Guide · Market mechanics · Updated 2026-09-19

A good till triggered order does not rest in the exchange order book

An instruction that survives the close is doing something the exchange order book cannot do. The book takes an order for the session. At the close, whatever has not matched is purged. Nothing rests there for a month waiting for a price to arrive.

So when a facility promises that your instruction stays alive until a price is touched, the first question is not what it does. It is where it is. The answer is that it is at your broker, in your broker's database, watched by your broker's process. The exchange has never seen it and does not know it exists. Every practical property of these orders, good and bad, is a consequence of that.

This is not a criticism of the feature. It is a genuinely useful thing to have if you cannot sit in front of a screen at eleven in the morning. But a convenience built on top of the order book behaves differently from the order book, and the differences are the part that generic explanations leave out.

The book has no slot for a month-long instruction

Order validity in the Indian equity segment comes in two forms. A day order lives for the session in which it was entered, and any unexecuted quantity is cancelled automatically at the close. An immediate-or-cancel order does not live at all: it executes against whatever is available at the instant it arrives, and the remainder is cancelled on the spot.

That is the whole menu. There is no validity in the regular book that spans a week, let alone a year. Board-approved policies published by trading members say so directly, in the course of explaining why the facility exists: orders are valid only for the day because the exchanges permit only day or immediate-or-cancel validity in the equity segment, so a client whose price is not reached has to enter the order again the next morning.

That daily re-entry is the problem the feature solves, and it is a real one. A person who decides to buy at a level twelve percent below the market and wants to be there when it arrives faces either a hundred consecutive mornings of order entry or a hundred chances to talk themselves out of it.

There are two engineering answers to that problem and they are not the same thing, even where the marketing name is similar. One is re-submission: the broker places a fresh day order every morning on your behalf. In that design the order really is in the book, every session, and it really does earn a queue position, but it is a different order each day. The other is trigger-watching: the broker holds the instruction and sends nothing at all until the condition is met. In that design there is no order in the book on any day until the day it fires.

Naming varies between brokers. Some offer one design, some the other, some both under different labels, some neither. The published policy is the document that tells you which one you have, and that is not a formality, because the two designs fail in different ways.

What the broker is doing while you are not watching

Strip the interface away and the mechanism is small. The broker's system compares the last traded price of the instrument against the trigger you stored. When the comparison turns true, it constructs an order using the price, quantity and order type you specified when you set the instruction, passes it through the broker's own risk management system, and transmits it to the exchange.

Two details in that sentence do more work than they appear to. The first is the reference price. The usual reference is the last traded price, which is a record of a completed trade, not a quote. A single print at a stray price can satisfy your condition. Equally, the best bid or offer can travel through your level without a trade printing there, in which case the condition is not satisfied even though the market visibly passed you.

The second is the phrase "passes through the broker's own risk management system". That system is entitled to reject the order. It checks funds, margin, holdings, product eligibility and any limit the broker applies to your account. None of that happened when you set the instruction. All of it happens at the instant it fires.

No queue position until it fires

Matching in the Indian equity market runs on price and then time. Among orders at the same price, the one that arrived first trades first. Priority is something an order earns by existing, and a pending instruction does not exist where the matching happens.

The consequence shows up in exactly the situations where you cared about the level. When a price band is reached and a large one-sided queue forms, only a fraction of that queue trades, and position in it decides who is in the fraction. An instruction that fires at the moment the band is touched arrives behind everyone who was already there. The same is true at any level where interest clusters, because the cluster forms before the price arrives, not at the moment it does. Our guide to circuit limits and how trading halts work covers what happens to the queue once a band is hit.

None of this makes the instruction useless. It makes it a different instrument from a resting limit order, and it means that the phrase "my order is in at that price" is not true of a pending trigger, however the app displays it.

The gap between the condition and the order

Between the moment the condition becomes true and the moment an order is live at the exchange, several things have to happen in sequence. The comparison has to run. In most implementations it is not a continuous watch but a check driven by market data updates or by a scheduled sweep, and either way there is an interval. The order has to be constructed. It has to clear risk validation. It has to be transmitted, accepted and acknowledged.

Each of those stages is fast in a calm market and slower in a busy one, and that is the structural problem. The burst of market data that carries the price through your level is the same burst that loads the broker's market data handler, the comparison process, the risk engine and the order gateway. The conditions under which a trigger is most likely to fire are the conditions under which the chain behind it is under most strain.

SEBI's own framework for broker systems is built on that observation. It requires brokers to plan installed capacity against the highest peak load observed during a calendar quarter, at not less than one and a half times that observed peak, and to raise alerts when utilisation goes beyond seventy percent of installed capacity. The regulator planned around peak load because peak load is what breaks things.

