Educational Reference
Calendar and Diagonal Spreads: Trading Time Instead of Direction
A calendar spread sells a near-dated option and buys a longer-dated one at the same strike. The near leg loses its time value faster than the far leg, which is the usual case, so the structure is routinely described as a way to harvest time. That description is true and incomplete, and the incompleteness is the expensive part. The same position is long volatility on the far leg and short volatility on the near leg, which means it can lose money while the underlying stands perfectly still. This page computes both halves and publishes the numbers, including the ones that undercut the tidy version.
The finding, stated first. On the illustrative structure computed here, thirty days of undisturbed time decay were worth 14,808 rupees per contract. A two point fall in implied volatility, with the index motionless, removed 1,634 rupees of that. A shift where near-dated implied volatility rose three points while far-dated rose one removed 3,261 rupees, and undid nineteen days of decay, even though average implied volatility had gone up. All figures illustrative and simulated.
Two different instruments share one name
Before anything can be computed, a piece of vocabulary has to be untangled, because the phrase "calendar spread" is used for two structures that have almost nothing to do with each other beyond the observation that both involve two expiry dates.
The first is the futures calendar spread. A trader holding a position in an expiring futures contract who wants to keep the exposure sells the near month and buys the next one, usually as a single spread order. The price difference between the two is the basis, and it is set mechanically by the cost of carry, meaning financing minus expected dividends. That structure is a position on the basis. It has no meaningful exposure to volatility, its two legs move almost one for one with the underlying and therefore cancel, and its economics are a question of interest rates and dividend timing. Our guide to rollover in futures covers that instrument, the recurring toll it charges, and how to read published rollover data, and it is the right page if the roll is what brought you here.
The second is the options calendar spread, and it is what this page is about. Here the two legs are options rather than futures, they share a strike, and they differ only in expiry. One is sold, one is bought. Nothing about the basis matters. What matters instead is that an option's premium contains time value, that time value decays towards zero as expiry approaches, and that it does not decay at a constant rate. A thirty-day option and a sixty-day option lose their time value at very different speeds over the same thirty days, and the whole structure exists to stand between those two speeds.
Two further distinctions are worth drawing while the vocabulary is being tidied, because both appear in the same conversations. A vertical spread has two legs at different strikes in the same expiry, so it is a position on direction with a cap on both ends. A calendar spread has two legs at the same strike in different expiries, so direction is close to neutral at inception and time is the axis of the trade. A diagonal spread changes both at once, and the last third of this page shows exactly what that does to the numbers.
The confusion between the futures version and the options version is not merely academic. Someone who has read that a calendar spread is a low-risk way to maintain an existing position, which is a reasonable description of a futures roll, and who then applies that intuition to a two-expiry options structure, has imported a mental model from an instrument whose principal risk is a financing rate into one whose principal risk is the shape of a volatility curve. The two are not related. Everything below concerns the options structure only.
The structure, stated in full so it can be checked
Every number on this page comes from one deterministic model of one structure, with the inputs stated openly so the arithmetic can be reproduced or disputed. The underlying is an illustrative index rather than a named instrument, which keeps the exercise reproducible and avoids implying that any particular Indian index would have produced any particular outcome.
