Surface SVI Power-Law Kernel Calibration Across Multi-Expiry Index Options

Calibrating 22 SPX expiration slices from 1-DTE dailies out to 3-year LEAPS with three global parameters plus an ATM variance curve θ_T allows Surface SVI (SSVI) to guarantee zero static butterfly and zero calendar spread arbitrage analytically across all log-moneyness and maturity pairs. Formulated by Gatheral and Jacquier (2014) and extended in our SVI & SABR Volatility Surface Calibration & No-Arbitrage Conditions cornerstone guide, SSVI resolves the central weakness of Gatheral raw SVI five-parameter slice fitting: independent slice hyperbolas inevitably cross in illiquid wings when calibrated one maturity at a time.

The SSVI Master Equation and ATM Variance Time Clock#

Instead of parameterizing total variance against calendar time T, SSVI treats ATM total implied variance θt = σBS2(0, t) × t as the intrinsic stochastic clock of the volatility surface: w(k, θt) = (θt / 2) × [ 1 + ρ·φ(θt)·k + √((φ(θt)·k + ρ)2 + (1 − ρ2)) ]

Here ρ ∈ (−1, 1) represents global spot-vol correlation, and φ(θ) > 0 is a smooth backbone kernel governing how ATM variance skew ∂w(0, θt)/∂k ÷ θt = (1/2)·ρ·φ(θ_t) decays with horizon. On equity index desks trading Cboe Volatility Index (VIX) and Euro Stoxx 50 options, empirical ATM skew follows a near-exact power law φ(θ) = η · θ^(−γ) with γ ∈ (0, 0.5]—and the modified power-law kernel φ(θ) = η / [θ^γ · (1 + θ)^(1 − γ)] to preserve bounded variance slopes at ultra-long maturities.

Global Power-Law SSVI Surface Calibration Pipeline#

Gatheral-Jacquier Global No-Arbitrage Theorems#

What makes SSVI indispensable for Dupire local volatility PDE finite-difference smoothing is that static no-arbitrage reduces to compact closed-form inequalities on θ_t and φ(θ):

Notice the algebraic elegance of the modified power-law form: whenever 0 < γ ≤ 1/2 and η(1 + |ρ|) ≤ 2, both Theorem 4.1 (calendar monotonicity) and Theorem 4.2 (Durrleman butterfly positivity) hold identically for every positive θ > 0 and every strike k ∈ (−∞, +∞)!

Heston-Like vs Power-Law Kernel Performance#

Gatheral and Jacquier originally tested two analytical kernels: the Heston-like kernel φ(θ) = [1 − (1 − exp(−λθ))/(λθ)] / (λθ) and the power-law kernel. Empirical tests across Cboe SPX and Eurex SX5E options confirm that Heston-like kernels flatten ATM skew far too slowly at long maturities and fail entirely to match 1-week to 2-week skew explosions—whereas power-law kernels with γ ≈ 0.47 mirror fractional volatility behavior analyzed in Rough Bergomi short-dated fractional skew models.

Seeding Local Volatility and Margin Stress Grids#

Even when an options market-making desk uses 5-parameter raw SVI inside the tightest liquid 30-day bid-ask spread, risk engines supervised by CFTC and ESMA use modified power-law SSVI as the prior regularization surface inside SVI & SABR Volatility Surface Calibration & No-Arbitrage Conditions and our Black-Scholes Option Calculator. Any individual strike quote that pulls raw SVI away from the SSVI prior must pay an explicit convexity penalty unless backed by tight executable liquidity.