Then there is the case where there is no latency problem at all and the outcome is still nothing like the trigger. Suppose an instruction is set to sell if the last traded price falls to 1,1 while the previous close was 252, and the instrument opens the next session at 236. The condition is satisfied at the first print. The order is generated at the first print. If it is a market order, it fills near the open, which is 4.8 percent below the trigger. The trigger did what it was asked to do and the number it named never traded. These figures are illustrative.

The framework that puts numbers on the words "the systems were up"

Because a pending instruction depends on the broker's systems rather than the exchange's, the reliability of those systems stops being a background concern and becomes part of the instrument. India has a specific framework for this, and it is worth knowing because it converts an argument into a record.

SEBI circular SEBI/HO/MIRSD/TPD-1/P/CIR/2022/160 dated 25 November 2022, in force from 1 April 2023, sets out a framework to address technical glitches in stock brokers' electronic trading systems. It defines a technical glitch as a malfunction in the broker's systems, including hardware, software, networks, processes or any product or service provided electronically, whether the broker's own or outsourced, which leads to stoppage, slowing down or variance in normal functions for a contiguous period of five minutes or more.

The last row is the one a client can actually use. The framework requires the exchanges to publish the instances of technical glitches occurring in brokers' trading systems along with the root cause analyses. That is a fact you can check before deciding how much of your risk management to leave sitting on one member's software over a three month window.

State this carefully, because it is easy to turn into an accusation and it is not one. Outages happen to everyone who runs systems at scale, which is precisely why the regulator chose measurement and publication rather than prohibition. The relevance here is narrow: a resting exchange order does not care whether your broker's servers are healthy, and a pending instruction does.

What the order becomes, and why that choice carries everything

A trigger is a condition. It is not an order type, and it does not confer any protection of its own. When it fires, whatever it generates is an ordinary order of a type you chose, and it carries every property and every risk of that type. This is the point at which people substitute the comfort of having set something for the discipline of knowing what they set.

Whatever it becomes, it becomes a day order. Member policies are explicit: once triggered, the order is placed as a normal order with day validity and appears in the normal order book, and irrespective of whether it executes or remains pending, it is not carried forward to the following day.

That produces the consequence most people have not thought through. An instruction that triggers and does not execute leaves you with nothing at all. The instruction is gone, because it was consumed by firing. The order is gone, because it was purged at the close. Published policies say so directly: instructions triggered but not executed for any reason have to be placed again if you still want them. The instruction is spent by firing, not by filling.

Conditions that are met and still produce nothing

Separately from timing, there is a set of states in which the condition is satisfied and no fill results, for reasons that have nothing to do with your broker's speed.

A price band. If the instrument is at its band, there is no trading against it in that direction. The order joins a queue that may never clear.

A halt or a suspension. Trading stops, the instruction may fire on the last print before the stop, and the resulting order sits in a book that is not matching.

A change in the instrument's status. If the security moves into a surveillance framework that restricts the product or the order type you chose, the order can be rejected on arrival even though the price is exactly where you wanted it. The surveillance frameworks guide sets out what those restrictions look like in practice.

A call auction. In the pre-open call auction the price is determined by an equilibrium calculation rather than continuous matching, and order behaviour there follows the auction rules, not the continuous ones.

A broker-side cap. Several published policies limit how many instructions a client may have pending at once, and at least one sets that limit at fifty. A new instruction beyond the cap simply is not accepted.

Product eligibility. Some members offer the facility only in the cash segment and only on delivery products, not on intraday products or derivatives. This is a member parameter, and it changes.

Validity is a broker parameter, not a market fact

There is no market-wide answer to how long one of these instructions lives, and this is where a reader is most likely to be carrying a wrong number. Across published member policies the stated validity ranges from two months to a year, expressed against different reference points.

The label on the feature tells you nothing about which of these you have. Good till cancelled, good till triggered and good till date are names, and the number behind the name is set by the member. This is exactly why the exchanges required the validity to be stated in a published policy rather than left to the interface.

The failure mode is quiet. An instruction set in the belief that it lives for a year, on a platform whose policy says sixty days, lapses without anything happening. You find out on the day the price finally arrives and nothing does.

The corporate action problem, and the rule that now governs it

This is the failure mode generic pages omit entirely, and it is the one most likely to catch a careful person, because it punishes exactly the behaviour the feature encourages: setting a level and leaving it alone.

A bonus issue, a stock split, a consolidation, a demerger or a scheme of arrangement changes the quoted price of a security without changing what a holder owns. The exchange applies an adjustment factor so that the price series remains continuous in economic terms. Your trigger is a number you typed into a form. It knows nothing about the factor.

Take an illustrative case. A sell instruction is set at 1,180 while the market is at 1,090, which is 8.3 percent above the market and a plausible target. A one-for-one bonus is then declared. On the ex-date the quoted price adjusts to around 545 and the holding doubles. The economics are unchanged. The instruction is not: the same 1,180 is now 116.5 percent above the market and will not be reached in any reasonable horizon.

