Rollover: Moving a Position to the Next Contract

A rollover ends a position in the contract about to run out and opens the same position in one that runs further out. A rollover is two trades, not an extension. With the spot price standing still, the further contract costs more only because it carries the reference asset for longer, so the difference is financing and not anybody's view about where the price is going.

A rollover comes out of one mismatch. A futures contract has a fixed end written into it by the exchange. A reason for holding a position very often has no end at all. The two facts cannot both be satisfied inside one contract, so the position has to be put down in the contract that is finishing and picked up again in one that runs further out. The only questions left worth asking are what that costs, and where the cost comes from.

A trader at the mandi keeps forty sacks of onions in a rented cold room. The rent agreement runs to the end of this month. The onions are not going anywhere at the end of this month. So a fresh agreement gets signed for the next three months, the old one is settled and closed, and the new one costs more than the old one did. Three months of room rent is more room rent than one month. Nobody at the mandi reads the larger figure as an opinion about onions. The trader has just rolled over, and everything below is that scene worked in rupees.

What is a rollover, and how many trades does it take?

Two, and knowing that is most of the battle. The position sitting in the contract that is about to run out gets closed by a matching trade the other way, placed inside the contract that is finishing. Then a position on the same side as before gets opened in a contract that runs further out. Two orders, two contracts, two separate obligationsSomething a party is bound to do whether it suits them on the day or not., and the second one has nothing to do with the first beyond the fact that the same party wanted both.

There is no such thing as extending a futures contract. Nothing is renewed, nothing is stretched, no date is pushed back and nobody agrees to carry on. The first obligation is discharged completely, and a second obligation with a different end date begins from a clean start. Almost every error below grows out of one picture: a single agreement being lengthened. A reader who has been told that a position was rolled forward should throw that picture away first.

The two trades are ordinarily placed together so that the party is never left standing without the exposure it wanted. Placing them together is a matter of practice rather than of arithmetic, and the practice differs. The stretch of days over which positions travel out of one contract and into the next, and whatever is permitted inside that stretch, is a matter for the exchange, working inside what the Securities and Exchange Board of India (SEBI) lays down at sebi.gov.in.

Two things a roll needs cannot be got out of arithmetic: the calendar dates the contracts run to, and the width of the gap between the closing trade and the opening one. Both belong to an authority. Where the days of a position have to be spoken about, they are numbered instead: the day the position moves, the day before it, the day after it. The real interval between the two trades is set by SEBI, sebi.gov.in, and by nobody else.

Why does the position have to be picked up and put down at all?

Because the contract runs out and the reason for holding the position does not. That is the whole of it. A contract that ends in three months stops being an instrument three months from now no matter how convinced anybody is that they want the exposure for two years. The end date is not negotiable and it is not a variable in anybody's model; it is written into the terms by the exchange before a single party turns up.

A party holding a position in a contract with three months left has exactly three routes open, and there is no fourth. All three are worth naming together, because a reader who has met only the first two keeps looking for one more.

The first route is to close the position and be finished. Place the reverse trade inside that very contract, the obligation cancels against itself, and the party walks away with whatever the running total came to. Closing is a decision to stop having the exposure, and it is a perfectly ordinary thing to do.

The second route is to hold on and do nothing. A position that is neither closed nor moved arrives at the end of the contract still open, and at that point the reference asset changes hands for real. Somebody delivers it and somebody pays for it and takes it in. Delivery is worked out in full separately, and it matters here only as the thing a rollover is an alternative to.

The third route is the rollover, and it exists so that a party can keep the same exposure past the end of a contract without ever having to take the reference asset in. Close in the contract that is finishing, open on the same side in one that runs further out, and the exposure carries on uninterrupted while the contract holding it changes underneath. What continues and what does not is worth stating precisely. The exposure continues. The intention continues. The contract does not, the price does not, and the running total does not, and the last two of those are where readers get caught.

What does the move actually cost, and where does that money go?

