Position Limits and Margin: Two Different Constraints
Two different questions, two different answers. The amount of a single contract that one participant may hold is settled by a cap, and a cap is a bound on quantity that stays put while the price moves. A margin requirement settles what must be sitting in the account against whatever is carried. The requirement is a share of exposure, and it gets re-struck at every close. Neither stands in for the other.
Both of these turn up on the same statement, both of them stop somebody from doing something, and that is enough for most readers to file them in the same drawer. The two constraints answer two questions that have almost nothing to do with each other. Asking the first is asking what size of obligation the arrangement is prepared to end up chasing from one party. Asking the second is asking what needs to be lodged in advance so that an obligation of any size can be chased successfully. With that split held firmly, everything else about the two falls into place.
Which figures below are prices, and which ones actually move money?
Run one test over every rupee that follows. Did it leave somebody's account? Rs 2,000.00/- did not, so it is a PRICE. Rs 160.00/- did, so it is a PAYMENT. And where several payments in both directions have already been set against each other, whatever survives that setting off is a NET. A net can therefore be small on a day when the payments behind it were anything but. A price and a payment can share a sentence quite comfortably while only one of the two ever reaches an account. Mixing the three is the commonest error anywhere in this machinery, and the easiest to commit.
Quantities take a second test. Multiply the price by how many units are referenced and what comes out is the EXPOSURE. On one unit that comes to Rs 2,000.00/-. A NOTIONAL works differently again. A notional is a multiplier printed on the contract, and no part of it is ever paid over by anybody. Exposure is the figure a margin requirement is struck on. A notional never enters that calculation.
Now notice something. Two figures here both read Rs 2,000.00/-: one of them is the spot price, the other is what a single unit carries as exposure. The two figures land on the same numeral because a single unit references exactly one unit, and that puts a multiplier of one between them. Neither was copied out of the other. At any size except one the two separate immediately. Twenty five units still start from a price of Rs 2,000.00/- while carrying Rs 50,000.00/- of exposure.
One more thing about the reference asset, and it bears directly on every collateral figure below: no credit of any kind reaches the account of whoever holds it. Were a payout to exist, it would land inside the exposure figure first, and the collateral worked out from that exposure would shift with it. The reference asset used in these figures was built without one.
Two figures above both read Rs 2,000.00/-: the spot price, and the exposure a single unit carries. Why?
What exactly does a position limit put a stop to?
A reader meets a cap at the moment it is felt rather than at the definition, so start there. A participant asks to add to a holding and is told that the addition will not be accepted. Nothing has gone wrong, nobody has failed to pay, and the account is in perfect order. The addition is simply not available.
Put formally, a cap says this: in one named contract, nobody acting as a single participant may hold beyond a stated amount, and that amount is stated against a base. The base is not a footnote to the bound; it is half of it. Counting the contracts gives one kind of bound. Express the same bound as a slice of open interestThe running count of contracts still standing open across everyone holding them, taken together. It rises as new positions are created and falls as they are closed out. and it becomes a different object entirely. Write it instead as a quantity of the referenced thing and it changes again. Three bases, three bounds, and a reader handed only a numeral has been handed nothing readable.
Why three shapes rather than one? Because the three shapes hold down three different dangers. A count of contracts is easy to check and easy to compare across participants. A share of what is standing open scales with the contract's own size, so it tightens on a thin contract and loosens on a busy one without anybody rewriting it. A quantity in the referenced thing speaks directly to what would have to be bought or sold if the position ever had to be closed. Each of those is a different theory of what a large holding is dangerous for.
Every cap, every base and every unit is set by the exchange under rules the Securities and Exchange Board of India (SEBI) writes, sebi.gov.in, and they differ by contract. No single value would be true of two contracts at once, so the row below carries the address that settles the figure instead of the figure.
Somebody recalls a participant's cap in this contract as a figure they remember clearly. What is still not known?
What exactly does a margin requirement ask for?
