<t>About TARNs, the "custom" is important. For simple TARNs, we can indeed augment the PDE as said in the article (which I did not know, thanks for the link). I know there is a similar trick for Asian options. But if the TARN condition and the Payoff become very bespoke (structurers like having fun)...

<t>Just in addition to Cuch's answer, I would say that MC is the only feasible solution for:- high dimensional problem (both for computation time and numerical convergence reasons)- general path dependent payoff (custom TARN for instance)- model explicitly designed for MCIn order to price financial ...

<t>Congratulations, Vladimir! The book is indeed really interesting.FrenchX, Andreasen has co-authored several papers with Andersen and Madan has co-authored one of Carr's most famous papers (FFT). And if you are interested in Monte Carlo, Glasserman's book can be precious.I think the price started ...

Sorry for seeing this thread quite late. As Antonio said, there are indeed 2 versions of the double Heston. Buehler's version is in 3D but the version we considered is in 4D. Hence we used Monte Carlo for numerical simulation.

.vsprops (Property sheets) are quite useful for setting properties which are common to several projects. For my projects, I use one debug.vsprops and one release.vsprops which manage 99% of the properties. Projects may redefine the properties if needed

<t>Well, the whole point of SABR is to capture the smile and beta is not the only one to be involved.Beta enables to move from lognormal to normal distributions (if you assume beta in [0,1], which is not always the case). Beta's role is actually not trivial, it controls the backbone of the smile.Sig...

<t>Can someone explain me that one (on Polter's website):3.3: I've experimented with the code int i = 2; i = i++;on several compilers. Some gave i the value 2, some gave 3, but one gave 4. I know the behavior is undefined, but how could it give 4?Because i is 2, the loop is executed twice. << ??? wh...

<t>Found on a perl forum. Seems that this code is unreadable for both the programmer and the compiler.=======================================In my experience, C and C++ compilers differ. Take the code:int x = 0;printf("%d %d %d\n",++x,++x,++x);A similar line in Perl (allowing for syntax differences)...

<t>There are few things involved here:- arguments are evaluated from right to left- all arguments are evaluated before calling the function- ++i increments i and passes the current value of i when the function is called (may have changed meanwhile)- i++ increments i and passes the "old" value of i (...

<t>5- brute force one:there are 1000*2^10 possibilities in your universe (1000 coins * 2^10 flip sequences).2^10 + 999 are only heads and 2^10 are with the two-headed coin.So the probability is 2^10/(2^10+999).smarter one:P(A|B) = P(A and B)/P(B) where A = the random coin is the two headed coin and ...

Almost all there is to be known on FX smile can be found in this article. Very detailed regarding the different conventions and how it affects the smile construction.

<t>The two solutions are possible.- If F is the forward rate, you may need to look at the LMM-SABR model, as proposed here (there exist some other SABR-LMM models)- If F is the swap rate, you indeed have to build a swaption volatility matrix. This would be an interesting reference.The first approach...

<t>Hi,Indeed the calibration depends on the interpolation method that you use to compute the slope at the money.In practice, if you have several market quotes, any reasonably smooth interpolation technique may be OK (or even linear). Remember this is only a heuristic to find an appropriate nu. You s...

Lung > Brain >Broccoli > Cauliflower > Brownian Motion > Norway > Wiener process > Great-Britain > Cantor setI like this ranking.

Have a look at this thread (oops not on Wilmott ...). Jackel's algo works very well. I haven't tried Axel's improvements yet. Maybe he can tell you more about it.