What causes the volatility smile in equity options?

The smile reflects four overlapping market features: (1) fat-tailed returns, actual returns exhibit kurtosis above what the normal distribution implies, so OTM strikes carry tail-risk premium; (2) leverage effect, equity volatility rises when prices fall, producing left-skewed implied distributions; (3) supply-demand imbalance, institutional hedgers buy OTM puts for portfolio insurance, bidding up downside premium; (4) jump risk around scheduled events. Models like SABR, Heston, Jump Diffusion, and Variance Gamma each capture different combinations of these features, which is why their smile shapes differ from each other.

Why do different options pricing models give different prices for the same option?

Each model makes different assumptions about how the underlying moves. Black-Scholes assumes constant volatility and continuous paths, producing a single fair value per option. Heston adds stochastic volatility, capturing the smile that Black-Scholes can't reproduce. Jump-diffusion models (Merton, Kou, Bates) add discontinuous price jumps for event risk. Variance Gamma uses pure-jump Lévy dynamics with explicit fat tails. Local Volatility fits the listed surface exactly by construction. The 'correct' price depends on which assumptions best match the underlying's actual dynamics, and the gap between two models is itself priced information about which assumptions the market is paying for.

What is a model-implied probability distribution?

A model-implied probability distribution is the histogram of expiration-spot outcomes that a given pricing model assigns. Black-Scholes implies a log-normal distribution. Heston implies a richer distribution shaped by stochastic vol and the leverage effect. Variance Gamma implies a fat-tailed Lévy distribution. The risk-neutral implied distribution from listed prices (computed via the Breeden-Litzenberger second derivative of the call-price curve) is what the market is collectively pricing; different models try to approximate this with their parameter set. The platform's probability views show the risk-neutral implied distribution from listed prices and overlays each model's implied distribution for comparison.

Which pricing model should I use for which situation?

Black-Scholes for fast pre-trade pricing, IV computation, and as the reference for divergence checks. Heston for surface-aware pricing where the smile and term structure matter jointly. SABR for fitting per-expiration smiles cleanly. Local Volatility when you need calibration-perfect repricing of the listed surface. Jump-diffusion (Merton, Kou, Bates) for options around earnings or known event jumps. Variance Gamma for fat-tailed Lévy returns. Binomial trees or PDE for American-exercise pricing. Monte Carlo for path-dependent exotics and risk-neutral probability distributions. The platform exposes all of these so you can compare.

Variance Gamma Model - Skew & Kurtosis Pricing

Last reviewed: May 17, 2026 by Options Analysis Suite Research.

What Is the Variance Gamma Model?

The Variance Gamma (VG) model is a pure-jump Lévy process for the underlying, introduced by Madan, Carr, and Chang in 1998. Where Black-Scholes uses Brownian motion (continuous paths, normal log-returns) and jump-diffusion uses Brownian motion plus discrete jumps, VG replaces Brownian motion entirely with a Brownian motion subordinated by a Gamma process, meaning time itself runs at a random rate. The result is a process with no continuous component but with infinitely many small jumps in any time interval.

VG matches a key empirical fact about asset returns: log-returns aren't normally distributed. They have fatter tails than normal and they're often skewed. VG produces exactly this: the resulting return distribution has explicit skew and kurtosis parameters, and the model's three parameters are directly interpretable as moments of the return distribution.

The Three Parameters

σ: volatility of the Brownian motion before subordination. Sets the scale of returns. Higher σ produces wider distributions in the same way Black-Scholes σ does, but with the additional fat-tail and skew structure VG inherits from the subordinator.
θ (theta): drift of the Brownian motion before subordination. Negative θ produces left-skewed returns (the equity-market norm where downside risk dominates upside potential). Positive θ produces right-skewed returns characteristic of certain commodities and emerging-market-currency options.
ν (nu): variance rate of the Gamma subordinator. Higher ν means time runs more erratically (long quiet periods punctuated by bursts of activity), which produces higher tail-heaviness. As ν → 0, VG converges to Black-Scholes (the subordinator becomes deterministic time).

The Subordinated Brownian Motion

VG's defining mathematical structure is subordination: instead of running a Brownian motion in calendar time, it runs in a random time scale set by a Gamma process. The Gamma subordinator is a non-decreasing Lévy process whose increments are Gamma-distributed, so the "time" experienced by the underlying is random, with some intervals seeing very small effective time elapsed and other intervals seeing large effective time elapsed. This produces the empirically-observed feature that asset-return volatility comes in bursts: most days have small moves, occasional days have very large moves, and the distribution of observed returns has heavier tails than the normal distribution VG would produce in deterministic time. The three VG parameters control the overall scale (σ), the skew direction (θ), and the burstiness of the subordinator (ν).