Here too the moment is more useful than the definition. Before a position may be carried at all, something has to be lodged against it, and it has to stay lodged for as long as the position stands. The lodged amount is not a fee, nobody keeps it, and it is not the price of anything. Collateral is encumberedPledged, and therefore no longer free to be used for anything else while the pledge stands. Money that is encumbered still belongs to whoever lodged it. money sitting where it can be reached.
A margin requirement is what has to be put up as collateral against a position, expressed as a share of the exposure that position carries. On the figures used throughout, an exposure of Rs 2,000.00/- carries Rs 160.00/- of collateral, being 8.0 per cent of it, and that 8.0 per cent was chosen for teaching rather than taken from anywhere. Real requirements come from clearing corporations under rules SEBI writes, sebi.gov.in, and they move.
Now the second limb, and it has to come in the same breath as the first. On its own the first flatters the arrangement. On its own the second frightens a reader out of understanding it. Each rupee of the Rs 160.00/- lodged is standing behind 12.50 rupees of the Rs 2,000.00/- exposure. Work an adverse move of 4.0 per cent on that exposure and the move is Rs 80.00/-, a PAYMENT. Set that against the Rs 160.00/- put up and half of it has gone: 50.0 per cent of the collateral, on a day the reference asset moved by one twenty fifth of its own price. The ratio between the two, rather than the exposure figure or the collateral figure taken separately, is what makes a position like this different to hold.
Two things sit outside what a margin requirement is, and they are both worth pushing away now so they stop crowding the comparison. The collateral put up is not the cost of carryWhat it costs to hold something from today until a later date, financing included. It belongs to how a forward price is built, and it is worked out separately. of the position. Cost of carry is a different quantity, built out of a financing rate. And what may be lodged, together with the haircutA discount applied to something lodged as collateral, so that a rupee of it counts as less than a rupee for the purpose it was lodged for. applied to it, is decided by SEBI, sebi.gov.in.
A 4.0 per cent adverse move takes 50.0 per cent of the collateral on one unit. Where does the 50.0 come from?
What does each of the two constrain, once both are on the table?
A cap constrains how much may be carried. A margin requirement constrains what must be sitting there against whatever is carried. Put anything between those two statements and a reader starts looking for an overlap. There is no overlap.
Followed out to the uncomfortable end, a cap can be satisfied perfectly well by somebody with nothing at all in the account. A bound on quantity says nothing whatever about backing. And a margin requirement can be satisfied perfectly well by somebody carrying a holding the arrangement would never have permitted. A share of exposure is a share of exposure at any size. Satisfying either of them says nothing about the other.
Set them side by side on the axes that actually differ, and what comes out is a table with no agreement anywhere in it. Six axes, six differences, and no row where the two constraints do the same thing.
The price of the reference asset climbs sharply. Choose one now, then read what follows: which of the two constraints moves?
Which of the two moves when the price moves?
A margin requirement is struck as a share of exposure, so it is re-struck every time the position is revalued. The requirement moves when the price moves, and it moves again whenever the share itself is changed. At Rs 2,000.00/- of exposure, the invented 8.0 per cent asks for Rs 160.00/-. Push the price to Rs 2,400.00/- and the same invented share asks for Rs 192.00/- on the same one unit. Nothing about the position changed. The number attached to it did.
A cap does not move when the price moves, at all. A cap is a bound on quantity, and a quantity is indifferent to what it happens to be worth this morning. Ten of something is ten of something at any price anybody cares to put on it.
Put together, those two produce the sentence worth carrying away. As a price rises, what has to be put up rises with it while how much may be carried stays exactly where it was. The two constraints are travelling in different directions on the same day, and a reader who was treating them as two versions of one idea now has two things to watch instead of one.
Of the six axes in the grid above, one of them explains most of the other five. Which?
How often does each one speak, and what does it say?
A cap is tested when a position is taken and when it is added to, and then it goes quiet. A holding that sits unchanged for a fortnight is not being asked about its cap on any of those mornings. Nothing about the holding has changed, and a cap only ever looks at quantity.
