Guide · Derivatives and risk · Updated 2026-09-18

A ban period is a limit on the security, not on your account, and since October 2025 you may open positions inside one

Every other limit a retail trader meets is a limit on the trader. Margin is computed on your positions. The client level position limit is a percentage of your exposure. Lot size fixes the smallest unit you may buy. All of them respond to something you did, and all of them can be seen coming.

The ban period responds to nothing you did. It is a property of the security, computed from the summed exposure of every account holding derivatives on it, and it lands on an account that has been dormant for a week exactly as it lands on one that has been trading all day. The usual reaction to it is wrong in both halves: the trader assumes they have been caught by a limit, and then assumes they are locked into whatever they hold. Neither is true, and the second error is the expensive one.

A limit on the security, and you are not what it measures

The market wide position limit, written as MWPL, is a ceiling on the aggregate open interest across all futures and all options contracts on a single underlying stock, summed over every participant and across exchanges. It exists because derivatives on one company settle against that company's shares, and the quantity of shares that can actually be delivered is finite. Without a ceiling, the promises written on a security can outgrow the security.

Read that definition for what it leaves out. It says nothing about who holds the positions, in what proportion, or with what intent. One institution could hold most of the aggregate, or two lakh retail accounts could hold a few lots each. The limit does not distinguish between those cases because the deliverability risk is identical in both.

Your own exposure is governed separately, by a client level limit expressed as a percentage of the same MWPL. Since 1 October 2025 those percentages are 10 percent of MWPL for a client or an NRI, 20 percent for a trading member's proprietary book, and 30 percent for a trading member's combined proprietary and client book, for a Category I foreign portfolio investor and for a mutual fund. Category II foreign portfolio investors sit at 20 percent, dropping to 10 percent where the investor is an individual, a family office or a corporate.

Those two tests share one input and nothing else. A retail account carrying a handful of lots is typically three or four orders of magnitude below its own 10 percent ceiling. It will meet the market's 95 percent long before it comes anywhere near its own, and it will meet it through the actions of strangers.

This is the whole reason a ban arrives without warning. There is no number on your own screen that was approaching anything.

Free float sets the ceiling, and since October it is not the only thing that does

Until 30 September 2025 the MWPL for a single stock was 20 percent of the number of shares held by non promoters, which is to say 20 percent of the free float, expressed as a number of shares. One input, one percentage, changing only when the share register changed.

From 1 October 2025 it is the lower of two numbers, with a floor beneath both, recomputed every three months on a rolling three month average.

The second leg is the change that matters, and it is a change of kind rather than of degree. Free float is a static property of the share register: it says how many shares sit outside promoter hands, and nothing at all about whether those shares move. Average daily delivery value measures how much stock is actually delivered in the cash market, so it says how deliverable the float really is. A large float that never turns over is not a large supply of deliverable shares, and the old formula could not tell the difference.

Which leg binds therefore varies by security, and it now varies over time for the same security.

The last row is the practical one. A security that rarely entered ban periods can start entering them because its cash market delivery fell, not because derivatives activity rose. The tolerance moved, not the behaviour. The quarterly recut is therefore a date worth knowing for anything you carry regularly in single stock derivatives.

This is the same family of control as the periodic resizing of contracts, which works on the other side of the problem by fixing how much exposure one lot carries. The reasoning behind that mechanism is set out in why lot sizes get revised and how the contract value band works.

Open interest is now a delta number, not a count of contracts

The figure measured against MWPL used to be notional open interest: add the open contracts, convert to shares, done. A long call and a short call of the same size each contributed their full notional, even though one gains when the stock rises and the other when it falls, and even when both sat in the same account.

Open interest is now computed at portfolio level as the net delta adjusted position across futures and options for one underlying. Long futures carry a delta of plus one times notional. Long calls run between zero and plus one. Long puts run between zero and minus one. A short call carries an effective delta that is negative, and a short put an effective delta that is positive. The result is futures equivalent open interest, usually shortened to FutEq OI: the number of futures your book is economically equivalent to.

The aggregation rule is the part most explanations omit, and it is the part that decides whether you are about to be banned. Each account's positions are netted into one signed number. Those signed numbers are then added across accounts without netting.

