# Volatility Swaps vs Variance Swaps Convexity Wedge Guide

> Derive the exact mathematical convexity adjustment K_vol_swap ≈ K_var - Var(sigma)/(2*K_var) between linear OTC volatility swaps and quadratic variance swaps.

Canonical: https://volsurfacedesk.com/desk-notes/volatility-swaps-vs-variance-swaps-convexity-wedge
Last updated: 2026-11-14T19:46:50.987Z

**Publication:** VolSurface Desk
**Category:** Vix Futures Variance Swap Convexity
**Published:** 2026-11-13T20:15:25.915Z
**Author:** Dr. Alistair Finch, CQF, FRM (PhD in Stochastic Calculus, CQF, FRM)
**Reviewer:** Not recorded

## Answer-first summary

Compare linear OTC Volatility Swaps against quadratic Variance Swaps: derive the second-order Taylor vol-of-vol convexity discount, Bessel-function Laplace replication issues, and why banks charge wider spreads on pure vol swaps.

## Key takeaways

- A Volatility Swap pays linearly on realized volatility: N_vega × (σ_realized − K_vol_swap), whereas a Variance Swap pays quadratically on σ_realized2!
- By Taylor expansion of √V around E^Q[V] = K_var2, the fair Volatility Swap strike MUST sit strictly BELOW the Variance Swap strike: K_vol_swap ≈ K_var − Var^Q(σ_real) / (2 · K_var)!
- Whereas an uncapped Variance Swap replicates statically with a model-free 1/K2 option strip, √RV is path-dependent on vol-of-vol—exposing dealers who hedge a Volatility Swap with a Variance Swap to short vol-of-vol gamma!
- In FX and commodities where clients dislike quadratic variance blowups, Volatility Swaps and Forward Volatility Agreements (FVAs) remain popular despite a 0.40–1.50 vol-point convexity spread.

## Related financial entities

- **Cboe Volatility Index (VIX)** (VIX) — product: Model-free variance-swap benchmark measuring 30-day expected annualized volatility of the S&P 500 Index derived from SPX option mid-quote prices.
- **Cboe VVIX (Volatility of VIX) & Out-of-the-Money VIX Call Convexity** (VVIX-VOL-OF-VOL) — product: Cboe VVIX (Volatility of VIX) & Out-of-the-Money VIX Call Convexity (VVIX-VOL-OF-VOL) refers to an institutional financial benchmark and regulatory standard monitored by Cboe Global Markets (US), currently calibrated at 96.4 (30-Day VVIX Benchmark) as of October 2026.
- **Over-the-Counter SPX Variance Swaps vs. Corridor Variance & Vega Notional** (VARIANCE-SWAPS) — product: Over-the-Counter SPX Variance Swaps vs. Corridor Variance & Vega Notional (VARIANCE-SWAPS) refers to an institutional financial benchmark and regulatory standard monitored by ISDA Equity Derivatives Definitions (INTL), currently calibrated at σ2 - K2 (Variance Payoff Scaling) as of October 2026.

## Analysis

When a corporate treasurer or macro PM first asks an investment bank derivatives desk for a swap paying **$100,000 per point of realized S&P 500 volatility**, they naturally expect linear **Volatility Swap** payoff N_vega × (σ_real − K_vol_swap). Yet the bank salesperson immediately steers them toward a quadratic **[Variance Swap](/desk-notes/otc-spx-variance-swap-vega-notional-convexity)** (N_var × (σ_real2 − K_var2)), quoting a much tighter bid-ask spread! Why do banks strongly prefer Variance Swaps while charging wider model reserves on linear Volatility Swaps?

Within the [VIX Term Structure, Variance Swaps, Vol-of-Vol VVIX & Tail Hedging](/topics/vix-futures-variance-swap-convexity) silo and our [VIX, Variance Swaps & Tail Hedging Guide](/desk-notes/vix-term-structure-variance-swaps-vvix-tail-hedging-guide), the answer lies in the **realized vol-of-vol convexity wedge** ([OTC SPX Variance Benchmark](/instruments/variance-swaps-over-the-counter-spx-variance-swaps-vs-corridor-variance-vega-notional)).

