I need to improve the speed of convergence of PRNG Monte Carlo. I'm opening a new thread for that purpose and I have question / need confirmation about the algorithm.

I'm pricing options with Heston, QE scheme.

  • For an European Down and In Put on a basket, I was thinking about using a vanilla put as a control variate (CV).

    I read that the correlation between the initial and modified (CV one) random variables had to be highly correlated to produce significant reduction. In the case my barrier is -30% from the Spot, I'm not sure this will be very correlated... What do you think ? By the way, there's no closed form formula for the Heston basket. How should I proceed then ?

  • For an American D&I put ? What's the CV ? Same thing, no closed form formula.

Edit: To be a little more precise

Let's say I'm dealing with a WorstOf basket: $$\text{Pay off}(T)=\max(0,K-\min(S^T_1,S^T_2))$$ What I was thinking as a CV is the following: $$CV_1 = \max(0,K-E(S^T_1)-S^T_2)$$ $$CV_2 = \max(0,K-E(S^T_2)-S^T_1)$$ Then this would be equivalent to pricing 2 vanilla put with new strikes $K_i=K-E(S^T_i)$ where $E(S^T_i)$ is given by the analytical Heston price.

Therefore the new payoff would be:

$$\text{New Payoff} = e^{-rt}\dfrac{1}{n}\sum\limits_{j=1}^{n}\text{Old Payoff} - CV_1^{(j)} - CV_2^{(j)} + E(S^T_1) + E(S^T_2)$$

The idea can be found in the paper below: https://www.researchgate.net/profile/Paolo_Pellizzari/publication/23744210_Efficient_Monte_Carlo_Pricing_of_Basket_Options/links/54a176970cf267bdb902bee9/Efficient-Monte-Carlo-Pricing-of-Basket-Options.pdf

An another idea could be as below but I don't know how to apply the methodology (pricing with a deterministic volatility as close as Heston vol) http://www.iaeng.org/IJAM/issues_v45/issue_1/IJAM_45_1_07.pdf

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.