Account D is the instructive row. It carries eight hundred shares of notional exposure and contributes nothing at all to the figure that triggers a ban, because its two short legs offset inside the account. Account C carries twelve hundred of notional and contributes three hundred and fifty. Under the old method the market wide figure here would have read 2,800. Under the current method it reads 900, and the same security is nowhere near a ban that it would previously have been pushed into.

The regulator's stated purpose for the change was to reduce instances of spurious ban periods in single stocks. Under notional counting, a market full of hedged books reported an enormous aggregate and reached 95 percent on exposure that did not exist in economic terms. Delta adjustment strips that inflation out at the account level while leaving genuine directional build up fully visible.

Two consequences follow, and they point in opposite directions. Because your book is netted before it enters the sum, a hedge you place for your own reasons also lowers the number that decides the ban. And because the sum is gross across accounts, nothing you do can offset what anybody else is doing. If open interest is unfamiliar ground, what open interest actually measures and what it does not is the prerequisite for the rest of this.

Ninety five to enter, eighty to leave, and the gap is deliberate

At the end of each trading day the exchanges and clearing corporations test whether the market wide FutEq OI in a security exceeds 95 percent of that security's MWPL. If it does, the security enters a ban period with effect from the next trading day. The ban does not then lift at 95. It lifts when the aggregate comes back down to 80 percent or below.

The asymmetry is deliberate. A single threshold would make the state of the security oscillate: cross 95, forced reduction pushes the aggregate to 94.5, the constraint lifts, positions rebuild, and 95 is crossed again before the next close. A security could enter and leave a ban several times in a week, and every entry and exit is itself a real order flow event, because a forced reduction is a wave of actual orders. Fifteen percentage points of separation means the aggregate has to genuinely unwind before normal trading resumes.

The teaching consequence is sharper than the design. The percentage alone does not tell you the state. At 88 percent of MWPL a security may be trading normally or may be in a ban period, and the only thing that distinguishes the two cases is which direction it arrived from. The published list is the fact. The ratio is context.

Increase and reduce are properties of your net delta, not of the words on the order

Here is the rule as it now stands. Any trading you do in a banned security after it has entered the ban must result in a reduction of your FutEq OI measured at the end of the day. Not per order. Not per leg. One signed number per underlying per client code, compared with the same number at the previous close.

Three qualifications complete it, and each one changes an answer. The floor is zero, so a book at plus ten may be reduced toward zero and a book at minus ten may be reduced toward zero. A change of sign does not count as a reduction, so moving from plus ten to minus eight is a breach even though the absolute value fell. And a passive increase caused only by the underlying moving is not a breach at all: hold one long futures lot, watch the stock rise five percent, and your delta rises with it while you have done nothing.

Read against that rule, the words buy and sell carry no information. Selling a call reduces a long book. Selling a put increases it. Both are sales.

The closing trade that counts as an increase. Suppose your book is net long delta because you hold a long future with a short call written against it. Buying that call back is a closing trade in every ordinary sense of the phrase. It also removes a negative delta, which raises your net. Under the rule it is an increase, and it is precisely the trade a person reaches for when they believe that closing anything is always permitted.

The roll. Rolling a long future from the near expiry to the next is two orders: sell the near contract, buy the far one. Net delta at the close is unchanged. Unchanged is not reduced, so the roll fails the test. This is why a ban that arrives in the final days before expiry is the case that actually costs something, and why the expiry calendar and the ban list should be read as one document rather than two.

Spreads inherit the sign of the book they land in

A spread has no fixed status under this rule, because the rule reads your book rather than your trade. That is the single most useful thing to understand about it.

A call debit spread, long a lower strike call and short a higher strike one, carries net positive delta. Placed into a book that is already net long, it increases the net and is blocked. Placed into a book that is net short by more than the spread's own delta, it reduces the net and is permitted. The same two orders, the same security, the same session, opposite answers, decided entirely by what you were already holding.

A short strangle sits near zero delta at inception and therefore barely moves the number. Barely moving is not moving downward. A structure whose net delta lands even marginally on the far side of your starting point fails the test, so an approximately delta neutral overlay is not automatically safe and should not be treated as the default way to keep trading a banned name.

The general statement is that during a ban, the only property of a proposed trade that the exchange cares about is the sign and size of the delta it adds to the net you already hold. The strikes, the payoff shape, the cost and the margin are irrelevant to permission. Margin remains a separate and simultaneous constraint that has to be satisfied on its own terms, and the way it is computed and maintained is set out in how margin is calculated and what maintenance actually means.