| Element | Setting | Why it is set this way |
|---|---|---|
| Underlying | An illustrative index at 24,000 | A round level makes the at-the-money case legible and keeps the arithmetic checkable by hand |
| Short leg | 24,000 call, 30 days to expiry, sold at 410.23 | At the money, so the premium is entirely time value and the decay effect is not contaminated by intrinsic value |
| Long leg | 24,000 call, 60 days to expiry, bought at 650.21 | Same strike, one expiry further out. Same strike is what makes this a calendar rather than a diagonal |
| Net cost | 239.98 per unit, a debit | The far leg costs more than the near leg, so the position is paid for rather than received. That amount is the maximum loss |
| Implied volatility | 13.0 percent near, 14.0 percent far | An upward-sloping term structure, the ordinary shape in a calm market. One volatility per expiry, no strike skew |
| Rates | Financing 6.5 percent, dividend yield 1.2 percent | Illustrative model inputs, stated rather than hidden. Both are continuous and both affect the far leg more than the near |
| Contract quantity | 75 units, illustrative | Gives a contract worth 18,00,000 rupees, inside the 15 to 20 lakh rupee minimum contract value band. Exchange lots are revised periodically and no lot should be quoted as a present fact |
| Pricing model | Black-Scholes, European, no early exercise | Index options in India are European, so the model matches the instrument. The point of the page is the relationships, not the last decimal |
The contract quantity needs a word of caution, because it is the input most likely to go stale. The minimum contract value for index derivatives was moved into a band of 15 to 20 lakh rupees under the October 2024 framework, and exchanges revise lot sizes periodically to keep contract value inside that band as index levels move. Lots were raised across the board in November 2024, when the headline index lot went from 25 to 75, and revised downwards again from January 2026 as indices climbed. Quoting any specific lot as a current fact therefore goes stale within months, and our note on lot sizes in the derivatives segment explains the mechanism. The 75 units used here is a modelling convenience chosen because it puts the contract inside the band, and nothing more.
One simplification is worth naming rather than burying. The model uses a single implied volatility for each expiry and no variation across strikes. Real option surfaces are not flat across strikes, and a downward-sloping call skew would make the sold leg in the diagonal section slightly cheaper than modelled. That would shift the diagonal numbers a little without changing their direction, and stating the assumption is better than quietly absorbing it.
What the position is worth when the near leg expires
The first thing to compute is what the whole structure is worth at the moment the near leg expires, because that is the natural horizon of the trade and because the answer is not the shape most people expect.
With an ordinary single-expiry structure, the diagram at expiry is a set of straight lines with kinks at the strikes, since every leg has resolved into intrinsic value. That is not what happens here. When the near leg dies, the far leg still has thirty days to live and therefore still carries time value. The position is being marked, not settled. The profile is the far leg's remaining value minus whatever the expired near leg was worth, which is a smooth curve with a single peak rather than a hockey stick.
The peak sits at the strike. If the index finishes at exactly 24,000, the expired near leg is worth nothing, the far leg is worth 437.42, and the position has gained 197.44 per unit, which is 14,808 rupees on a seventy five unit contract and 82.3 percent of what was paid for it. That is the best case, it requires the index to do nothing at all for thirty days, and it is the number the structure is usually sold on.
The breakevens are the more informative pair of numbers. They sit at 23,573 and 24,587, which is a fall of 1.78 percent or a rise of 2.45 percent from the starting level. The profitable band is 1,014 points wide, or 4.23 percent of the index. Stated that way it sounds generous. Stated against the movement the near leg is itself priced for, it does not.
Here is the comparison that matters. The near leg was sold at an implied volatility of 13 percent. Over thirty days that implies a one standard deviation move of 3.73 percent, or 894 points. The profitable band, at 1,014 points, is about 57 percent of a two standard deviation span. In other words the structure needs the index to move less than the option market itself expects it to move, and it needs that for a full month. That is a coherent position to hold, but it is a view, not a mechanical harvest, and the difference is worth being clear about before any money is involved.
Running the same structure across 200,000 simulated thirty-day paths at a realised volatility of 13 percent, which is exactly the level the near leg was priced at, the index finished inside the profitable band 42.6 percent of the time. The mean outcome across all paths was a loss of 1,546 rupees per contract and the median was a loss of 2,104 rupees. The fifth percentile was a loss of 15,670 rupees and the ninety fifth was a gain of 12,782 rupees. Raise realised volatility to 16 percent and the share of paths finishing inside the band falls to 35.4 percent. At 19 percent it falls to 30.1 percent. Those are properties of a peaked profile meeting a distribution wider than itself, and they are invisible in the diagram, which is precisely why the diagram alone is a poor basis for judging the structure.
The two tails are not symmetrical, and the asymmetry is worth a sentence. On the downside the loss reaches the full debit, 17,999 rupees, and it is entirely gone below 21,690, because the far leg's remaining time value collapses to nothing. On the upside the loss stops short of the full amount. At 30,000 the position has lost 10,626 rupees rather than the whole debit, because the far leg's strike now sits far below the index and it retains intrinsic value against the near leg that expired against it. A structure whose worst case is capped is easier to size, and this one is capped in both directions but deeper in one of them.