What Does Variance Gamma Capture?

Fat-tailed returns: kurtosis is explicit and tunable through ν, controlling how much weight sits in the wings of the implied distribution.
Skew: θ < 0 produces the left-skew that equity returns exhibit, with downside outcomes more probable than what the symmetric normal distribution would imply.
The volatility smile: VG's skewed and leptokurtic distribution naturally produces smile shape without needing stochastic-volatility dynamics. The smile emerges from the static return distribution rather than from time-varying vol.
A characteristic function in closed form, making FFT and Fourier-COS pricing fast and calibration tractable.
Risk-neutral moments (skewness, kurtosis) that map directly to interpretable parameters, useful as diagnostics for understanding what the market is pricing.

What Doesn't Variance Gamma Capture?

Stochastic volatility: VG's volatility is constant in the same sense Black-Scholes' is. Vol-of-vol risk needs a stochastic extension (CGMY or VG-CIR), where the vol process is itself random and produces dynamics beyond what the static VG can model.
Continuous diffusion: VG is pure jump, which produces unrealistic high-frequency dynamics for some applications. Hybrid models that combine VG with a continuous component (jump-diffusion-with-pure-jump-extras) are the practical next step.
Term-structure changes that aren't a function of the three parameters: VG can fit a single tenor cleanly but jointly fitting term structure may leave residuals that a richer Lévy-stochastic-vol hybrid would capture.
Path-dependent exotics: VG's pure-jump nature means barrier-and-lookback pricing under VG is theoretically possible but numerically harder than under continuous-path models.

VG in the Lévy Family

Variance Gamma sits within the broader Lévy-process family of pricing models: CGMY (Carr, Geman, Madan, Yor) generalizes VG with four parameters, NIG (Normal Inverse Gaussian) provides a different subordinated Brownian motion, and other Lévy specifications offer still more flexibility. All of these share the closed-form characteristic function property that makes FFT-based pricing fast. The platform exposes VG as the canonical Lévy-family example because it has the cleanest three-parameter interpretation and the smallest calibration challenges; CGMY and other extensions are accessible through the Python SDK for research purposes but aren't exposed as primary surfaces in the web UI.

When Should You Use Variance Gamma?

Fitting a single-expiration smile where the empirical kurtosis and skew need an explicit parameterization rather than an implicit one through stochastic-vol parameters.
Pricing options around tail events where the explicit fat tail of VG is more honest than the under-tailed Black-Scholes.
Risk management: VG-implied VaR and ES respect the empirical distribution shape rather than assuming lognormality.
Research on return distribution properties where explicit skew and kurtosis parameters are useful diagnostics for understanding what the market is pricing.
As one of the cross-model surfaces in a divergence view: VG's disagreement with Black-Scholes specifically isolates the fat-tail premium.

How OAS Uses VG

The platform exposes Variance Gamma as one of the alternative-model surfaces calibrated to listed prices, with FFT-based pricing for fast evaluation. The model-divergence view often shows VG and Black-Scholes disagreeing on OTM strikes; the gap reflects the fat-tail premium the market is pricing that Black-Scholes ignores. VG is also one of the eight models in the regime-detection cross-model fit-error suite; when VG fits the surface best relative to the other seven calibrated models, it indicates the listed prices are emphasizing fat-tail features that smoother models can't reproduce.

Calibrate VG in the pricing calculator

Related Concepts

Jump Diffusion (vs) · Black-Scholes · Heston · SABR · Local Volatility · FFT Pricing · Volatility Skew · Volatility Smile · Tail Risk · Risk-Neutral Density · Implied Volatility · Model Divergence · Calibration · Model Landscape

Try this model in the pricing calculator

This section is part of the Options Analysis Suite Documentation. Browse the full model index or compare alternatives in the pricing calculator.

As of the latest snapshot, AAPL has an ATM implied volatility of 21.6%, IV rank 28% (percentile 11%); 20-day realized vol 16.3%. 25-delta skew is +1.1%, meaning OTM puts trade richer than OTM calls. The IV here is the input that pricing-model walkthroughs (Black-Scholes, Heston, SABR, local-vol) take as their starting point: each model decomposes the same observed quote into different latent dynamics (constant vol, stochastic vol, surface-fitted vol, etc.) which is why two models can agree on price but disagree on Greeks and on how vol will evolve.