Forward Volatility
The volatility of a window that has not started yet, hidden inside two that have prices.
Quick Answer
Forward volatility isolates the movement priced for a future window — say the month starting 30 days out — after stripping the volatility of the intervening period. You extract it because variance, not volatility, adds across time, so you difference two spot variances and re-annualise. It is what a forward-starting or calendar position is really exposed to.
Definition of Forward Volatility
Forward Volatility is the volatility implied for a future window between two dates, extracted from the term structure so that variance stays additive across the two periods; quoted annualised.
Key takeaways on Forward Volatility
- Forward volatility is the volatility the market implies for a future window between two dates, extracted from the two spot implied volatilities that straddle it.
- Variance is additive in time and volatility is not, so you must work in variance: subtract σ₁²T₁ from σ₂²T₂, divide by the window, take the square root.
- On an upward-sloping term structure the forward volatility sits above both spot numbers — a one-point slope from 12.9% to 13.9% implies about 14.8% for the second month.
- A negative forward variance is not a market state but an arbitrage flag: one of the quotes is wrong, or the surface admits a calendar arbitrage.
- A calendar spread is fundamentally long or short forward volatility, and the gap between the forward volatility you trade and the volatility the window realises is the whole economics of the position.
Forward volatility is where the term structure stops being a list of prices and becomes a set of bets on specific future windows. The discipline it forces — always convert to variance, difference, then convert back — is the difference between a trader who thinks a calendar spread is about time decay and one who knows it is about the volatility of a month that has not started yet. Read the spot curve and you see averages. Extract the forward volatilities and you see what the market is actually charging for each stretch of the future, including the one your position will still be alive to feel.
Forward Volatility at a glance
| Attribute | Value |
|---|---|
| Section | Core Volatility |
| Measures | Volatility of a future window |
| Direction | Forward-looking |
| Units | Annualised % |
| Level | Advanced |
Forward Volatility in simple words
Forward volatility answers a question the option screen does not display directly: not "how volatile is the next 60 days", but "how volatile is the second 30 days, the stretch from day 30 to day 60". You cannot read it off any single quote, because no option covers only that middle window. Instead you back it out of the two you can see — the 30-day implied volatility and the 60-day implied volatility. The trick that makes it possible is that variance adds up over time, so the variance of the 60-day period minus the variance of the first 30 days leaves the variance of the second 30 days, and its square root is the forward volatility.
Here is the part that surprises people. If the 30-day implied volatility is 12.9% and the 60-day is 13.9% — a rise of just one point — the forward volatility for the window between them is not 13.9%, and not somewhere in between. It is about 14.8%, higher than either. A gently upward-sloping term structure hides a forward window that is materially more expensive than either number you started with, because the later month has to carry all the extra variance the longer option is charging for, packed into a shorter time.
How a forward volatility sits above the spot curve it came from
Variance adds, so the forward window is the steeper one
Spot implied volatilities for 30 and 60 days, and the forward 30-to-60-day volatility implied by them.
Forward Volatility: professional explanation
Variance is additive in time; volatility is not
This single fact is the whole subject. Under the standard assumption of independent returns, variances over consecutive non-overlapping periods add: the variance of a two-month window is the variance of the first month plus the variance of the second. Volatilities do not add, because volatility is the square root of variance, and square roots do not add. So to isolate the second month you must work in variance — total variance to 60 days minus total variance to 30 days — and only convert back to a volatility at the very end. Anyone who tries to reason about forward volatility in volatility units rather than variance units gets it wrong, usually by assuming the forward number sits between the two spot numbers, which it almost never does.
The formula, and why the forward number exceeds the spot
Total variance to expiry is σ²·T, so the variance of the forward window between T1 and T2 is σ₂²T₂ − σ₁²T₁, and the forward volatility is the square root of that divided by the length of the window: σ_fwd = √((σ₂²T₂ − σ₁²T₁) / (T₂ − T₁)). When the term structure slopes up, σ₂ exceeds σ₁, and the later period must carry not just its own share of variance but enough extra to drag the whole average up from σ₁ to σ₂. Squeezing that surplus into the shorter forward window makes σ_fwd larger than σ₂. A one-point rise from 12.9% to 13.9% over the second month becomes almost a two-point rise in the forward volatility, to about 14.8% — the leverage of working in variance.
A negative forward variance is an arbitrage flag
The quantity under the square root, σ₂²T₂ − σ₁²T₁, can in principle come out negative — if the longer-dated total variance is less than the shorter-dated total variance. That would require a downward-sloping term structure steep enough that the market implies less total movement over 60 days than over 30, which is impossible for non-overlapping windows: you cannot un-move. So a negative forward variance is not a strange market state; it is a signal that one of the two quotes is wrong, stale, or that the term structure as quoted admits a calendar arbitrage — you could sell the shorter option and buy the longer and lock in a mispricing. Practitioners use exactly this check to sanity-test a quoted volatility surface: no arbitrage-free surface produces a negative forward variance anywhere.