The decay differential, modelled day by day
The engine of the structure is that two options with the same strike and different expiries lose their time value at different rates. That statement is easy to make and rarely quantified, so here it is quantified.
Pin the index at the strike so that nothing but time moves, and track both legs every day for thirty days.
Over the thirty days the near leg gives up its entire premium of 410.23. The far leg, over the identical period, gives up 212.80, falling from 650.21 to 437.42. The near leg loses 1.93 times as much as the far leg, and that difference, 197.44 per unit, is the whole gain available to the structure if the index cooperates completely.
The rates diverge because time decay is not linear. An option's time value falls roughly with the square root of the time remaining, which means the rate of loss accelerates as expiry approaches and accelerates hardest at the very end. In the first week the near leg shed about 8.1 points a day. In its final week it shed about 26.4 points a day, a rate 3.2 times higher. The far leg, which never gets close to its own expiry during this window, went from 6.5 points a day to 7.9, an acceleration of only 1.2 times. The whole structure is a bet on that gap between 3.2 and 1.2.
The practical consequence is severe and is the part most descriptions omit. Because the near leg's decay is concentrated at the end, so is the structure's profit. The final seven days delivered 65.7 percent of the entire thirty-day gain. The final five days delivered 58.1 percent. The first fifteen days, which is half the life of the position, delivered 16.1 percent. Netted across both legs, a day in the first week was worth about 126 rupees on the illustrative contract.
Three things follow. The first is that a calendar spread held for two weeks and then closed has collected a small fraction of what the diagram promises, while having been exposed to every adverse move in the meantime. The second is that the position is at its most fragile exactly when it is at its most valuable, because the days that carry the profit are the days when the near leg's sensitivity to a move in the underlying is also at its most violent. Our page on expiry day and how time value collapses into it covers that acceleration in its own right. The third is the reason the next section exists: if two thirds of the reward is confined to a single week, then anything that can move the position by more than a week's worth of decay in a single session deserves more attention than the decay itself.
The volatility exposure the time story hides
A calendar spread is described as a time trade. It is also, unavoidably and simultaneously, a volatility trade, and the volatility exposure is larger on a day-to-day basis than the time exposure it is supposed to be built for.
The mechanism is simple. An option's sensitivity to implied volatility grows with the time remaining, because a longer-dated option has more time in which a change in the expected pace of movement can matter. The sixty-day leg therefore has more volatility sensitivity than the thirty-day leg. Since the far leg is bought and the near leg is sold, the position is net long volatility. Our explainer on what implied volatility is and how it is solved for is the pillar for that idea, and the arithmetic below is its consequence in this particular structure.
In the illustrative position the far leg moves 38.11 per volatility point and the near leg moves 27.17, so the net is 10.94 per unit, or 820 rupees per contract for every one point change in implied volatility. Hold the index at 24,000, let no time pass at all, and move only the volatility inputs.
A two point parallel rise in implied volatility adds 1,645 rupees. A two point parallel fall removes 1,634 rupees. Set that against the time engine: seven full days of undisturbed decay were worth 879 rupees. A two point move in implied volatility, which can happen in a single session, is worth almost twice a week of the thing the structure is nominally built to collect. Recovering a two point fall takes twelve days of decay. That is the first uncomfortable result, and it is only the parallel case.
The parallel case is also the unrealistic one. Implied volatility does not move by the same amount at every expiry, and the pattern of how it moves differently is the whole substance of the position. Near-dated implied volatility is more reactive than far-dated implied volatility, because a surprise that is expected to resolve quickly raises the price of near-dated protection much more than it raises the price of protection two months out. That is why the curve tends to steepen in calm markets and flatten or invert when something unsettling happens.
Model that directly. Let near-dated implied volatility rise three points while far-dated rises one. Average implied volatility across the two expiries has gone up by two points, the index has not moved, and no time has passed. The position loses 3,261 rupees. Push the near leg to plus four while the far stays at plus one and the loss becomes 5,304 rupees, and it takes nineteen days of decay to recover the first of those.