A margin requirement is tested at every close, without exception. A position that has not been closed still has to be revaluedGiven a fresh price although nobody has closed the position, so the account can be brought up to date. How that fresh price is arrived at is settled separately., and the collateral due against it is worked out from whatever that revaluation produces. There is nothing occasional about it. The requirement speaks daily, and it speaks whether or not anything interesting happened.
The consequence of each one biting is different in kind, and that difference decides what a reader has to be ready for. Reach a cap and nothing arrives: no demand, no payment, no entry in the account. The position holder loses options. The next thing they wanted to do is no longer available to them. Fail to keep the collateral where it needs to be and something very concrete arrives. A demand for money lands, and meeting it takes cash out of the account the same day.
One of these two takes away the ability to act and the other takes money out. A participant can prepare for the second with a pot of unencumbered cash and cannot prepare for the first that way at all. The consequence of passing a cap, and the consequence of leaving a demand unmet, are both decided by SEBI, sebi.gov.in.
A position reaches its cap on Monday. On Tuesday, the collateral against a different position falls short. What arrives in each case?
Who is each of these two constraints protecting?
Neither one of them is there for the position holder, and the reasons they are not differ. A reader who has just learned that both constraints restrict them naturally assumes both were designed with them somewhere in mind.
A cap exists for the arrangement and for everybody standing inside it. Here is why. Size measured against the referenced thing is what makes a holding hard to leave: one big enough on that measure cannot be unwoundGot out of by taking the opposite side of the position, rather than by waiting for the end date to arrive. quietly. Leaving means trading in size, and trading in size has market impactThe price movement that trading itself causes. When enough of anything is sold quickly, the level obtained for the last part is worse than the level available at the start.. And the price that shifts is the one used to mark everybody else's holdings in the same contract. So a single exit reaches into accounts that had nothing to do with it. Making that impossible is what the cap exists for.
A margin requirement exists for the promise. Its whole job is to make sure that what is owed can actually be collected from whoever owes it. The timing follows straight from that job. Collateral is demanded early, at a point where nothing has been closed and nobody has yet been asked to produce money they do not have. Leave the demand until a loss has crystallised and the party owing it has usually acquired other problems by then. Every rupee a margin requirement asks for is aimed at the collectability of an obligation rather than at the size of one.
Neither constraint was designed around any one participant's circumstances, so a reader hunting for one comes away empty handed. One of the two was struck for the market as a whole and the other for the enforceability of an obligation, and neither has ever looked at a single participant's plans.
Could a big enough collateral demand simply replace a cap?
Most readers actually arrive with this question, usually in the form of a suspicion that caps are redundant. To stop people carrying enormous positions, the thinking goes, make the collateral punishing enough and the problem solves itself. The answer is no, and the arithmetic says so rather than an opinion.
Take one unit first. The exposure is Rs 2,000.00/-, the collateral at the invented 8.0 per cent is Rs 160.00/-, and each rupee lodged is therefore carrying 12.50 times its own value. Now take an invented holding of twenty five units. Exposure Rs 50,000.00/-, collateral Rs 4,000.00/- at that same invented share, and dividing the one by the other returns 12.50 times once more. Every ratio the collateral produces is identical at both sizes, and it would be identical at any size tried.
The matching ratio is not a fluke of these particular numbers. The result falls out of the definition, and one line of algebra shows why the size disappears entirely.
| L | the exposure carried for each rupee of collateral, in times |
| E | the exposure the position stands on, in rupees, taken from the price and the number of units |
| M | the collateral put up, in rupees, read off the statement |
| m | the share of exposure the collateral has to be, here an invented 0.08 |
Raising what has to be put up makes a position more expensive to carry and never once makes it impossible to carry a larger one. Somebody with more collateral simply buys a proportionally larger holding at the higher requirement, and the arrangement is exactly where it started, facing a position it cannot see the size of. A constraint that scales with the thing it is meant to constrain is not a bound at all. Such a constraint is a price.