This is the number a calendar spread actually trades
A calendar spread — sell the near expiry, buy the far expiry at the same strike — looks like a bet on time decay, and beginners describe it that way. What it is really long is forward volatility. The near leg you sold covers today to T1; the far leg you bought covers today to T2; the position that survives after the near leg expires is exposure to the window from T1 to T2, priced at the forward volatility when you put it on. If you paid a forward volatility of 14.8% and the second month realises 12%, the structure loses regardless of how gently the spot curve sloped. The gap between the forward volatility you paid and the volatility the window actually realises is the entire economics of the trade, and the spot curve you read off the screen conceals it.
Forward volatility around a scheduled event
The term structure is rarely a smooth upward line; it bulges around events. When an RBI policy decision or a Union Budget falls between T1 and T2, the forward volatility across that window jumps far above the neighbouring windows, because all the event's expected movement is concentrated in one session inside it. This is how a desk isolates the market's price for an event: take the forward volatility across the expiry that contains the event and the forward volatility across the expiry just before it, and the difference is what the option market is charging for that single day. It is also why a calendar spread placed across an event is a specific bet on whether the event delivers the movement its forward volatility is pricing — a cleaner event trade than a naked long option, and a more dangerous one to misjudge.
The whole curve the forward volatility is extracted from
At-the-money implied volatility of NIFTY across expiries, from the near weekly out to three months.
Forward Volatility formula
Forward volatility between two dates
σ_fwd = √( (σ₂²·T₂ − σ₁²·T₁) / (T₂ − T₁) )
Work in variance, never in volatility. Total variance to each date is σ²·T; the forward window's variance is the difference, and its square root over the window length is the forward volatility. If the term is annualised, express T in years; if you use days, use days consistently on top and bottom — the ratio is unit-free either way. A negative quantity under the root means the two quotes admit a calendar arbitrage.
- σ_fwdThe forward volatility — annualised — for the window running from T1 to T2. The unknown being solved for.
- σ₁Spot implied volatility to the nearer date T1, annualised as a decimal (0.129 = 12.9%).
- σ₂Spot implied volatility to the farther date T2, annualised as a decimal (0.139 = 13.9%).
- T₁Time to the nearer date, in years (calendar days ÷ 365) or in days — used consistently with T₂.
- T₂Time to the farther date, in the same unit as T₁. T₂ − T₁ is the length of the forward window.
The variance-additivity identity it rests on
σ₂²·T₂ = σ₁²·T₁ + σ_fwd²·(T₂ − T₁)
Read left to right, this says the total variance to T2 is the variance of the first window plus the variance of the forward window — the additivity of variance over non-overlapping periods. Rearranging it for σ_fwd gives the forward-volatility formula above. If the right-hand pieces do not add up to the left, the surface is not arbitrage-free.
How to extract a forward volatility from two implied volatilities
- Pick the two dates that define your window: the near expiry T1 and the far expiry T2. The forward window is the stretch from T1 to T2.
- Read the two at-the-money implied volatilities off the chain — σ₁ to the near date, σ₂ to the far date — and express both as decimals.
- Convert both to total variance by multiplying the square of each by its own time: σ₁²·T₁ and σ₂²·T₂. Use days or years, but the same unit for both.
- Subtract: forward variance = σ₂²·T₂ − σ₁²·T₁. If this is negative, stop — the quotes admit a calendar arbitrage and at least one is wrong or stale.
- Divide the forward variance by the window length T₂ − T₁ to get the forward variance per unit time.
- Take the square root and, if you worked in days, annualise. The result is the forward volatility for the window.
- Compare it against the two spot numbers. On an upward-sloping curve it should sit above both; if it does not, recheck your arithmetic before you trust it — the counter-intuitive answer is usually the correct one.
Forward Volatility in practice: worked example
NIFTY worked example
NIFTY at 24,000. The 30-day at-the-money implied volatility is 12.9% and the 60-day is 13.9% — a term structure that slopes up by a single point. What is the volatility the market implies for the second month, the window from day 30 to day 60? Work in variance. Total variance to 30 days is 0.129² × 30 = 0.016641 × 30 = 0.49923; total variance to 60 days is 0.139² × 60 = 0.019321 × 60 = 1.15926. The forward window's variance is 1.15926 − 0.49923 = 0.66003, over a window of 60 − 30 = 30 days, so the forward variance per day is 0.022001, and its square root is 0.14833 — a forward volatility of about 14.8%. Interpret it: a one-point rise in the spot curve implied a nearly two-point jump in the second month alone. If you sell a calendar spread here you are short that 14.8%, and if the second month realises anything less you keep the difference — the spot curve never showed you that number.