Now run it the other way. Let near-dated implied volatility fall three points while far-dated falls one. Average implied volatility has gone down by two points, which is supposedly bad for a net long volatility position, and the structure gains 3,246 rupees. The direction of the volatility move told you nothing. What mattered was the change in the relationship between the two expiries.
The rule the model produces is clean enough to remember. For the position to hold its value, the far leg's implied volatility has to move at least 0.71 points for every one point the near leg's moves. If the near leg rises three points, the far leg needs 2.14 points just to leave the position flat. Any shift in which the front of the curve outruns the back is a loss, whatever the average is doing.
| What changes | Effect on the position | What it tells you |
|---|---|---|
| Seven days pass, nothing else | plus 879 | The engine the structure is built for, and the slowest-moving of the three |
| Index rises 1.5 percent | minus 232 | Near neutral at inception. Net delta is 0.018, so small moves barely register |
| Index falls 1.5 percent | minus 1,225 | The direction exposure is not symmetric, because the far leg's value falls faster below the strike |
| Implied volatility rises 2 points at both expiries | plus 1,645 | Net long volatility. Almost twice a week of decay, available in one session |
| Implied volatility falls 2 points at both expiries | minus 1,634 | Twelve days of decay undone with the index motionless |
| Near rises 3 points, far rises 1 | minus 3,261 | The case that surprises people. Average volatility rose and the position still lost |
| Near falls 3 points, far falls 1 | plus 3,246 | Average volatility fell and the position gained. The level is not the exposure |
Read that table as a ranking and the argument of the page falls out of it. The largest single-variable move is not time and it is not direction. It is a change in the shape of the volatility curve, and it is roughly four times a week of decay. A structure introduced as a way to harvest time turns out, on any horizon shorter than the last week of the near leg's life, to be dominated by something that is not time at all.
This also explains a common and puzzling experience. A trader puts on a calendar spread ahead of a known event, reasoning that elevated near-dated implied volatility means the sold leg is expensive and therefore the structure is well priced. That reasoning is sound as far as it goes. What it omits is that the elevated near-dated volatility is the thing most likely to collapse when the event resolves, and a collapse concentrated in the near leg is precisely the shift that helps the position, while the run-up to the event, where near-dated volatility climbs faster than far-dated, is precisely the shift that hurts it. The structure is therefore exposed to the event twice, in opposite directions, at different times, and the diagram shows neither.
Change the strike as well and direction walks in
A diagonal spread differs from a calendar spread in exactly one respect: the two legs no longer share a strike. That single change converts a position that was close to direction-neutral into one that is not, and the size of the conversion is easy to underestimate until it is computed.
Take the same long leg, the sixty-day 24,000 call, and instead of selling the thirty-day 24,000 call, sell the thirty-day 24,400 call. Everything else is held identical.
Three things change at once. The first is cost. The sold leg is now out of the money and therefore cheaper, at 230.79 against 410.23, so less is received and the net cost rises from 239.98 to 419.42 per unit. That is 31,457 rupees against 17,999, an increase of 74.8 percent. Since the debit is the maximum loss, the worst case has deepened by the same proportion.
The second is direction. Net delta at inception moves from 0.018 to 0.192, roughly eleven times as much. The calendar was very nearly indifferent to a small move in the index. The diagonal is not indifferent at all, and anyone holding one is holding a directional position whether or not they describe it that way.
The third is where the position wants the index to finish. The calendar peaks at 24,000, which is where the index already is, so its best case is that nothing happens. The diagonal peaks at 24,400, which requires a rise of 1.67 percent, and at that point it is worth 20,241 rupees against the calendar's best of 14,808. The diagonal offers more at its peak, and it demands a correct directional call to get there.