Here is the setting, and one part of it is deliberately frozen. One unit of the invented reference asset stands at a spot price of Rs 2,000.00/-. Financing costs 6.50 per cent a year. The reference asset yields its holder not one paisa over the months in question. And the spot price is held at Rs 2,000.00/- across every single line of the working below, on purpose. Every rupee of difference that appears then has only one place it could have come from.

Take the contract about to run out first. The near contract has three months to go, so the financing on it accruesTo build up little by little as time passes, rather than landing in one lump on one day. over a quarter of a year. Run Rs 2,000.00/- at 6.50 per cent a year across a quarter of a year and Rs 32.50/- falls out. Stack that onto Rs 2,000.00/- and the contract comes out at Rs 2,032.50/-.

Now the contract being moved into. The contract being entered has twelve months to go. Twelve months of that same rate on Rs 2,000.00/- reaches Rs 130.00/-. Stack that onto the spot price and the contract comes out at Rs 2,130.00/-. Both figures are worked the same way, off the same unmoved spot price, with the same rate, and the only input that differs between them is how long each one has to run.

The rate is quoted annualisedStated as a full year of it even where the stretch being measured is shorter, which means the figure has to be scaled down before it is used.. The three month figure is therefore a quarter of the twelve month figure rather than the same size. Financing here is charged pro rataSplit in proportion to a share of something, so a quarter of the time carries a quarter of the amount. across the months, and that proportionality is what makes every number in the rest of this guide predictable rather than something that has to be looked up.

Rs 2,130.00/- less Rs 2,032.50/- is Rs 97.50/-, and that is what the move costs. Now check it the other way, without touching either contract price. Take Rs 2,000.00/- through nine months at the 6.50 per cent annual rate and Rs 97.50/- is what accumulates. The two routes land on the same figure to the paisa, and they do it because they are the same calculation approached from opposite ends. Nine further months is exactly what was bought, so nine further months of financing is exactly what was paid.

Look at what dropped out of that subtraction. The spot price appears on both sides and cancels. The cancellation is the algebraic version of what the held figure was doing in the worked example. Whatever the reference asset is worth, the gap between two contract prices struck off it is the financing across the months between them. Move the spot price and both contract prices move together; the gap changes size in proportion but it never changes into something else.

Does the cost change if fewer months are bought?

The cost does change, and it changes in a way that can be predicted without being told. A single roll cost is easy to distrust; a reader who is shown one figure will reasonably suspect it was chosen. So here is the same contract about to run out, with three months to go and standing at Rs 2,032.50/-, against every rung it could be moved into.

Every figure in that table has been checked twice. Rs 2,065.00/- less Rs 2,032.50/- is Rs 32.50/-, and Rs 2,000.00/- financed over a quarter of a year at 6.50 per cent a year piles up Rs 32.50/-. Rs 2,097.50/- less Rs 2,032.50/- is Rs 65.00/-, and half a year of the same rate on the same amount gives Rs 65.00/-. The bottom row has already been seen. The cost of a roll is the carry over the months bought, and nothing else is in it.

One thing in that table looks like a coincidence and is not. Rs 32.50/- turns up twice: as the carry to three months on the contract being left, and as the cost of buying three further months. Rs 97.50/- turns up twice as well, as the carry to nine months and as the cost of nine further months. Both pairs are forced by the arithmetic. The carry over a stretch of months does not care whether that stretch sits at the beginning of a contract or in the middle of a ladder, so the same multiplication produces the same rupees in both places. Reading either coincidence as a signal about anything would be reading a multiplication table as news.

Two consequences follow, and neither is obvious until the table is laid out. The cost scales with the months, so a three month roll costs a third of a nine month roll, exactly. And the cost scales with the spot price, so the same roll on a reference asset standing at Rs 4,000.00/- would cost twice as much in rupees while costing precisely the same in proportion. A roll cost is not a fixed amount, not a fee, and not something a party can shop around for. A roll cost is an amount of financing, and the only things that set its size are the rate, the time and the price of the thing being referenced.

Take the control down to the low end and something useful happens. At no further months the position would be entering a contract of the same length it is leaving, so nothing is bought and nothing is paid, and both bars vanish. The low end settles what a roll cost actually is. A roll cost is not an administrative charge for the act of moving. A roll cost is the price of time, and where no time is bought the price of it is nil.