## Taylor Expansion of √RV and the Exact Convexity Trade P&L

Let V = σ_real2 denote terminal annualized realized variance with risk-neutral expectation E^Q[V] = K_var2. Expand the square-root function σ_real = √V to second order around K_var2 (or equivalently apply the exact algebraic identity σ2 − K2 = 2K(σ − K) + (σ − K)2):

## Quantitative Pricing Model: Volatility Swaps vs Variance Swaps Convexity Wedge Guide

```formula
K_vol_swap = E^Q[ σ_real ] = K_var − E^Q[ (σ_real − K_var)2 ] / (2 · K_var) ≈ K_var − Var^Q(σ_real) / (2 · K_var)
- Exact Convexity Package P&L (Long $N_vega Variance Swap @ K_var vs Short $N_vega Volatility Swap @ K_vol_swap):
- PnL_convexity(T) = N_vega × [ (σ_real − K_var)2 / (2 · K_var) − (K_var − K_vol_swap) ] !
- Notice the Pure "Straddle on Realized Volatility": You pay fixed upfront wedge (K_var − K_vol_swap) and earn quadratic payoff (σ_real − K_var)2 / (2·K_var) whether realized volatility finishes WAY ABOVE or WAY BELOW K_var!
- Worked 6-Month SPX Pricing (K_var = 17.60v, Implied Vol-of-Realized-Vol SD[σ_real] = 5.40v, N_vega = $100,000):
- Fair Convexity Wedge = (5.40)2 / (2 × 17.60) = 29.16 / 35.20 = 0.83 vol pts => Fair K_vol_swap = 16.77v! Break-even realized vol corridor = 17.60 ± 5.40 = [12.20v, 23.00v]!
```

Compare this **realized** vol-of-vol wedge directly against the **implied** forward VIX Jensen wedge analyzed in [Cboe VVIX & VIX Futures Convexity Wedge](/desk-notes/vvix-vol-of-vol-jensen-inequality-vix-futures-wedge), and test option vega nonlinearity in our [Black-Scholes Option Calculator](/calculators/black-scholes-calculator).

## Why Dealers Charge Model Reserves on OTC Volatility Swaps

![Dealer hedging comparison showing static vanilla replication of Variance Swaps versus dynamic SLV vol-of-vol risk on Volatility Swaps](/api/figure?type=flow&title=Why%20Dealers%20Charge%20Model%20Reserves%20on%20OTC%20Volatility%20Swaps&subtitle=Static%201%2FK%C2%B2%20model-free%20hedging%20vs%20unspanned%20realized%20vol-of-vol%20correlation%20risk&steps=Variance%20Swap%3A%3AE%5EQ%5B%E2%88%AB%CF%83%C2%B2dt%5D%20fixed%20uniquely%20by%20single-expiry%20vanilla%20smile%20via%20-2%2FT%20*%20E%5EQ%5Bln%28S_T%2FF_0%29%5D!%7CSmile%20Invariance%3A%3AHeston%2C%20Bergomi%2C%20and%20Dupire%20LV%20ALL%20give%20the%20EXACT%20same%20Variance%20Swap%20strike%20K_var!%7CVolatility%20Swap%3A%3ADupire%20Local%20Vol%20underprices%20Var%28%CF%83_real%29%20by%2040%25-60%25%20compared%20to%20Rough%20Bergomi%20%2F%20SLV!%7CDynamic%20VIX%20Hedge%3A%3ADealer%20hedges%20Vol%20Swap%20%3D%20Var%20Swap%20minus%20dynamic%20OTM%20VIX%20%2F%20forward-variance%20straddles&metrics=DUPIRE%20LV%20BIAS%3A%3AOverprices%20K_vol%3A%3AUnderstates%20Vol-of-Vol!%7CSLV%20%2F%20BERGOMI%3A%3ATrue%20Var%28%CF%83_real%29%3A%3AMatches%20VVIX%20Dynamics%7CVAR%20BID-ASK%3A%3A~0.25%20-%200.40v%3A%3ATight%20Static%20Hedge%7CVOL%20SWAP%20BID-ASK%3A%3A~0.65%20-%201.10v%3A%3AEmbeds%20Model%20Reserve)

## Head-to-Head: OTC Variance Swap vs OTC Volatility Swap vs FVA

| Derivative Feature | OTC Variance Swap | OTC Volatility Swap | Forward Volatility Agreement (FVA) |
| --- | --- | --- | --- |
| **Terminal Payoff Formula** | N_var · (σ_real2 − K_var2) | N_vega · (σ_real − K_vol) | Cash value of ATM straddle at T1 (σ_imp(T1,T2) − K_FVA) |
| **Settlement Underlying** | **Realized** daily log-returns | **Realized** daily log-returns | **Implied** ATM volatility at forward start T1! |
| **Model Dependency** | **Model-Free** (up to jumps/cap) | **Strongly Vol-of-Vol Dependent!** | Forward-smile / cliquet model dependent ([see FVA note](/desk-notes/forward-starting-variance-swaps-fva-calendar-carry)) |
| **Crash Tail Risk for Seller** | Extreme unless [2.5× Capped](/desk-notes/capped-25x-vs-uncapped-variance-swap-crash-risk) | Linear ($100k stays $100k at 60v!) | Linear on forward implied ATM vol |

