Dupire Local Volatility from Option Prices and Arbitrage-Free Surfaces
Summary
The document asks how to estimate local volatility with Dupire’s formula using finite differences in option prices across maturity and strike. Its answer stresses that local volatility is a function of time and the underlying price, rather than a volatility assigned to an individual option. The formula is intended to recover underlying dynamics consistent with the prices of European vanilla options as a group.
For practical evaluation at arbitrary time and spot points, the answer recommends first building a surface from observed option prices, commonly by interpolating implied volatility, and then differentiating that surface. The surface must avoid butterfly and calendar spread arbitrage for a valid local volatility function. The sample price differences in the question alone do not establish that the calculation is reliable: the answer gives no numerical result and does not discuss data quality, interpolation choices, or numerical stability. It points to research on arbitrage-free smoothing as further reading.
Key ideas
- Local volatility describes underlying dynamics as a function of time and spot, not a single option’s volatility.
- Dupire calculations require option prices across strikes and maturities.
- A surface interpolator can provide prices at arbitrary points before derivatives are estimated.
- The implied volatility surface should be free of butterfly and calendar spread arbitrage.
Tags
Full text
# Dupire Formula question
# Dupire Formula question
I want to calculate the local volatility from Dupire's formula:
$\sigma _{VL}^{2} (K,T,S_{0}) = \frac{\frac{\partial C}{\partial T}}{\frac{1}{2} K^{2} \frac{\partial^2 C}{\partial K^2}}$
So I use the aproximation forms as :
$\frac{\partial C }{\partial T} \cong \frac{C(K, T + \Delta T) - C(K,T - \Delta T}{2 \Delta t}$
$\frac{\partial^2 C}{\partial K^2} \cong \frac{C(K- \Delta K, T) - 2C(K,T) + C(K + \Delta K,T )}{(\Delta K)^{2}}$
With this date to get the local volatility of the option SPX(288.5,April 15), should I do this?:
$\frac{\partial C }{\partial T} \cong \frac{2.02 - 1.51}{2 * 3}$
$\frac{\partial^2 C}{\partial K^2} \cong \frac{2.03 -2*1.65 + 1.4}{0.5^2}$
## Answer by LocalVolatility (score 2)
https://quant.stackexchange.com/a/45468
I first want to clarify one statement. You write "the local volatility of the option ...". A local volatility is, unlike an implied volatility, not a property of an option but instead a function of time and the underlying assets price. I.e. an option does not have an implied volatility but the local volatility function describes the dynamics of the underlying asset and is found such as to match the market prices of all European vanilla options jointly.
Regarding your actual problem. Typically you want to be able to evaluate the local volatility surface at arbitrary (time, spot)-points, independent of the market strikes and expiries. In a finite-difference scheme for example, you need to be able to evaluate the local volatility at all grid points. A common approach is thus to construct an a surface interpolator first from the existing European options prices and then evaluate the partial derivatives of this interpolator. While you could interpolate prices directly, it is more common to construct an interpolated implied volatility surface. A crucial property in order for local volatilities to exist is for this surface to be free of butterfly and calendar spread arbitrage.
Here are some references to get you started:
- Fengler, Matthias R. (2009) "Arbitrage-Free Smoothing of the Implied Volatility Surface", Quantitative Finance, Vol. 9, No. 4, pp. 417-428
- White, Richard (2013) "Local Volatility," Open Gamma Quantitative ResearchShown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.