Could a contract further out ever cost less than a nearer one?

Yes, and the reason is worth naming even though the figures here cannot demonstrate it. Put a payout into the reference asset and the payout comes off the financing before the contract price is settled. The spot price can then end up standing above the later contract price. Whoever holds the reference asset over those months collects something, and that collection reduces what holding it costs. Push the collection above the financing and the cost of holding goes negative, at which point a contract running further out is worth less than a nearer one and the party rolling into it receives money rather than paying it.

The reference asset priced above hands over nothing at all during the months it is held, so its later contracts always come out the dearer ones. Every figure above is struck off an asset that produces no receipt of any kind, so the cheaper-further case can be described but not worked.

Two things follow that are worth carrying away. The direction exists and it is real. And it is a property of the referenced thing rather than of the contract: nothing about a futures contract makes a later one dearer or cheaper, and everything about what the referenced thing does while somebody holds it does. A reader who meets a market where the further contract is cheaper has not met a different kind of contract. The underlying thing is what differs, and the contract is not.

What happens to what was lodged, and to the count of contracts?

Two things move that a reader is likely to assume stay put. The first is the collateral. A position that gets closed has its lodgement released, and a position that gets opened has a lodgement placed against it. The release and the placement are two separate events on two separate positions, and nothing travels from one to the other under its own steam.

On the illustrative figures, an initial margin of 8.0 per cent of the Rs 2,000.00/- of exposure is Rs 160.00/- a unit. Rs 160.00/- comes back as the near position closes and Rs 160.00/- goes down against the position being opened. The 8.0 per cent is a teaching number rather than a requirement anybody faces. A party's actual lodgement before a position may be carried is worked out by clearing corporations, and SEBI settles the method behind it at sebi.gov.in, publishing both in one place. The lodgement differs by contract. The lodgement differs by day.

Why does the difference matter, when the amount happens to be identical on both sides here? Because it is identical only because the illustration made it so. The two positions sit in different contracts with different lengths left to run, and what gets asked of each of them is worked out separately by the people who work it out. A party planning a move needs the incoming lodgement arranged as a thing in its own right, not as an amount it expects to find already in place.

The second thing that moves is the count of contracts outstanding. A closing trade takes the position out of the contract being left, and an opening trade puts one into the contract being entered, so the count falls in one and rises in the other. The count itself, how it is read and what a roll does to it are covered separately.

What is left behind when the position moves?

Three things do not travel, and the third one is where money gets misread.

The agreed price does not travel. A position opened in the later contract is opened at that contract price, Rs 2,130.00/- in the worked example, and the Rs 2,032.50/- it used to be measured against belongs to a contract that no longer holds this party. The history does not travel either. Whatever the running total on the closed position came to has been settled to that contract close and is finished, and the position in the new contract begins its own running total at nil on the day it opens.

And the reference point does not travel, which is the consequence that matters: after a roll, the price the position is measured against is the price the new contract was opened at. Setting Rs 2,032.50/- against a settlement price in the twelve month contract produces a number, because subtraction always produces a number. The result is not a payoff, not a profit and not anything else. It is a price from one contract subtracted from a price in a different contract, and the two are not measures of the same thing.

Four words get used loosely around a roll and are worth pinning down. Rs 2,000.00/-, Rs 2,032.50/- and Rs 2,130.00/- are all PRICES, either agreed or worked out. Rs 97.50/- is the distance between two of those prices, and a distance is the cost of moving rather than a fourth kind of figure. No PREMIUM figure appears above, because a premium is not how anybody enters either of the arrangements worked out so far. PAYOFF names whatever the position delivers when its contract finishes, counted before anything spent to reach that day. PROFIT is that same figure with the spending taken out.

Payoff and profit walk to the same figure on a futures position, and they do it because nothing was handed over to open one. So nothing waits at the end to be subtracted, and the two words land on one number. Where something is paid to get in, they part company, and a reader who has only ever watched them coincide will subtract nothing on the day subtracting was the whole job.