The scenario strip at the foot of the figure is the clearest way to see the trade. If the index does not move, the calendar earns 14,808 and the diagonal earns 1,350. If the index rises 1.5 percent, the calendar earns 4,606 and the diagonal earns 18,148. If the index falls 3 percent, the calendar loses 6,986 and the diagonal loses 20,444. The two structures are not versions of the same idea at different settings. They are answers to different questions, and the diagonal's answer requires an opinion about direction that the calendar does not.
One asymmetry in the diagonal deserves attention because it is genuinely useful and rarely stated. The calendar has two breakevens and loses money outside both. The diagonal, in this configuration, has only one. Its lower breakeven is at 23,967, and above that level it does not cross back below zero however far the index rises. At 30,000 the diagonal is still worth 5,916 while the calendar has lost 10,626. The reason is structural: the long leg's strike sits below the short leg's strike, so in a large rise the long leg's intrinsic value exceeds the short leg's by the width between them, and that width is a floor. The price of that floor is the higher debit, and the arithmetic of paying for caps and floors is the subject of our page on vertical spreads and what a cap actually costs.
The margin benefit that stopped applying on expiry day
Everything above is about the value of the position. What it costs to hold is a separate question, and one specific rule change altered it materially for these structures.
Margin on derivatives positions is not computed leg by leg in isolation. Where two positions partly offset each other, the clearing system recognises the offset and requires less margin than the sum of the two standalone requirements. A calendar spread is an obvious candidate: a short option in one expiry and a long option at the same strike in another expiry do genuinely hedge one another over most outcomes, and the margin framework reflected that with a calendar spread benefit.
SEBI's October 2024 derivatives framework changed this for one specific day. Circular SEBI/HO/MRD/TPD-1/P/CIR/2024/132, dated 1 October 2024, removed the calendar spread margin benefit for contracts expiring on the day, with effect from 1 February 2025. On the near leg's expiry date, the offset between the two expiries is no longer recognised, and the legs are margined without it. The same framework added two percent of additional extra loss margin on short option positions expiring that day, effective 20 November 2024, and raised the minimum contract value into the 15 to 20 lakh rupee band from the same date.
The regulator's stated reasoning concerned expiry-day behaviour generally, noting that expiry-day options were being held for average periods measured in minutes with no discernable benefit towards sustained capital formation. Whatever view one takes of that, the consequence for a two-expiry structure is arithmetically direct, and it lands in an awkward place.
Recall the decay computation. Two thirds of the structure's entire thirty-day gain arrives in the final seven days, and 58.1 percent arrives in the final five. The position is therefore worth the most, and is closest to realising it, on precisely the sessions when the capital required to keep it open steps up. A structure that was margined as a hedged pair for twenty nine days is margined without that recognition on the thirtieth, on the day its remaining value is concentrated.
Three practical consequences follow, and none of them requires a specific margin figure to state. The first is that carrying a calendar spread into the final session is a different capital commitment from carrying it the day before, and the difference is a rule, not a market condition, so it is entirely predictable and can be planned for. The second is that closing the near leg before its expiry date, which the charge model later on this page assumes, is not merely a convenience: it sidesteps a step-up in requirement that arrives on a known date. The third is that the change alters what the structure is worth relative to the capital it ties up, which is a different question from what it is worth relative to what was paid for it, and the second question is the one the payoff diagram answers.
What cannot honestly be put on this page is a rupee margin figure. Margin under the exchange's risk-array system depends on parameters that are set daily and vary by underlying, and no primary source was available to fix a defensible number. The mechanism and the effective date are verified and dated. The size of the step-up on any given day is not, and inventing one to make the section tidier would defeat the purpose of computing anything.
Five chargeable orders, and the cost that is not a charge
A calendar spread is two legs, and a calendar spread that is rolled is more than two. Running the illustrative structure through entry, one roll of the short leg, and a final exit is five separate chargeable orders.
The rates are taken from the primary instruments. Securities transaction tax is 0.15 percent of the premium on an option sale, a rate raised in stages and most recently set by section 159 of the Finance Act 2026 with effect from 1 April 2026, and it falls on the seller only. Stamp duty is 0.003 percent of the premium on the buy side only, under Schedule I Article 56A(d) of the Indian Stamp Act 1899. The exchange transaction charge is modelled at 3,250 rupees per crore of premium turnover, the rate notified for index options by an exchange notice of 27 September 2024 with effect from 1 October 2024. Goods and services tax at 18 percent applies to brokerage and to the exchange charge, and not to the transaction tax or the stamp duty, which pass through under the pure agent route. Brokerage is an illustrative flat 20 rupees per order.
