Quantitative Markets Research Lab

Module 07 · Flagship Study

From Black-Scholes to Stochastic Volatility: Why Market Reality Requires Better Models

A study in how a beautiful model fails, what the failure teaches, and what a disciplined repair looks like — ending where all good modeling ends, in calibrated humility.

Revised · numerical recovery study, not a market calibration

Reading time
Approximately 12 minutes
Prerequisites
Option pricing fundamentals, implied volatility, and stochastic-process notation
Contents · six sections

I · The model

What Black-Scholes claims#

The 1973 model makes one structural bet: that a single number, \(\sigma\), describes the randomness of an asset for the life of an option. Under geometric Brownian motion with constant volatility, continuous frictionless hedging, and a constant rate, every European option on the same underlying and expiry must be priced by the same \(\sigma\). This is not a minor implication — it is the testable implication. Invert the formula on market prices and the implied volatility should come back flat across strikes:

$$\sigma_{\text{imp}}(K, T) = \text{const} \quad \text{for all } K.$$

The relevant historical claim is narrower than “the smile began in 1987.” Rubinstein’s S&P 500 option study documents a pronounced and persistent post-crash skew in the 1987–1992 sample, while emphasizing that the observed surface is inconsistent with the constant- volatility model.2

II · The refutation

What any trading day violates#

Take the model's assumptions in order and hold them against an ordinary day of market data. Returns are not Gaussian: daily index returns show excess kurtosis and negative skew; moves the model rates as once-in-millennia arrive within careers (the arithmetic is in Note VI). Volatility is not constant: it clusters — quiet weeks follow quiet weeks, violent days follow violent days — and it moves inversely with price, the leverage effect. Prices jump: earnings, policy surprises, and liquidity spirals produce discontinuities no continuous diffusion generates. Each violation leaves a fingerprint in option prices, and the market, which prices repricing risk whether or not the model does, records them all in one curve:

Implied volatility across strikes for a 6-month option (S₀ = 100). The smile-consistent series uses points and a solid line; Black-Scholes uses a dashed flat line. The slider in section IV moves the dashed line, and the comparison table supplies the underlying values.

Out-of-the-money puts trade at implied volatilities several points above at-the-money; out-of-the-money calls below. Traders are paying up for crash protection and the market is quoting that premium through the Black-Scholes formula, using implied volatility as the unit. The smile is the market's standing annotation on the model: your Gaussian is too thin on the left, and your constant σ is not constant.

III · The repair

Heston's intuition: let variance breathe#

Heston (1993) promotes variance from parameter to process:1

$$dS_t = \mu S_t\,dt + \sqrt{v_t}\,S_t\,dW_t^S, \qquad dv_t = \kappa(\theta - v_t)\,dt + \xi\sqrt{v_t}\,dW_t^v, \qquad d\langle W^S, W^v \rangle_t = \rho\,dt.$$

Three mechanisms do the work. Mean reversion (\(\kappa\), \(\theta\)): variance is pulled toward a long-run level, so volatility shocks decay — this produces clustering, and it makes long-dated smiles flatter than short-dated ones, as observed. Vol-of-vol (\(\xi\)): variance itself is risky, which fattens both tails of the return distribution and gives the smile its curvature. Correlation (\(\rho < 0\)): the leverage effect, wired in explicitly — negative return shocks arrive with rising variance, skewing the distribution left and tilting the smile downward toward high strikes. With \(\rho\) near \(-0.7\), the model generates the equity index skew almost as a matter of course. And because the model has a closed-form characteristic function, whole volatility surfaces can be calibrated in seconds — a large part of why Heston, and not a dozen worthier candidates, became the industry workhorse.

IV · The comparison

Black-Scholes vs. Heston, strike by strike#

Below, the flat-σ model meets Heston-generated target prices. The Heston call values are precomputed offline from the Heston characteristic function (Heston 1993, "little trap" formulation, Simpson-rule integration of the P₁/P₂ integrals; parameters in the table note). The Black-Scholes prices are computed live at whatever flat volatility you choose. No single σ reconciles the two columns — that is the entire point.

