Heston Stochastic Volatility Model: Pricing the Smile

Key Takeaways

• Heston stochastic volatility model drives variance with a CIR mean-reverting process correlated with spot; negative rho produces equity put skew.
• A calibrated SPX Heston model typically shows rho near -0.7, kappa near 2.0, and vol-of-vol xi near 0.5, all shifting with market regime.
• A common mistake: treating Heston parameters as stable forecasts, they are calibrated daily to the surface and can drift significantly across weeks.
• Heston prices cliquets and forward-start options differently than local volatility models because it implies a different forward-smile dynamic.

What It Is

Steven Heston published "A Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options" in the Review of Financial Studies in 1993. The model replaces Black-Scholes' constant volatility with a second stochastic process for variance and derives a semi-analytic price for European options via Fourier inversion.

The model's structural advantage over Black-Scholes is that it generates a realistic volatility smile directly from the underlying dynamics, not as an ad-hoc overlay. Its key innovation is a non-zero correlation parameter between asset returns and variance, which produces the negative skew observed in equity index options.

The Intuition

Black-Scholes assumes volatility is a constant number the trader simply has to guess. That is wrong in two obvious ways. First, volatility varies over time, clustering in stressed periods and calming in quiet ones. Second, volatility is typically negatively correlated with returns on equities, a pattern known as the leverage effect: when stocks fall, volatility tends to spike, and vice versa.

Heston puts both of those features directly into the pricing model. Variance itself follows a mean-reverting square-root process (the same process Cox, Ingersoll, and Ross used for interest rates). A correlation parameter links the Brownian motion driving the stock to the Brownian motion driving variance. Tune those parameters and you can match observed implied volatility surfaces reasonably well.

How It Works

The Heston model specifies two coupled stochastic differential equations:

dS_t = mu * S_t * dt + sqrt(v_t) * S_t * dW1_t
dv_t = kappa * (theta - v_t) * dt + xi * sqrt(v_t) * dW2_t
corr(dW1, dW2) = rho

Where:

• S_t is the underlying price
• v_t is the instantaneous variance (volatility squared)
• theta is the long-run mean variance
• kappa is the speed of mean reversion
• xi is the volatility of variance ("vol of vol")
• rho is the correlation between asset returns and variance
• mu is the drift of the underlying

The variance process is a CIR (square-root) process. It stays non-negative if the Feller condition 2 * kappa * theta > xi^2 holds.

European call prices are computed via the characteristic function of log-returns and Fourier inversion. The pricing formula is:

C = S * P1 - K * exp(-r * T) * P2

Where P1 and P2 are probabilities derived from integrating the characteristic function. The formula is closed in the sense that it requires only one-dimensional numerical integration, which executes in milliseconds.

Worked Example

Calibrating Heston to S&P 500 index options on a typical day might return parameters like v0 = 0.04 (spot variance of 4 percent, roughly 20 vol), theta = 0.06 (long-run variance of 6 percent, roughly 24.5 vol), kappa = 2.0 (mean reversion in about half a year), xi = 0.5 (vol of vol), and rho = -0.7 (strong negative correlation).

The negative rho of -0.7 is the key driver of the equity skew. When the index falls along a path, variance tends to rise on that path, so out-of-the-money puts end up with higher expected payoffs and higher prices than Black-Scholes would give. The resulting implied volatility surface slopes down from low strikes to high strikes, matching the market.

A flat rho = 0 produces a symmetric smile. A positive rho flips the skew the other way, which is sometimes what you see in commodities during supply squeezes where upside volatility dominates.

Common Mistakes

1. Treating Heston as a forecast. Heston is a pricing model, calibrated daily to the observed surface. The calibrated parameters drift over time. Using yesterday's theta and kappa as today's price assumption is a quick way to mis-hedge a book.

2. Ignoring the Feller condition. If 2 * kappa * theta <= xi^2, variance can hit zero, which breaks the square-root term and produces numerical instability. Some calibrations produce parameters that technically violate Feller. Practitioner implementations either enforce the condition during calibration or use reflecting boundary schemes in Monte Carlo.

3. Over-fitting the smile. Heston has five parameters. A Black-Scholes fit has one. You can always fit the smile better with more parameters. The question is whether the fit generalizes across dates. Out-of-sample Heston parameters move around, and the model mis-prices exotic options whose value depends on forward smile dynamics.

4. Confusing Heston with local volatility. Both models reproduce the observed vanilla smile, but they imply different forward skew behavior. Heston implies that the smile flattens and shifts with the spot. Local volatility models (Dupire) imply that the smile is sticky to strike. Those assumptions price barrier and cliquet options very differently.

5. Using closed-form pricing on American options. Heston's closed form is for European exercise only. American options under Heston require finite-difference PDE solvers or Monte Carlo with Longstaff-Schwartz regression, both substantially more expensive than vanilla Black-Scholes early-exercise handling.

Frequently Asked Questions

Q: What is the Heston stochastic volatility model in simple terms? Heston extends Black-Scholes by making volatility itself random. It assumes variance follows a mean-reverting process correlated with stock returns, that correlation produces the negative skew observed in equity index options without adding it as an external overlay.

Q: How does the Heston model affect investment decisions? When pricing exotic options like barriers or cliquets, the choice of model matters because Heston and local volatility imply different future smile shapes. Using the wrong model prices forward-starting volatility exposure incorrectly, which can cost real money on structured products.

Q: What is a real-world example of Heston calibration? S&P 500 calibration yields rho around -0.7, meaning when the index drops, variance tends to spike. That correlation is what makes out-of-the-money SPX puts persistently more expensive than out-of-the-money calls in any realistic model.

Q: How can practitioners use the Heston model effectively? Re-calibrate parameters daily to the current vanilla surface. Check the Feller condition (2κθ > ξ²) to ensure variance stays non-negative. For American options under Heston, use finite-difference PDE solvers rather than the closed-form formula.

Q: How is the Heston model different from local volatility (Dupire)? Both reproduce today's vanilla surface, but they imply different future smiles. Heston implies the smile shifts with spot; local volatility implies the smile is sticky to strike. Those differences price barrier options and cliquets very differently.

Sources

1. Heston, S.L. (1993). "A Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options." Review of Financial Studies, 6(2), 327-343. https://www.ma.imperial.ac.uk/~ajacquie/IC_Num_Methods/IC_Num_Methods_Docs/Literature/Heston.pdf
2. Oxford Academic. "Closed-Form Solution for Options with Stochastic Volatility with Applications to Bond and Currency Options." https://academic.oup.com/rfs/article-abstract/6/2/327/1574747
3. Natenberg, S. Option Volatility and Pricing: Advanced Trading Strategies and Techniques. McGraw-Hill. https://archive.org/details/optionvolatility00shel
4. Xilinx Vitis Libraries. "Heston Model Closed-Form Solution." https://xilinx.github.io/Vitis_Libraries/quantitative_finance/2020.1/methods/cf-ht.html

Disclaimer

This article is educational content only and is not financial advice. Nothing here is a recommendation to buy, sell, or hold any security. Consult a licensed advisor before making investment decisions.