome.

The figure makes one point that the arithmetic alone hides. Look at where the untraded band sits: it straddles the trigger exactly. That is not a coincidence of the drawing, it is the structure of the problem. A gap that matters is by definition one that jumps your level, so the band of prices that never traded is precisely the band your stop was sitting in. The stop was never touched across twelve sessions of ordinary movement, coming no closer than thirty paise, and then it was bypassed in a single instant without ever being tested. This is why treating a stop as a floor under an overnight position is a category error. The floor was never installed. What you actually have overnight is an exposure, and the only thing that governs how much that exposure can cost you is how much of it you are holding.

Suppose no gap and no lock: an ordinary session, and your trigger fires cleanly on a real print. Even here the trigger has not set your price. The released order arrives at an order book, which is a ladder of resting bids at descending prices, each with a finite quantity attached. A market order does not get the price; it gets the best bid until that bid is exhausted, then the next one down, then the next, until the whole quantity is filled. Your actual fill is the quantity-weighted average of every level it consumed. That average is not a property of your order and not a property of your trigger. It is a property of the ladder that happened to be standing there at that millisecond.

Which means the same stop, on the same trigger, in the same instant, produces wildly different outcomes depending only on the depth beneath it. The figure below walks one identical released order, a sell of 500 shares triggered at ₹98.00, through two different books drawn to the same quantity scale. In the liquid book there are hundreds to over a thousand shares resting at every tick, so the order is absorbed inside two levels and barely dents the ladder. In the thin book each level holds a few dozen, so the order consumes every visible level and keeps going. Neither average fill is asserted: both are computed from the bars you can see, by taking each level in turn until the 500 is complete. The liquid book returns ₹97.99, one paisa from the trigger. The thin book returns ₹95.66, two rupees thirty-four away. Nothing about the trader, the instruction or the timing differed. Only the depth did.

The practical content of that figure is that liquidity is not a background condition, it is the price. Traders tend to treat slippage as bad luck, an unpredictable tax that sometimes lands. It is closer to a reading you could have taken in advance: the depth beneath your position is visible before you ever need it, and a stop on an instrument whose entire bid side is thinner than your own order is not really a stop, it is a hope. It also follows that slippage is not uniform through the day. It clusters in a few identifiable windows, and the pattern is always the same: the wider the spread and the thinner the resting depth at the instant your order is released, the further your fill drifts from your trigger.

The takeaway is not to avoid stops in those windows, which would defeat the purpose of having one. It is to choose the order type for the window you are in. Where depth is thin and slippage risk is high, a well-placed limit stop refuses the worst prints and accepts a small chance of not filling. Where the instrument is deep, a market stop still fills within a tick or two and buys certainty cheaply. And because none of these windows can be timed precisely, the constant that protects you across all of them is not the order type at all. It is size: an exposure small enough that even a violent, badly priced fill is an ordinary loss rather than an account event.

Gaps and thin books both still end in a trade. The third failure does not, and it is the one that most cleanly separates a stop from a guarantee. Indian exchanges apply daily price bands to cash-segment securities, commonly at 2, 5, 10 or 20 percent of the previous close depending on the security, and orders priced outside the band are rejected outright. When heavy one-way pressure drives a stock to its lower circuit, it locks. The book does not become thin; it becomes one-sided. There are willing sellers stacked at the band and effectively no buyers, because no one is permitted to bid below it and no one wants to bid at it. A released sell order is accepted, is perfectly valid, and joins a queue that cannot clear. Your trigger fired correctly, your order is live, and it cannot fill, because a trade needs two sides and one of them is absent. The mechanics of the bands themselves are set out in the guide on circuit limits on NSE.

Notice what the right-hand panel does not contain. In the slippage case there was a bad price available and your order took it; here there is no price at all, and no order type invents one. An SL-M cannot sweep a book with nothing in it. An SL-L cannot be filled at a limit no one will meet. The stop is simply stranded until the lock breaks, which happens only when buyers return or the band is revised, and the fill you eventually get can be far below the level you drew. For securities that have derivatives the mechanism is gentler but not absent: instead of a hard freeze, the operating price range flexes during the day, relaxing in steps in the direction of the move so that trading can continue. That materially reduces outright locks, and it is one of the real differences between segments. It does not change the underlying truth, which is worth stating plainly because it survives every rule change: a stop is a request to trade, and where there is no counterparty, no instruction you can write will manufacture one.

Here is the part most current explanations either get wrong or omit, and it is specific to India. For years the standard advice was that a stop-loss market order was the safest choice, because it guaranteed you an exit. Read against the lifecycle above, you can already see the flaw: a market order does not promise a good exit, it promises any exit, and in a book with nothing in it those are opposite things. In thin option strikes that advice was quietly dangerous for years. In 2021 it stopped being quiet.