BANKNIFTY worked example
BANKNIFTY at 52,000 shows how quickly the sign of the slope changes the story. Suppose its term structure is inverted ahead of a bank-heavy results season: the 30-day implied volatility is 18.0% and the 60-day is only 16.5%, because the near month contains the earnings cluster and the far month is calmer. Total variance to 30 days is 0.18² × 30 = 0.972; to 60 days it is 0.165² × 60 = 1.6335. Forward variance is 1.6335 − 0.972 = 0.6615 over 30 days, giving 0.022050 per day and a forward volatility of √0.02205 ≈ 14.8%. So the second month is implied at about 14.8%, well below the near month's 18% — the inverted curve has front-loaded the volatility into the first window. A trader who wanted exposure to the quiet second month, not the noisy first, would structure a calendar to isolate that 14.8%, and the lesson is that an inverted curve makes forward volatility cheaper than spot, exactly the mirror image of the NIFTY case.
Lot sizes used above (NIFTY 65, BANKNIFTY 30) are those in force as of July 2026 (NSE circular NSE/FAOP/70616, effective 28 October 2025); NSE revises them periodically. Figures exclude brokerage, STT, exchange charges, stamp duty and GST. Examples are teaching scenarios built on round numbers — they are not historical quotes, not backtests and not trade calls.
Advantages & limitations of Forward Volatility
What it is good for
- It isolates a future window. Forward volatility is the only clean way to ask what the market implies for the second month alone, rather than for an average over both months that buries the answer.
- It exposes term-structure information the spot curve hides. A gentle one-point upward slope translates into a nearly two-point forward move, a magnification you simply cannot see by reading the spot numbers off the screen.
- It is the correct pricing input for calendar and forward-starting structures. Anything whose exposure begins in the future — a calendar spread, a forward-start option — must be valued off forward volatility, not spot implied volatility.
- It provides a built-in arbitrage check. A negative forward variance anywhere on a quoted surface flags a stale or mispriced quote immediately, which is why desks compute forward variances across the whole term structure as a consistency test.
- It localises the price of an event. The difference between the forward volatility across the expiry containing an event and the window just before it is the market's price for that single session, which no spot number reveals on its own.
Where it breaks down
- It assumes returns are independent across the two windows. If volatility is autocorrelated — a turbulent first month tends to bleed into the second — the clean additivity of variance is only an approximation, and the extracted forward number is biased.
- It is only as clean as the two quotes it came from. Forward volatility differences two implied volatilities, and differencing amplifies noise: a stale or wide quote on either expiry can throw the forward number off by several points, especially when the two variances are close.
- It is undefined when the arithmetic breaks. If the term structure as quoted implies a negative forward variance, there is no real forward volatility at all — the correct output is to flag an arbitrage, not to report a number.
- It says nothing about the skew of the forward window. The formula uses at-the-money implied volatilities and delivers an at-the-money forward volatility; the shape of the forward smile, which a real forward-starting structure is exposed to, is a harder object the simple identity does not touch.
- It is a fragile guide near an event whose timing is uncertain. If the exact date of a catalyst moves — a policy meeting rescheduled, results deferred — the window that was supposed to contain it may not, and the forward volatility computed on the old assumption misprices the structure built on it.
Common misconceptions about Forward Volatility
Misconception: A calendar spread is purely a bet on time decay.
Reality: Both, but its defining exposure is to forward volatility. It is described as a time-decay trade because you collect the near option's faster theta, but what survives after the near leg expires is a position in the forward window's volatility, and that is what decides the outcome.
Misconception: Forward volatility tells you the direction of the market.
Reality: No. Like every volatility measure it describes the expected size of movement in a future window, not its sign. A forward volatility of 14.8% says the second month is priced to move that much annualised; it says nothing about whether the market will rise or fall over that window.
Common mistakes with Forward Volatility
- Assuming the forward volatility sits between the two spot numbers. On an upward-sloping curve it sits above both, often by more than the slope itself, because variance additivity magnifies the difference. Reasoning in volatility units instead of variance units produces this error every time.
- Reading a longer-dated implied volatility as the volatility of the later period. A 60-day implied volatility is an average over both months; the second month alone is implied to be more volatile than that average whenever the curve slopes up.
- Ignoring a negative forward variance instead of treating it as a red flag. If the numbers under the root go negative, the answer is not a small or imaginary volatility to be rounded away — it is that one of the quotes admits a calendar arbitrage and should not be traded on.
- Mixing time units between the two variances. Multiplying one implied volatility by days and the other by years, or by 365 and 252 inconsistently, produces a forward number that is silently wrong, because the whole method depends on σ²·T being on one scale.