Somebody proposes that a high enough collateral requirement would make caps unnecessary. What does the arithmetic say?
So which of the two decides the size actually carried?
Whichever of the two binds first, and which one that is depends entirely on the particular case. There is no general answer.
Take a party with a great deal of collateral and a large intended position. Their account will fund far more than the arrangement is willing to see in one pair of hands, so the cap is what stops them, and it stops them at a size their collateral would have covered comfortably. Now take a party with a modest amount lodged. The lightly collateralised party runs out of collateral long before the cap comes into any kind of view, and for them the cap is simply never the operative constraint. Same contract, same rules, opposite experience.
Which of the two binds is a fact about the party, not about the contract. Settling it in the abstract would require knowing what a party holds, what it could lodge tomorrow, and what it is trying to do, and none of that is written into the contract.
Two parties intend an identical position. One has a great deal of collateral, one has very little. Which constraint binds for each?
Why does one of these two constraints carry no figure at all?
Look back over everything above and one asymmetry stands out. The margin side has been worked in rupees at two sizes. The cap side carries no figure at all, in any unit and against any base. The asymmetry is deliberate, and the reason for it is arithmetic rather than caution.
Here is the margin side, laid out so both sizes can be read against each other in one place.
| What is being measured | One unit | An invented holding of twenty five units |
|---|---|---|
| Exposure the position stands on | Rs 2,000.00/- | Rs 50,000.00/- |
| Collateral put up, at an invented 8.0 per cent | Rs 160.00/- | Rs 4,000.00/- |
| Exposure carried per rupee of collateral | 12.50 times | 12.50 times |
The bottom row is the finding. Multiplying the holding by twenty five multiplied the exposure by twenty five and the collateral by twenty five, and left the relationship between them exactly where it was. A margin requirement is therefore a cost of carrying a position rather than a bound on how large that position may be.
The cap side cannot be laid out the same way. Every cap, every base and every unit belongs to a decision the exchange takes under rules SEBI writes, sebi.gov.in, and those decisions differ by contract. Writing one down would be inventing a fact. The shape of the two constraints can be set side by side where their sizes cannot, and the shape survives every revision either figure is put through. Six axes of difference survive any revision. A pair of numbers would not.
The error that gets made here, and what it costs
Somebody works out how large a position they can carry by looking only at what they are able to put up. Rs 4,000.00/- of collateral, divided by Rs 160.00/- due on each unit at the invented 8.0 per cent, funds twenty five units. Twenty five units becomes the plan, and it goes into a spreadsheet, into a note to somebody senior, and into the sizing of everything the position was meant to sit against.
The mistake is not the beginner's error. Readers who have understood the collateral machinery well are the ones who make it, and every step of their arithmetic is correct. The correctness is what makes it dangerous. There is nothing to catch. The division is right, the percentage is applied properly, and the answer of twenty five is the true answer to the question the planner asked.
The cost is the plan. The size that may be carried was never that arithmetic's to settle. A cap is struck against a base the planner has not looked at, it does not move when the price moves, and it can bite at a size the collateral would have funded without straining. When that surfaces, the plan gets remade late and under time pressure, and whatever part of it depended on the position being that size has to be dropped.
Two questions kill it, and they kill it better than a warning would. One of them asks what has to be put up, and that is a question about the promise. The other asks how much may be carried, and that is a question about the arrangement. Only the first of those is answered by anything sitting in the reader's own account.
Somebody sizes a position purely from the collateral they can lodge. What have they left out, and when do they find out?
Who has to tell these two apart in ordinary work?
Start with something with no derivative in it at all. A bank sanctions a borrower a home loan up to a stated amount, and separately asks that borrower to put in a share of the flat's price before it releases anything. The sanction and the share are the two constraints in the shape most people have already met. The sanction is a bound: it does not rise because the flat chosen got more expensive. The share the borrower has to find is a proportion: it rises with the price of the flat, every time, and it rises again if the bank changes the share. Finding more of one's own money does not raise the sanction by a single rupee, and being sanctioned a large amount does not reduce what has to be found. A wedding hall behaves the same way. The hall seats what it seats, whatever the caterer's advance turns out to be.