The mirror case is worse. A buy instruction set comfortably below the market can be crossed by the adjusted price on the very first print after the ex-date, and fire into a purchase at a price that means something completely different from what it meant when it was set. Our note on how a demerger splits a cost base covers the same discontinuity from the tax side.

India now has a specific rule for this, and it is recent enough that most live explanations of these orders predate it. NSE circular NSE/INSP/62528 dated 21 June 2024, and BSE Notice No. 20240622-2 dated 22 June 2024, on the policy for handling good till cancelled orders offered by members to clients, required every trading member offering good till cancelled, good till triggered or similar orders to frame a clear and transparent policy. The exchanges specified what the policy must contain.

A follow-up circular, NSE/INSP/63789 dated 6 September 2024, went further on where the policy has to live: members were advised to make it part of the account opening form and kit, to display it on their website, and to communicate it to existing clients. That is why every member offering the facility now publishes a document with a title along the lines of "policy on handling of good till cancelled and good till triggered orders". It is the single most useful page about your own instructions, and almost nobody reads it.

Get the boundary right. This is a disclosure and process obligation on the member. It does not transfer the outcome to the member, and published policies are explicit that the decision to cancel, reset or retain is the member's, made with regard to exchange directives, the nature of the action and system capability, and that the client remains responsible for monitoring pending instructions and acting on the notice. What the rule guarantees you is notice, and notice only helps someone who reads it.

In practice the three treatments look like this. Cancellation is common for actions that change the share count or the identity of the instrument, such as a split, a bonus, a consolidation, a merger, a demerger, a scheme of arrangement, a delisting or a suspension. Price and quantity adjustment, where a member's systems support it, follows the exchange-prescribed adjustment factor. Retention without change is for actions that do not materially move the price or the quantity, which in most policies means certain cash dividends.

Nothing is reserved while it waits

A pending instruction is not an order, so nothing is blocked against it. No margin is earmarked, no funds are held, no holdings are locked. Member policies put the obligation on the client directly: ensure the necessary funds, margins or holdings are available.

There is a second, structural reason no balance sits behind a waiting instruction, and it is worth knowing because it is often assumed otherwise. Under SEBI circular SEBI/HO/MIRSD/MIRSD-PoD-1/P/CIR/2023/084 dated 8 June 2023, as modified by SEBI/HO/MIRSD/MIRSD-PoD-1/P/CIR/2023/110 dated 30 June 2023 and effective from 1 September 2023, stock brokers and clearing members are not permitted to retain clients' funds on an end-of-day basis. Client funds are upstreamed to the clearing corporation, in cash, as a lien on a fixed deposit receipt or as a pledge of units of an overnight mutual fund scheme.

So the money is not sitting at the member overnight, waiting for your instruction, even in principle. The sufficiency test happens at the instant of firing, which is the instant you are least able to do anything about it. A buy instruction set three months ago against a balance you have since withdrawn produces a rejection at precisely the moment the price arrived. The same logic runs on the sell side through holdings: an instruction against shares you have since sold, pledged or transferred fails the holdings check the moment it fires.

Margin is the least intuitive of these, because the requirement itself moves. A derivatives instruction set when the requirement was one number can fire on a day when volatility has raised it, which is again the same day the price reached your level. Our guide to how margin requirements are maintained through a position's life covers why that number is not a constant.

What the feature is honestly for

Both halves of this are true and a fair description has to carry both.

It is a real convenience, and the benefit is larger than the obvious one. The obvious benefit is that you do not have to re-enter an order every morning for three months. The less obvious and more valuable one is that you do not get a hundred opportunities to move the level. A decision made once, calmly, away from the screen, and then left alone, is a different decision from one re-made daily under the influence of whatever the price did that morning. For someone with a job and a plan, that is worth a great deal.

It is also not a resting exchange order, and treating it as one is where the trouble starts. It has no queue position. It depends on systems other than the exchange's. It is spent by firing rather than by filling. Its validity is a number your broker chose. It can be cancelled or reset by a corporate action. It reserves nothing, and it is checked for funds and margin at the worst possible moment. And what it becomes when it fires carries every risk of that order type, undiminished by the fact that a condition preceded it.

The competent use follows from that list rather than from any tip. Read your own broker's published policy and note the validity and the corporate action treatment. Decide deliberately what the resulting order should be, and accept the trade between certainty of fill and control of price rather than pretending the trigger removes it. Diarise your pending instructions against the corporate actions of what you hold, and read the notice when it comes. Keep the funds or the holdings that the instruction will need, because nothing is holding them for you.

That is the difference between using an execution tool and outsourcing a decision to one. The tool is good. It is a tool for carrying out a judgement you already made, and it has never been a substitute for making it.