- Treating a calendar spread as a pure time-decay trade and forgetting it is long forward volatility. If the forward window realises less volatility than you paid for, the structure loses even if the theta arithmetic looked favourable when you opened it.
- Extracting a forward volatility from overlapping windows. The additivity identity requires non-overlapping periods; differencing a 45-day and a 60-day implied volatility gives the forward 45-to-60 window, not the 0-to-15 window, and confusing which window you have isolated wrecks the trade.
How professionals use Forward Volatility
Volatility desks live on the term structure and therefore on forward volatility. A relative-value trader does not ask whether NIFTY implied volatility is high in absolute terms; they compute the forward volatility across each pair of expiries and ask which forward window looks rich or cheap relative to the others and relative to what the underlying is likely to realise in that window. A forward window that prices far above its neighbours without an event to justify it is a candidate to sell via a calendar; one that prices too low ahead of a known catalyst is a candidate to buy. The entire book can be expressed as a set of long and short forward-volatility positions across the curve.
Structurers price forward-starting products — cliquets, forward-start options, some structured notes — directly off forward volatility, because the client's exposure does not begin until a future date and valuing it off spot implied volatility would misprice the deal from the outset. Risk managers, meanwhile, use the forward-variance non-negativity condition as a hard constraint when fitting a volatility surface: any surface that produces a negative forward variance anywhere is rejected as arbitrageable before it is ever used to mark a book, because a surface that admits an arbitrage will mark positions at prices the market would never honour.
Forward Volatility: frequently asked questions
How is forward volatility calculated?
By working in variance. Multiply each spot implied volatility's square by its own time to get total variance, subtract the near total from the far total to isolate the forward window's variance, divide by the window length, and take the square root. The formula is σ_fwd = √((σ₂²T₂ − σ₁²T₁)/(T₂ − T₁)).
Why is forward volatility higher than spot volatility?
Because variance is additive and the later window must carry enough extra variance to lift the whole average from the near volatility up to the far one. On an upward-sloping curve that surplus, squeezed into a shorter window, makes the forward volatility exceed both spot numbers — a one-point slope can become nearly a two-point forward move.
Why can't I just average the two implied volatilities?
Because volatility does not add or average over time — variance does. Averaging 12.9% and 13.9% to get 13.4% ignores that the second window has to account for all the variance the longer option prices above the shorter one, which pushes the true forward number to about 14.8%, above both inputs.
What does a negative forward variance mean?
It means the quotes are inconsistent. A longer window cannot have less total variance than a shorter one it contains, so a negative forward variance signals a stale or mispriced quote, or a term structure that admits a calendar arbitrage. The correct response is to flag it, not to report a volatility.
What is the difference between forward volatility and spot volatility?
Spot implied volatility covers a window starting today; forward volatility covers a window starting in the future. A 60-day spot implied volatility averages the next two months, whereas the forward 30-to-60 volatility describes the second month alone, which on an upward curve is more volatile than the average.
Can forward volatility be lower than spot volatility?
Yes, when the term structure is inverted. If the near expiry contains an event and the far one is calmer, the far implied volatility is lower and the forward window can price below the near spot volatility. An inverted curve front-loads volatility into the first window and leaves a cheaper forward window behind it.
Does variance really add up over time?
Under the assumption that returns are independent across periods, yes: the variance of two consecutive non-overlapping windows is the sum of their variances. This is the foundation of the whole calculation. When volatility is autocorrelated the additivity is only approximate, which is one of the method's limitations.
What happens to forward volatility if one quote is stale?
It can be thrown off badly. Forward volatility differences two variances, and differencing amplifies noise, so a stale or wide quote on either expiry — especially when the two variances are close — can move the forward number by several points. Always use fresh, liquid at-the-money quotes.
Voice search questions about forward volatility
Natural-language questions people ask about forward volatility.
What does forward volatility actually tell me?
It tells you how volatile the market thinks a future window will be — say the second month, not the whole two months. You cannot read it off the screen because no option covers only that window, so you back it out of the 30-day and 60-day implied volatilities using the fact that variance adds up over time.
Why is the forward number higher than both the quotes I started with?
Because you are working in variance, not volatility. The far option prices more total variance than the near one, and all that extra variance belongs to the second window, which is shorter. Packing the surplus into a shorter window pushes its volatility above both spot numbers — a one-point slope becomes almost two.
Is this the number my calendar spread is really trading?
Yes. A calendar spread looks like a theta trade, but what it is genuinely long or short is forward volatility — the volatility of the window between the two expiries. If that window realises less than the forward volatility you paid, the structure loses, however good the time-decay arithmetic looked.
People also ask about forward volatility
Questions best answered by their own dedicated pages:
Sources & references for Forward Volatility
Published 9 July 2026. Educational content only — not investment advice.