Generally when we implement Dupire's local volatility model, we follow the steps below:

  1. Calculate implied volatility from given historical data
  2. Fit the implied volatility skew. So we also know the corresponding call option prices.
  3. Use the Dupire's Formula to calculate the local volatility $$\sigma_{loc}^2(K,T) = \frac{\partial_TC(K,T) + rK\partial_KC(K,T)}{\frac{1}{2}K^2\partial_{KK}C(K,T)}$$.
  4. Use the SDE $$dS_t = r_tS_tdt + \sigma_{loc}(S_t, t) S_t dW_t$$ to do Monte Carlo simulation to get any call option price we want.

I have a few questions:

  1. What's the differences between the call option prices from step 2 and step 4?
  2. Using historical data, we can generate discrete points in the implied volatility surface. If we fix those points and regardless of the shape of the volatility skew, could we perfectly recover the historical call option prices using Dupire's Formula regardless of simulation path?

Is there a formal proof or a detailed explanation of the above two?


1 Answer 1


A few thoughts on the steps please:

1):You would normally calibrate Dupire based on current option prices; so you won't calculate implied volatility from historical prices of the underlying (assuming this is meant), but the current option prices. The prices are usually quoted in terms of implied volatilities, so in most cases you wont need to calculate the implied volatility.

2): This is one of the many possible ways to approach the calibration problem, and is needed to generate more granular input data - market price quotes (which could be in the form of IV) can be very sparse, and we need more granular data along both the strike and the maturity dimension.

3): You will need some finite approximation formulae to estimate the derivatives that appear in this formula, but I assume this is implicit in the statement.

Now regrading the two questions:

1): If the model is correctly implemented, the price produced by the MC should match the input prices. The only conceptual difference between the two prices could be down to the fact that the market prices are quoted in terms of Black Scholes' IV, and the Dupire in terms of market prices directly, but that is only a presentation matter.

2): The short answer is no, as Dupire's model uses and tries to explain the current option prices (not the historical or future).

Hope this helps!

  • $\begingroup$ Thank you very much for your comment. (1). You mentioned the price in step 4 matches step 2 but not step 1(which is historical data). Is that because of the imperfect fitting of the IV skew? What if we can perfectly fit the historical price so that prices in 1 and 2 are the same? (2) You said no. Does that mean we cannot recover the historical prices at all or it is path dependent? It is very abstract and could you further explain it? Thank you. $\endgroup$
    – Lefair
    Mar 6, 2020 at 23:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.