Now the same distinction in work. Somebody lending against a business that carries derivative positions is trying to answer a question about survival: on a bad week, how much cash walks out of this borrower's account before anything else gets paid? The survival question is a margin question end to end. A cap has never taken a rupee out of anybody's account, so the cap is close to irrelevant to it. The lender wants to know the exposure, the share it attracts, and what unencumbered cash sits behind that share.
Somebody reading a participant's disclosures has the opposite worry. A disclosure reader is trying to work out whether the participant can keep doing what they have been doing, and that is a cap question. A participant already close to a bound has a business that cannot grow along its current line however much collateral it raises, and no amount of reading the collateral figures will show that. The two readers are working through the same document and taking away entirely different things from it.
The two constraints fail in ways that do not warn each other, so somebody inside the participant has to hold both at once. Collateral can be ample on the morning a cap stops an addition. A cap can be miles away on the afternoon a demand for money arrives that nobody had cash ready for. Preparing for one of the two is not preparation for the other, and treating them as one constraint means being ready for neither at the moment it matters.
What is settled by the authority
SEBI settles the ceiling on one participant's holding in a single contract, at sebi.gov.in. SEBI decides what a position has to be backed with, and the working that produces the figure, at sebi.gov.in. SEBI fixes what a member may carry in total across all the participants it acts for, at sebi.gov.in. SEBI rules on how far a lodged balance may fall before something happens, and on what that something is, at sebi.gov.in. SEBI keeps the further demand made during a day, and the point in the day when it is made, at sebi.gov.in.
Not one of those five rows can carry a fixed value. Each of the five belongs to a decision SEBI revises on its own schedule. Two contracts sitting side by side can carry different answers in the same row on the same morning. So the cells stay open and the address does the work the value would have done.
A principle about how markets that clear centrally should behave across borders does exist, and the International Organization of Securities Commissions (IOSCO), iosco.org, is where it comes from. SEBI's rendering of that principle is what binds anybody operating in India, rather than the principle itself. Where a rate or currency arrangement is agreed between two parties instead of on an exchange, the Reserve Bank of India, rbi.org.in, is the address.
The 8.0 per cent used in every calculation above sits outside this block entirely. The figure was chosen for teaching, it is labelled that way wherever it appears, and it should never be read as the empty row's missing value.
What neither constraint settles, and why
By this point a reader will have started working out how large a position could be carried. The instinct is right. Neither constraint above was struck with any one participant in mind, and neither one of them answers that question either. A cap was struck for the arrangement. A margin requirement was struck for the promise. Nothing in either has ever looked at a particular account.
The inputs that would settle it sit with the position holder. The purpose of the position comes first. A holding taken to cancel out a risk the business is already running is not the same animal as a holding taken on its own account. Then there is what has already been lodged, and separately what could be lodged tomorrow morning if the phone rang before the market opened. And the last input is a bad day rather than a typical one. Averages are no use for this. The figure that matters is what lands in the account when everything moves against the position at once.
The twenty five unit holding used above is a second size, introduced to show that the ratio does not budge.
Which body settles each figure
| Source | What is confirmed there | Site |
|---|---|---|
| SEBI | The largest holding one participant may run in a single contract | sebi.gov.in |
| SEBI | What collateral a position attracts, and how that figure is arrived at | sebi.gov.in |
| SEBI | A member's total across all the participants it acts for | sebi.gov.in |
| SEBI | The floor a lodged balance is measured against | sebi.gov.in |
| SEBI | A further demand inside a trading day, and its timing | sebi.gov.in |
| Reserve Bank of India | Rate and currency arrangements agreed between two parties | rbi.org.in |
| IOSCO | Cross-border principles for markets that clear centrally | iosco.org |
The reference asset, its starting price, the position holder and the 8.0 per cent collateral share are invented.
Educational material. Not advice on any investment, tax, budget or market position.