A freak trade is a momentary, extreme print in an illiquid contract, typically when a large order meets a near-empty book and sweeps it clean in one motion. During 2021 several option strikes spiked violently for a few seconds before snapping back to fair value. On 20 August 2021, for instance, a Nifty call strike printed a jump of several hundred percent within seconds. The damage was purely mechanical, and it is the exact scenario the depth figure above describes: resting SL-M orders on those strikes triggered on the spike and, being market orders, took whatever absurd price the thin book offered, filling instantly at levels no rational trader would ever accept, before the print corrected. The order type that promised a certain exit delivered one, at a catastrophic price. The exchange response came in two distinct steps, and keeping them straight is the whole point.

On the first measure: from 27 September 2021, under NSE circular NSE/FAOP/49677, stop-loss market orders were discontinued for index and stock options. A stop on an option is now necessarily a limit order, which by construction cannot take an unbounded price. On the second: from 31 October 2022, under NSE circular NSE/FAOP/54242, NSE introduced Limit Price Protection in the equity derivatives segment. LPP defines a reference price as the simple average of trade prices over the last 30 seconds, revised at 30-second intervals through the day, and rejects any incoming limit-priced order that falls outside a dynamic band around it. The band is calibrated to the contract: broadly, options priced above about ₹50 are allowed to move within a percentage band of roughly 40 percent of the reference price, while cheaper options get an absolute rupee allowance instead of a percentage, subject to a minimum floor, so that a near-zero-priced strike is not held to a meaninglessly tiny band.

Read together, the two measures bound a released order from both directions: options can no longer fire an unbounded market order at all, and any limit-priced order that does fire is checked against a live fair-value band and rejected if it is absurd. It is worth stating the boundary precisely, because this is exactly where secondary sources blur it: LPP validates the price of limit-type orders in the derivatives segment, while equity cash-segment securities are governed by their own daily price bands rather than by LPP. Two different mechanisms, two different segments, frequently conflated. If you want one defensive habit that survives all of it, it is the one the exchanges effectively legislated: on a thin instrument, never rely on a market exit, and place a limit wide enough to fill but tight enough to refuse a freak.

By this point a pattern has emerged that is easy to miss when stops are discussed generically. The instruction you write is the same in every segment; the rulebook that receives it is not. A stop on a large-cap share, a stop on an index future and a stop on an option strike are three different objects wearing the same name, and the differences are not stylistic. In the cash segment you may use either order type, and your worst case is a daily price band that can lock you in. In equity derivatives your worst case is bounded by LPP, but in options the choice of order type has been removed from you: SL-M is not accepted, so every option stop is a limit order and inherits the limit order's failure, non-fill. The instrument does not merely change the odds. It changes which failures are available.

The segment where those rules bite hardest is also the one carrying the heaviest retail participation and the worst outcomes. Regulator data is blunt on the point: about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, aggregate net losses exceeding ₹1.8 lakh crore (SEBI, September 2024). Execution mechanics are plainly not the whole of that story, and it would be dishonest to pretend otherwise. But there is a specific contribution worth naming. A trader who sizes an option position on the assumption that a stop caps the loss at a known number has mis-specified the risk before the trade begins, because in options the stop is a limit order and the limit order's failure is that it does not fill. The loss in that scenario is not the one that was planned for and refused; it is the one that was never bounded at all.

All of the above describes rules as they stand on 17 July 2026. Order-type availability, band parameters and protection thresholds are set by the exchanges and revised from time to time, sometimes at short notice, and a page is a poor place to learn that a specification moved. Treat every figure in this section as the shape of the mechanism rather than a live number, and verify the current specification with the exchange or your trading member before you rely on it for anything with money attached.

One more variable changes what a stop is, and it is the one people reach for when they discover their protection quietly vanished at 3:30. Not every stop lives and dies within a single session. A good-till-triggered (GTT) order is a standing trigger held at the broker, with a long validity window, commonly up to a year, that submits your order to the exchange only when the trigger condition is met. Because it persists across sessions, it is the natural way to keep a protective level under a delivery holding you intend to carry for weeks or months, without re-entering the order every morning and without discovering on the wrong morning that you forgot. An intraday day stop, by contrast, exists only for the current session: if it does not fire, it lapses at the close and must be placed again. Neither is better; they answer different questions, and the question they both answer is how long does the trigger stay alive.

Two properties of a GTT are routinely misunderstood, and both are the lifecycle from the first section reasserting itself. First, the GTT trigger is typically valid once: when it fires, it submits the order and its job is finished, so if that submitted order does not fill, the GTT does not keep trying on your behalf. Second, and far more important, triggering a GTT does not guarantee a fill. The GTT only places the order. From that instant it is an ordinary order that still needs a counterparty, and it can be gapped through, rejected outside a price band, stranded at a locked circuit, or left unfilled at its limit exactly like any other stop. A GTT solves persistence across time, which is a real problem and worth solving. It does not solve, and structurally cannot solve, any of the execution risks that begin the moment the order is released, because those risks live downstream of everything a GTT does.

