# Forward Volatility: Extracting a Future Window from Two Expirations

Learn how total variance—not volatility itself—is subtracted across expirations to infer a future interval, and where the calculation can mislead.

Canonical: https://wiki.fcontext.com/options/forward-volatility/
Fact checked: 2026-07-22

> For educational purposes only; not investment advice.

<a id="answer"></a>

## Direct answer

**Forward volatility** is the annualized volatility implied for a future interval, such as day 30 through day 90, inferred from volatility information covering today through each endpoint. If `σ1` and `σ2` are consistently measured annualized volatilities for times `T1 < T2`, the constant-variance approximation is:

`σ²forward = (σ²2T2 − σ²1T1) ÷ (T2 − T1)`.

The operation subtracts **total variance** (`σ²T`), not volatility percentages. The result is a market-implied, model- and quote-dependent number—not a guarantee of realized volatility and not automatically a tradable price.

<a id="mechanism"></a>

## Why variance is additive across time

Under the simplifying assumption that variance accumulates through non-overlapping intervals, total variance from today to `T2` equals variance from today to `T1` plus variance from `T1` to `T2`. Solving for the missing interval gives the formula above. Taking the square root converts forward variance back to annualized volatility.

The inputs must describe comparable objects: same underlying, timestamp, annualization convention, and a consistent strike or moneyness treatment. A single option's Black-Scholes IV at two unrelated strikes is not the same as a model-free expected-variance measure. Skew, jumps, dividends, rates, settlement, stale quotes, and interpolation can all change the answer. For event analysis, total variance is often more informative because a short event window can raise a near expiration sharply while adding less to a longer annualized IV.

<a id="example"></a>

## One-month 40%, three-month 30%

Suppose comparable one-month and three-month annualized IVs are `40%` and `30%`. Use `T1 = 1/12` and `T2 = 3/12` years:

- One-month total variance: `0.40² × 1/12 = 0.013333`.
- Three-month total variance: `0.30² × 3/12 = 0.022500`.
- Remaining variance: `0.022500 − 0.013333 = 0.009167`.
- Remaining time: `2/12 = 0.166667` year.
- Forward volatility: `sqrt(0.009167 ÷ 0.166667) = 23.45%`.

So the market inputs imply about `23.45%` annualized volatility for months two and three under these assumptions—not `30%`, and certainly not `40% − 30% = 10%`. The one-month number may contain a concentrated event premium. Bid/Ask inputs and a full surface calculation can produce a range rather than one precise result.

<a id="risks"></a>

## Calculation and trading checklist

- Use exact year fractions and the same day-count and annualization convention.
- Capture both surfaces at the same timestamp and retain Bid, Ask, strike, forward, rates, and dividends.
- Compare like moneyness—preferably forward moneyness or Delta—not merely the same dollar strike.
- State whether inputs are single-strike IVs, ATM interpolations, variance-swap rates, or an index methodology.
- Reject or investigate negative forward variance; it often signals inconsistent quotes, interpolation, or input definitions.
- Recalculate across Bid/Ask and nearby strikes to expose estimation uncertainty.
- Separate an inferred metric from a position. A calendar spread also carries Delta, Gamma, Theta, skew, execution, and assignment risk.
- Stress the event moving dates, being canceled, or producing a different magnitude than implied.

<a id="misconceptions"></a>

## Common misconceptions

- “Subtract the two IVs.” Variances weighted by time are subtracted.
- “Forward volatility predicts what will happen.” It is an implication of current prices and assumptions.
- “A high forward number identifies direction.” Volatility is primarily about dispersion, not up versus down.
- “Two expirations are enough without matching strikes.” Skew makes unmatched comparisons unreliable.
- “Negative forward variance is a market forecast.” It is usually a warning that inputs or methodology are inconsistent.
- “A calendar spread directly trades only forward volatility.” Its actual P/L depends on the whole surface and path.

<a id="related"></a>

## Related topics

- [Event volatility](/options/event-volatility/)
- [Implied volatility](/options/implied-volatility/)
- [Forward-start options](/options/forward-start-options/)
- [Calendar spread](/options/calendar-spread/)

<a id="sources"></a>

## Primary and academic sources

- [Cboe: VIX White Paper](https://www.cboe.com/tradable_products/vix/vix_white_paper/)
- [Cboe: VIX Term Structure](https://www.cboe.com/tradable_products/vix/term_structure/)
- [OCC: Characteristics and Risks of Standardized Options](https://www.theocc.com/company-information/documents-and-archives/options-disclosure-document)
- [Carr and Wu: A Tale of Two Indices](https://doi.org/10.3905/jod.2006.616865)