One rate could not be verified and is therefore omitted rather than guessed. The regulator's turnover fee is verified at 10 rupees per crore for the cash segment under regulation 41(1) of the SEBI (Stock Brokers) Regulations, 2026, but no primary source was available for the corresponding derivatives-segment figure. Leaving it out means the totals below understate the true cost by a small amount, and saying so is better than filling the gap with a plausible-looking number.
The result is the inconvenient finding running in the opposite direction to the usual one. All five orders come to 286.07 rupees. Against a debit of 17,999 rupees that is 1.59 percent, and against the best case of 14,808 rupees it is 1.93 percent. Securities transaction tax is the largest statutory component at 116.14 rupees, but the illustrative flat brokerage at 100 rupees across five orders is not far behind it. For a two-leg structure of this size, the statutory stack is not what makes the position difficult. Multi-leg structures with four legs and frequent adjustment are a different arithmetic, which our page on the iron condor and its eight chargeable events works through, and the contrast between the two is instructive.
What is larger here is the cost that never appears on a contract note. Longer-dated options are less actively traded than near-dated ones, so the gap between bid and ask on the far leg is wider than on the near leg. That gap is paid twice, once entering and once exiting, and it is paid on the leg that is hardest to trade. At an illustrative two points of spread crossed on each of those two far-leg orders, the cost is 300 rupees on a seventy five unit contract, which on its own exceeds the entire statutory stack of 286.07. That figure is an assumption rather than a verified market rate, and it is offered as a sensitivity, not a measurement. The point survives the uncertainty: for this structure the liquidity of the far leg is a larger consideration than the tax on it.
How the structure fails
Every structure has a set of ways it goes wrong, and the useful version of that list is the one where each entry is traced to a number rather than a caution.
| Failure mode | What it costs in the model | Why the payoff diagram does not show it |
|---|---|---|
| The underlying travels | The full debit of 17,999 below 21,690, and 10,626 at 30,000 | The diagram does show this, but it says nothing about how likely the move is. At the near leg's own implied volatility the index finished outside the band 57.4 percent of the time |
| The volatility curve steepens | 3,261 for near plus three against far plus one, with the index still | The diagram fixes volatility. Every point on it assumes the curve has not moved, which is the assumption most likely to fail |
| Implied volatility falls across the board | 1,634 for a two point parallel fall, undoing twelve days of decay | Net long volatility is invisible in a payoff line, which is drawn at a single volatility |
| Closing early | The first fifteen days deliver 16.1 percent of the gain | The diagram is drawn at one date. It has nothing to say about the path to that date |
| Rolling the short leg repeatedly | Each roll is two more chargeable orders and two more crossings of the spread | A payoff diagram is drawn for one cycle, and a rolled position is many |
| The far leg is thinly traded | An illustrative two point spread crossed twice is 300 rupees, above the whole statutory stack | Every diagram is drawn at mid prices, which is a price nobody transacts at |
| Holding into the near leg's expiry date | The calendar spread margin benefit no longer applies on that day, from 1 February 2025 | The diagram measures value against premium paid, never against capital tied up |
| Assignment on the short leg | Not modelled here. European index options remove it; single-stock options do not | The diagram assumes the position is held intact to the horizon it is drawn for |
The pattern across that table is worth naming. Almost every failure mode is something the payoff diagram structurally cannot display, because a payoff diagram is a single slice: one date, one volatility surface, one set of mid prices. It is a photograph of an assumption. The structure's real behaviour lives in the dimensions the photograph flattens, which is why a page about calendar spreads that consists of a diagram and a paragraph has communicated the least interesting part of the subject.
What the computation is actually for
None of this establishes that a calendar spread is a good structure or a bad one. It establishes what the structure is a position in, which is a prior question and a more useful one.
