Hi,1) You can search simplex method like Nelder-Mead2) 23.53% is a good upfront payment for this period, you can check another article's with data and you will see the same kind of upfrontJ

Hi,It's a little bit strange when you compute Expectation of Variance volatility of x(k,n)(t) and x(l,p)(s) doesn't appear.With this correction , we will through the volatility of OU process the dependency of Swap rate

<t>Hello,i have a little question about calibration on this article.Mr Piterbarg explains that the equation for swaptions volatiliies involves and model skews . But i don't see where in the equation of swaptions volatilities model skews Beta(t,n) is involved ?I thought in the function g(x) but it's ...

<r>Hi, You can use V.Piterbarg's paper on PRDC, in this paper V. Piterbarg gives a proxy scheme for calibration :<URL url="http://papers.ssrn.com/sol3/papers.cfm?abstract_id=685084RegardsJ"><LINK_TEXT text="http://papers.ssrn.com/sol3/papers.cfm? ... 84RegardsJ">http://papers.ssrn.com/sol3/papers.cf...

Hello,In this thread looks the attached file by V.Piterbarghttp://www.wilmott.com/messageview.cfm?catid=4&threadid=35696J

<t>why "semi"?Because you use a numerical method which is a gaussian quadrature.Explanation of normal approximation (Ni asset nominal, Ri recorate rate of asset i):For each T_i you compute this expected loss and after you integrate it : EL(Ti)[A-B] = int(EL(Ti)[A-B|V]f(V)dV)ProtectionLeg=sum_i{(EL[T...

<t>integration is very very fast with 32 points for example. Moreover there are only 20 dates.For example compare a pricing in a case of homogeneous portfolio with Monte-Carlo and with semi-analytical method. You will see a big gap between the two computation's timeMoreover, Analytic method is more ...

<r>It's exact you must construct a tree with this recursive relation :<URL url="http://db.riskwaters.com/global/events/im_cdotraining2005/precourse_cdos/All%20your%20hedges%20in%20one%20basket.pdffor"><LINK_TEXT text="http://db.riskwaters.com/global/events/ ... ket.pdffor">http://db.riskwaters.com/g...

Recursive loss like for example "All your hedge in your basket" is very very fast compare with Monte carlo methodyou can use a normal approximation like D.Shelton to improve time of computation for example.J

Yes, "homogeneous" or "heterogenous" and no large pool assumptioni advise you articles of John Hull and JP Laurent.J

you can not fit the market prices with a monte-carlo pricer, you must use a semi-analytical model : 1 factor model with gaussian copula.

Pricing Interest Rate-Sensitive Credit portfolio derivativesP.Ehlers and P.J. Schonbucherhttp://www.defaultrisk.com/pp_price_76.htm

a) no s/(1-R) = hazard_rate = lSo Proba_default = 1 - exp(-hazard_rate *T)l = 0.01213, p = Proba_default = 0.0588631c) P(N(T) = k) = (n,k) * p^(k) * (1-p)^(n-k) (N number of defaults)d)EL[0-33] = 33 * P(L > 33) + sum( k = 0 , round(33/(1-R)), k * (1 - R) * P(N(T) = k))etc .....

google -> http://www.fitchratings.com/dtp/pdf4-06/vcco1411.pdf

<t>Can you share Pedersen's Lehman paper please ?JI don't understand your problemIn fact the first problem is to find the initial default intensity (or default probability p_x(0,T) for all T) which fits the tranche spreadsAfter that you can take a special diffusion for your intensity (HW, ...) but y...