Scope boundary No dated option chain, bid–ask spread, or observed implied-volatility surface enters this example. It is a numerical recovery and misspecification audit, not an empirical Heston calibration. The fixed Heston grid is the target and flat Black–Scholes is the candidate.

Objective and residual definition#

For the displayed one-parameter comparison, the diagnostic objective is the vega-weighted implied-volatility root-mean-square error

$$J(\sigma)=\sqrt{\frac{\sum_i \nu_i\,[\sigma-\sigma_i^{H}]^2}{\sum_i \nu_i}}, \qquad \nu_i=\operatorname{Vega}_{BS}(K_i,\sigma_i^{H}).$$

A genuine market calibration would replace the generated targets with timestamped option quotes and should weight residuals by measurement quality such as bid–ask spread. Exposing the objective is essential: the parameter vector is otherwise impossible to interpret or reproduce.

Flat Black-Scholes volatility#

19.2% is the ATM implied vol of the Heston surface — the "best single number" available.

Six-month European call prices across strikes: precomputed Heston values shown as diamonds against live Black-Scholes at the chosen flat σ shown as a solid line.
Pricing error, BS minus Heston. Positive and negative bars use different fill patterns as well as different sides of zero. Matching at-the-money leaves the wings mispriced — cheap crash protection on the left, overpriced upside on the right.

Heston parameters: S₀ = 100, r = 2%, q = 0, T = 0.5y, v₀ = 0.04, κ = 2.0, θ = 0.05, ξ = 0.6, ρ = −0.7. Values computed offline in Python (Simpson rule, 4,000 nodes on [0, 200]) and hardcoded; implied vols recovered by bisection. Inspect the Heston implementation at commit ad24c499958 and the derivatives notebook at the same commit.

V · The residue

What stochastic volatility still misses#

Heston is a better description, not a true one. Its diffusive variance cannot move fast enough to explain the steep smiles of short-dated options — a market pricing overnight jump risk needs jumps, which is why practice moved to Bates (Heston plus Merton jumps) and beyond. Its single variance factor forces the whole term structure of skew to move in lockstep, while real surfaces twist. Its calibrated parameters, re-fit each morning, drift in ways the model says they should not — a slow-motion confession that the model is a curve-fit to a deeper process. And the modern frontier (rough volatility, with Hurst exponents near 0.1) suggests volatility's memory is structurally unlike anything a Markovian diffusion can produce. Each generation of models recapitulates the same pattern: capture the last generation's known failure, inherit an unknown one.

VI · Conclusion

Model risk and epistemic humility#

The deepest lesson of the smile is not that Black-Scholes is wrong — every model is wrong — but that markets metabolized its wrongness. The formula survived its own refutation by becoming a quoting convention: implied volatility is the market's chosen coordinate system for expressing exactly how non-Black-Scholes the world is. That is the relationship with models this lab advocates. A model is an instrument for locating structure in uncertainty — for saying precisely where reality diverges from an idealization, and by how much. Used that way, Black-Scholes remains among the most useful wrong ideas ever written down. Used as prophecy — as a claim that the tails are thin because the formula needs them to be — the same equation has helped sink funds and banks. The mathematics did not change between those two uses. The humility did.

Credo Models clarify uncertainty. They do not eliminate it. The moment a model's user forgets the second clause, the model stops being an instrument and becomes a liability with excellent typography.

Scholarly record

Sources and implementation references#

  1. Heston, S. L. (1993), “A Closed-Form Solution for Options with Stochastic Volatility,” Review of Financial Studies 6(2), 327–343. doi:10.1093/rfs/6.2.327 ↩
  2. Rubinstein, M. (1994), “Implied Binomial Trees,” Journal of Finance 49(3), 771–818, especially pp. 771–775 on the post-1987 S&P 500 pattern. doi:10.1111/j.1540-6261.1994.tb00079.x ↩
  3. Black, F., and M. Scholes (1973), “The Pricing of Options and Corporate Liabilities,” Journal of Political Economy 81(3), 637–654. doi:10.1086/260062