The honest description is that a calendar spread is three positions welded together. It is long the difference in decay rates between two expiries, which is the part that gets described. It is long the far expiry's implied volatility and short the near expiry's, which is the part that dominates on any short horizon and which cost 3,261 rupees in a single modelled shift where the underlying did not move at all. And it is very slightly long direction, which becomes substantially long direction the moment the strikes are separated into a diagonal. Anyone holding the structure holds all three, whether or not all three were intended.
That framing changes what the relevant question is. It is not whether time decay works, because the computation shows plainly that it does: 197.44 per unit over thirty days with the index pinned. It is whether the holder has a view on the shape of the volatility curve between two expiries, since that view is embedded in the position regardless. A structure entered for a reason it does not primarily express is a structure whose outcome will feel random, because the variable actually driving it was never being watched.
This matters more in the Indian derivatives market than the arithmetic alone suggests. SEBI found that about 93% of individual traders in equity derivatives made net losses over FY22 to FY24, with aggregate net losses exceeding Rs 1.8 lakh crore (SEBI, September 2024). Multi-leg structures are frequently presented as the sophisticated alternative to the outright positions that produced that record. The computation on this page suggests a more careful reading: a structure with more legs has more ways to be right and correspondingly more ways to be wrong, and the additional exposures do not announce themselves. A two-leg position with a hidden term-structure exposure is not obviously safer than a one-leg position with an obvious directional one. It is differently exposed, and the difference is only visible to someone who has computed it.
The general habit is the transferable part. Take any structure, hold every variable still except one, move that one, and write down what happens in rupees. Do it for time, for direction, and for volatility separately, and then do it again for volatility where the two expiries move by different amounts. The exercise takes an afternoon, it requires nothing more advanced than an option pricing formula, and it reliably surfaces at least one exposure that was not part of the reason the position was put on. That habit, applied before capital is committed rather than after, is a large part of what separates a considered position from a hopeful one, and it is the method we teach.
FAQ
Frequently asked questions
Is a calendar spread in options the same as a calendar spread in futures?
No, and the shared name causes real confusion. A futures calendar spread buys one expiry of a future and sells another expiry of the same future, so it is a position on the basis between the two, which is set by financing against dividends. An options calendar spread sells a near-dated option and buys a longer-dated option at the same strike, so it is a position on the difference between how fast the two premiums decay and on the shape of the implied volatility curve across expiries. The instruments differ, the risks differ, and only the word is shared.
What actually makes a calendar spread profitable?
Two things at once, and only one of them is usually mentioned. The near leg loses its time value faster than the far leg, which works in the holder's favour if the underlying stays near the strike. The far leg also carries more volatility exposure than the near leg, so the position gains when implied volatility rises and loses when it falls. In the illustrative structure computed on this page the whole thirty-day decay differential was worth 14,808 rupees per contract, and a two point fall in implied volatility on its own removed 1,634 of that.
Can a calendar spread lose money if the underlying does not move at all?
Yes, and this is the case that surprises people. The position is long volatility on the far leg and short volatility on the near leg, so what matters is not the level of implied volatility but the relationship between the two expiries. In the illustrative model, near-dated implied volatility rising three points while far-dated rose only one produced a loss of 3,261 rupees per contract with the index perfectly still, even though average implied volatility across the two expiries had gone up by two points.
How much of the profit arrives in the final days?
Most of it. In the illustrative computation, with the index pinned at the strike, the final seven days before the near leg expired delivered 65.7 percent of the entire thirty-day gain and the final five days delivered 58.1 percent. The first fifteen days, half the life of the position, delivered 16.1 percent. The near leg's daily decay ran at about 8.1 points a day in its first week and about 26.4 points a day in its last, a rate roughly 3.2 times higher.
What is the difference between a calendar spread and a diagonal spread?
A calendar spread changes only the expiry between the two legs and keeps the strike the same. A diagonal spread changes both. Changing the strike introduces a directional component that the calendar does not have. In the illustrative comparison, moving the short strike 400 points higher lifted net delta from 0.018 to 0.192, roughly eleven times as much, raised the cost of the position by 74.8 percent, and moved the point of maximum value from an unchanged index to one that had risen 1.67 percent.