## P&L of "Long Variance / Short Volatility Swap" Package ($100k Vega)

![Payoff profile of Long Variance minus Short Volatility Swap convexity package across realized volatility outcomes from 6v to 42v](/api/figure?type=bars&title=P%26L%20of%20%22Long%20Variance%20%2F%20Short%20Volatility%20Swap%22%20Package%20%28%24100k%20Vega%29&subtitle=A%20pure%20quadratic%20straddle%20on%20realized%20volatility%20centered%20at%20K_var%20%3D%2017.60v%20%28Wedge%20%3D%200.83v%29&steps=Dead%20Calm%20%28%CF%83%20%3D%209.0v%29%3A%3A%2B%24127%2C200%20PROFIT!%20Variance%20swap%20concavity%20loses%20less%20than%20short%20volatility%20swap%20gains!%7CAt%20Strike%20%28%CF%83%20%3D%2017.6v%29%3A%3A-%2483%2C000%20max%20bleed%3B%20equals%20exact%20upfront%200.83v%20convexity%20discount%20paid%20to%20enter%7CUpper%20Break%20%28%CF%83%20%3D%2023.0v%29%3A%3A%240.00%20break-even%20boundary%20%28%2B5.40v%20move%20recovers%20full%20upfront%20convexity%20cost%29%7CCrisis%20Spike%20%28%CF%83%20%3D%2040.0v%29%3A%3A%2B%241%2C342%2C400%20PROFIT!%20Quadratic%20variance%20surges%20far%20past%20linear%20short%20vol%20swap%20liability!&metrics=MAX%20BLEED%20%40%20K_VAR%3A%3A-%2483%2C000%3A%3A-0.83v%20Upfront%20Wedge%7CDEAD-CALM%20GAIN%3A%3A%2B%24127%2C200%3A%3AAt%209.0v%20Realized%20SPX%7CCRISIS%2040v%20GAIN%3A%3A%2B%241%2C342%2C400%3A%3A16.2x%20Max%20Bleed%20Risk!%7CASSET%20FAVORITE%3A%3AFX%20%26%20Macro%20RV%3A%3APure%20Vol-of-Vol%20Play)

Notice that remarkable first bar (**+$127,200 profit at dead-calm 9.0v realized vol**): buying a Variance Swap against selling a Volatility Swap profits in **both** extreme market crashes AND historic dead-calm summer compressions!

## Frequently asked questions

### How large is the fair strike difference between a 1-year S&P 500 Variance Swap and a 1-year Volatility Swap?

Suppose 1-year SPX Variance Swap trades at K_var = 19.00 vol points and the risk-neutral standard deviation of 1-year realized volatility is ~6.2 vol points (Var(σ) = 38.44). Then K_vol_swap ≈ 19.00 - 38.44 / (2 × 19.00) = 19.00 - 1.01 = 17.99—a full 1.01 vol-point discount!

### Why can’t a dealer replicate a Volatility Swap statically using vanilla European SPX options?

Vanilla European options only depend on the terminal distribution of S_T, which determines expected integrated variance E^Q[∫ σ_t2 dt] uniquely via ln(S_T), but does NOT determine E^Q[√(∫ σ_t2 dt)]! Two stochastic vol models with identical vanilla smiles can have different realized vol-of-vol and thus different Vol Swap prices!

### What happens to a dealer who sells a Variance Swap and buys a Volatility Swap at equal initial Vega?

That pair ("Variance minus Volatility Convexity Trade") is pure SHORT realized vol-of-vol! Taking Payoff_short_var + Payoff_long_vol = -N_vega · (σ_real - K_var)2 / (2·K_var) + N_vega·(K_var - K_vol_swap): the trader earns the upfront convexity wedge if σ_real finishes near K_var, and loses quadratically if σ_real crashes to 8 or spikes to 50!

### How does Carr-Lee Bessel-function replication attempt to hedge Volatility Swaps?

Carr & Lee (2008) showed that ONLY under zero correlation (ρ = 0) between spot and volatility can √RV be replicated via fractional power/Bessel option weights! Because equity SPX exhibits strong negative correlation (ρ ≈ -0.75), Bessel replication breaks down on equity indices and works better in symmetric G10 FX pairs.


## Source URLs

- https://www.isda.org/book/2002-isda-equity-derivatives-definitions/
- https://www.cboe.com/tradable_products/vix/vvix/
- https://www.bis.org/publ/qtrpdf/r_qt2006b.htm

## Citation

- APA: Dr. Alistair Finch, CQF, FRM. (2026). Volatility Swaps vs Variance Swaps: Exact Vol-of-Vol Convexity Pricing. VolSurface Desk. https://volsurfacedesk.com/desk-notes/volatility-swaps-vs-variance-swaps-convexity-wedge
- Canonical article: [Volatility Swaps vs Variance Swaps: Exact Vol-of-Vol Convexity Pricing](https://volsurfacedesk.com/desk-notes/volatility-swaps-vs-variance-swaps-convexity-wedge)
- Site research index: [VolSurface Desk](https://volsurfacedesk.com/llms-full.txt)