Why your order window says blocked when the rule says permitted

The regulation permits a risk reducing fresh position. A great many order windows do not, and the gap between those two facts is where most of the confusion about ban periods now lives.

The reason is in the shape of the test. The obligation settles at the end of the day, on a net delta that cannot be known when an order is entered, because delta moves with the underlying for the rest of the session and because your own later orders will move it too. A member has to decide at eleven in the morning what the clearing corporation will only calculate at the close.

The asymmetry of consequences settles the question. A rejected order costs the client an opportunity. A permitted order that turns out to have raised the end of day delta costs the member a penalty and a line on a surveillance report. Faced with an uncertain forward looking test and a certain backward looking sanction, a risk system blocks and a compliance team supports it.

So the working position for a trader is threefold. The regulation permits more than your screen does. The screen is not wrong to be conservative, given who pays. And the way to find out what is actually available to you in a banned security is to attempt the reducing trade and read the rejection, rather than to assume the whole security has been frozen and act on that assumption.

The penalty, and who the exchange actually bills

The long standing published penalty for creating or increasing a position in a security during a ban period is 1 percent of the value of the increase, subject to a minimum of 5,000 rupees and a maximum of 1,00,000 rupees, applied per instance.

Two features of that structure explain the behaviour it produces. The minimum means a trivially small breach is not trivially penalised: a one lot excess whose 1 percent computes to a few hundred rupees is still charged at five thousand. The maximum means a very large breach is capped in money terms, so the monetary deterrent stops scaling, and the exchange reserves the ability to escalate beyond money for repetition, through higher margin requirements or through restrictions on trading rights.

The billing is the part traders misread. The penalty is raised on the trading member, who then recovers it from the client whose position caused it. The member is on the hook first for a limit the client crossed. That single fact is the complete explanation for pre-emptive order rejection, and it is why arguing with a risk system about what the circular permits is an argument aimed at the wrong party.

Under the current framework the clearing corporations were directed to frame a joint standard operating procedure covering the monitoring of end of day delta positions and the penalty structure attached to that specific test. The operational detail therefore sits with the clearing corporations rather than in the circular itself, so confirm the schedule that applies to your account with your own member rather than assuming the historical numbers carry across unchanged.

Where the list comes from, and when it reaches you

The securities in ban period list is produced by the exchanges from the end of day test and published after the close, applying to the following trading day. It names securities, never accounts, and it carries the utilisation figure for each security alongside the name.

Since 3 November 2025 the exchanges also monitor MWPL utilisation during the session, at a minimum of four randomly timed snapshots. That monitoring drives surveillance action such as additional surveillance margin and entity level concentration checks, and it feeds fortnightly reporting. It does not create an intraday ban. The ban itself still turns on the end of day test, which means a security can spend an afternoon above 95 percent and not be banned if it closes below.

The reading order that actually helps anyone carrying single stock derivatives overnight is short. Read the list on the evening it is published rather than on the morning it takes effect, because the list applying to tomorrow exists tonight and tonight is when you still have every option. Read the utilisation figure next to the name, because a security at 92 percent and climbing should be treated as one that will be listed. And read both against your own expiry calendar, because a ban that lands in the last week before expiry removes the roll, which is often the only action the position needed.

What the constraint is actually for

MWPL is not an anti speculation measure and it is not aimed at retail participation. It is a deliverability constraint. Derivatives on a single company settle against that company's shares, the quantity of shares that can actually be delivered is finite, and since October 2025 that quantity is measured directly rather than inferred from the share register alone.

Seen that way the design reads cleanly and nothing in it is arbitrary. The limit is derived from free float and delivery because those are the supply of the thing being promised. It is measured in delta because delta is the economic size of the promise rather than its paperwork. It binds on the market rather than on the participant because the risk it addresses is a market risk. And it permits reduction at every moment because the purpose was never to trap anyone, only to stop the aggregate growing.

The judgement it asks of you is small and specific, which is what makes the widespread confusion about it so costly. Know before the close whether something you hold is likely to be listed. Know which direction your book's delta points, as a signed number rather than as a feeling about your positions. Know that reduction is always available and that the definition of reduction is arithmetic rather than intuition. That is the whole of it, and it is a fair example of the general case: a constraint understood mechanically costs you one decision, and the same constraint understood as a rumour costs you a position.