What did SEBI change about calendar spread margins on expiry day?
The October 2024 derivatives framework, circular SEBI/HO/MRD/TPD-1/P/CIR/2024/132 dated 1 October 2024, removed the calendar spread margin benefit for contracts expiring on the day, with effect from 1 February 2025. The offset previously recognised the partial hedge between two expiries and reduced the margin required. On the expiry day of the near leg that recognition no longer applies, so the legs are margined without the offset. The same framework added two percent of extra loss margin on short option positions expiring that day, effective 20 November 2024.
How wide is the profitable range on a calendar spread?
Narrower than most people expect, and the honest way to see it is against the move the near leg is itself priced for. In the illustrative structure the breakevens sat at 23,573 and 24,587, a band 1,014 points wide, or 4.23 percent of an index at 24,000. A one standard deviation move over the same thirty days at the near leg's own implied volatility of 13 percent was 894 points. The profitable band was therefore about 57 percent of a two standard deviation span, which is why it is worth computing rather than assuming.
Do the charges make a calendar spread unviable at retail size?
On the illustrative arithmetic, no, and that is the inconvenient answer in the other direction. Running the structure through entry, one roll and exit is five chargeable orders, and the statutory stack plus an illustrative flat brokerage came to 286.07 rupees against a debit of 17,999 rupees, about 1.59 percent. What is larger is the cost that is not a charge. Crossing an illustrative two point bid and ask spread on the far leg at entry and again at exit costs 300 rupees on a seventy five unit contract, more than the entire statutory stack.
Why does the value profile curve instead of forming a hockey stick?
Because only one of the two legs has expired. At the moment the near leg dies it is worth its intrinsic value and nothing more, but the far leg still has thirty days of life and therefore still has time value. The position is being marked, not settled, so the profile is the far leg's remaining value minus the near leg's intrinsic value, and that is a smooth curve with a peak at the strike rather than a set of straight lines with a kink.
What is the worst case on a calendar spread?
The loss is limited to the net amount paid, which in the illustrative structure was 17,999 rupees per contract, and it is reached when the underlying travels far enough that the far leg's remaining time value collapses. The two tails are not symmetrical. In the model the entire debit was gone below 21,690, while a rise to 30,000 produced a loss of 10,626 rupees rather than the full amount, because the far leg's strike sits below the level and it retains intrinsic value against the expired near leg.
Method note
How the numbers on this page were produced
Every figure comes from a single deterministic model, written in Python with numpy only and seeded so that it reproduces identically on each run. Options are priced with the Black-Scholes formula for a European call on an underlying paying a continuous dividend yield, using the inputs listed in the setup table. The normal distribution function is built from the error function rather than an external library, so the model has no dependency that could change its output between runs. Breakevens are located by bisection on the computed value curve rather than read off a chart.
The decay computation holds the underlying at the strike and reprices both legs once per day for thirty days, so the only variable moving is time. The volatility computation holds the underlying and the clock fixed and moves only the implied volatility inputs, separately for each expiry, which is what allows the term-structure cases to be isolated. The distribution of outcomes uses 200,000 lognormal paths at a stated realised volatility, with zero drift in log terms; that is a modelling assumption, not a forecast, and a different drift would move the numbers.
Charge rates are taken from the primary instruments cited in the charge section. Brokerage is an illustrative flat amount and is labelled as such. The derivatives-segment regulator turnover fee could not be verified from a primary source and is omitted, so the cost totals understate the real figure slightly. No margin figure is quoted anywhere on this page, because margin depends on risk parameters set daily that could not be verified; only the rule change and its dates are stated. The model uses one implied volatility per expiry with no variation across strikes.
All results are illustrative and simulated. They are not a track record, they are not a forecast, and they are not an indication of what any structure would produce in a live account. Nothing here is a recommendation to enter, hold or avoid any position. The purpose of the exercise is to show the relationships between time, direction and volatility inside a two-expiry structure, which are properties of the instrument rather than of any particular market.
Related