Quantities need the same care. One unit multiplied by the Rs 2,130.00/- the later contract was opened at gives Rs 2,130.00/- of notionalA size used purely as a multiplier when working out what moves. No amount of that size is ever handed across by anybody.. Not a rupee of that money has left anybody's hands. The same unit multiplied by the spot price of Rs 2,000.00/- instead gives Rs 2,000.00/- of exposure. Exposure is the figure the collateral is struck against. The quantity riding on one contract is fixed by the exchange rather than chosen by a party, under SEBI at sebi.gov.in. The arithmetic above therefore runs on one unit and leaves the multiplication to whoever knows the real amount.

Whose rules fill in the open rows?

Everything above is arithmetic and it does not date. Every practical detail around the arithmetic does date. The jurisdictionThe territory whose rules a matter falls under, which decides who gets to answer a question about it. those details belong to is India, so they belong to SEBI. Six of them arise here.

One figure sits above that drawing rather than inside it, and the difference is worth being clear about. The 8.0 per cent behind the Rs 160.00/- is a teaching number, put there so the collateral arithmetic had something to work on, and it belongs to no rulebook. The rows in the drawing are the opposite: they are real requirements with real values, and those values are held by the authorities named inside them.

The rows are drawn empty rather than left out. A requirement nobody ever sees written down is the more expensive kind of ignorance, because the reader does not know the question is there to ask. Drawing the row and leaving it open teaches two things at once: that the requirement is real, and that its value is somewhere else. A conventionA way of doing something that everybody in a market has agreed to follow, settled by practice rather than worked out from arithmetic. also hides inside several of those rows. Practice differs between contracts even where the rule above them is one rule, so a figure lifted from one contract would not carry across to another.

What starts on the desk the week the near contract runs short?

Four separate pieces of work begin, in different places and with different people. A roll is four jobs rather than one order.

The first is a piece of arithmetic done before any order goes anywhere. Somebody prices what nine further months costs on the exposure being carried, off the current spot price and the current financing rate, and the answer is a rupee amount that has to be recognised as a cost rather than discovered afterwards on a statement. Everything established so far is that first job.

The second is a lodgement problem. The collateral against the position being opened has to be found while the collateral against the position being closed is still tied up, so for a stretch of the day two lodgements are live at once even though only one exposure is intended. Anybody planning a move on real size has to have arranged for that overlap in advance. How long the overlap lasts depends on how long money takes to travel after a trade, one of the six open rows above.

The third is a bookkeeping correction, and it is the one that gets missed. The price the position is measured against has to be changed on the sheet, on the day, from the price in the contract that closed to the price the new contract opened at. Miss it and every figure the sheet produces afterwards is a subtraction across two different contracts, the struck line in the drawing further up.

The fourth is a check against a ceiling that is not in these notes. SEBI settles the ceiling on a single holding, at sebi.gov.in, and a position that fitted comfortably inside the contract being left has to fit inside the contract being entered as well. The ceiling question has a real answer, and SEBI is where the answer lives.

Notice what is common to all four. Not one of them requires anybody to hold a view about the reference asset. A roll is a piece of maintenance on an exposure that already exists, and the arithmetic behind it would look exactly the same whoever was doing it and whatever they thought.

Is the cost of a roll a reason to pay it?

Rs 97.50/- has just left a position simply to keep it alive, and the next question forms itself without help: is that worth paying? The arithmetic cannot answer that one.

An answer needs four pieces of information. What was the position opened to do? How much longer does that reason run? Which other exposures already sit against it? And what does the authority permit during the days when positions travel across? All four belong to the party holding the position and to nobody else. The carry calculation has also never been run against a stretch of real months, so it says nothing about how often a roll turns out to have been worth its cost. Working out the price of something is a long way from deciding to buy it.

A reading skill is worth taking away, rather than an instruction. A roll cost met anywhere can afterwards be read for what it is made of, for which two inputs set its size, and for what it does not contain. Because a reading skill survives every change in the numbers, it outlasts any recommendation.

The reference asset, the forward buyer and the forward seller are invented.
Educational material. Not advice on any investment, tax, budget or market position.