Put the four mechanics side by side and the conclusion assembles itself. Take one trade with one intention: entry at ₹100, trigger at ₹98, an intended loss of ₹2.00 per share. That number is real and you genuinely chose it. Now ask what it survives contact with. In a deep book on an ordinary session it survives almost intact, at ₹2.01. Walked through a thin ladder it becomes ₹4.34, because the depth was not there. Jumped by an overnight gap it becomes ₹6.00, because the price never traded. Stranded at a circuit lock until the lock breaks, it becomes larger still, and how much larger is not yours to decide. Each of those realised prices is carried over from the figures above rather than invented for the conclusion, so the R-multiples underneath them are arithmetic, not assertion.

Read the shape of it. The intended loss is a single line, fixed by your trigger, and every realised outcome sits to the right of it. Nothing sits to the left, because no mechanic on this page can deliver you a better exit than you asked for; the distribution of surprises is one-sided by construction. That is what it means to say a stop bounds your intent, not your loss. The trigger is a statement about what you were willing to accept. The market, at the instant of release, decides what you actually get, and it consults the book, the calendar and the price band, but never your intention. To be clear about what this argument is and is not: it says nothing about how often any of these happens, and nothing here forecasts anything. It says only that the trigger has no vote once the order is released.

Which puts the weight on the one input the market cannot overrule. Look at the bottom strip of that figure: halve the size and every outcome halves with it, in rupees, at once, without changing a single ratio. Size is the only term in the whole system that you set unilaterally and that scales every branch simultaneously. Order type chooses which failure you face and is worth choosing well. Placement decides whether the level means anything at all, which is why it gets its own guide. But neither of them bounds the outcome, and size does, which is why it is the last line of defence rather than the first, and why it belongs upstream in risk management rather than in a stop-loss checkbox. The four failures below are the complete catalogue: each has a cause you can name and a mitigation you can apply, and not one of them is bad luck.

From those mechanics a compact checklist follows, current as of 17 July 2026 and concerned with the execution choice only. 1. On a liquid index or large-cap, prefer SL-M: the book is deep, so a near-certain exit costs a tick or two. 2. On a thin or small-cap stock, prefer a wide SL-L: cap the price against a sweep and accept a small non-fill risk. 3. On options, the choice is already made: SL-M is not accepted, so place a limit wide enough to fill and let LPP reject the truly absurd. 4. For a delivery holding carried overnight, use a GTT for persistence, and remember it neither fills inside a gap nor conjures a counterparty. 5. In the open, expiry and event windows, assume wider slippage and lean toward limit control on anything not deeply liquid. 6. Underneath all five, size the position so that the worst mechanical fill is survivable, because that is the only line that holds when the other five do not.

None of this is an argument against stops, which would be a strange conclusion to draw from a page that spends its length taking them seriously. A stop remains the single most useful instrument a retail trader owns, for a reason that has nothing to do with the guarantee it does not provide: it converts an open-ended decision made under pressure into a closed one made in advance. What the mechanics change is what you are entitled to believe about it. A stop is a commitment about your own behaviour, not a contract with the market, and it should be sized as though the number on the screen is the best case rather than the worst. Deciding the exit before the entry, and then sizing it so that the market's answer is survivable whatever it turns out to be, is exactly the discipline the method we teach is built to install.

• Discontinuation of SL-M orders in options. NSE circular NSE/FAOP/49677 discontinued stop-loss market (SL-M) orders for index and stock options with effect from 27 September 2021, following a series of freak-trade episodes; stop-loss limit orders were recommended in their place. Establishes that an option stop on NSE must now be a limit order.
• Limit Price Protection (LPP). NSE introduced LPP in the equity derivatives segment effective 31 October 2022 (circular NSE/FAOP/54242, dated 28 October 2022): a reference price equal to the simple average of trade prices over the last 30 seconds, revised every 30 seconds, with limit-priced orders outside a dynamic band rejected. Establishes the current bound on where a released limit order can execute. nseindia.com
• Order-type behaviour and the GTT facility. NSE and BSE order-type specifications define market, limit, SL-M and SL-L order types and the last-traded-price trigger behaviour, and the good-till-triggered facility provides a standing trigger with a long validity window whose firing does not guarantee a fill. Establishes the lifecycle used throughout this guide.
• Price bands and the operating range. NSE applies daily price bands to cash-segment securities (commonly 2, 5, 10 or 20 percent), rejecting orders beyond them, while securities in the derivatives segment use a dynamic operating range that flexes intraday. Establishes the circuit-limit mechanics in the failure catalogue.
• Indian retail derivatives outcomes. The Securities and Exchange Board of India study of individual traders in the equity derivatives segment found that about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, aggregate net losses exceeding ₹1.8 lakh crore (SEBI, September 2024). Used here only as the context for why option execution mechanics are worth getting right. sebi.gov.in