Reputation
888
Top tag
Next privilege 1,000 Rep.
Edit questions and answers
Badges
2 10
Newest
 Yearling
Impact
~10k people reached

Aug
28
comment The best way to generate market scenarios
The question is still too broad, there are books on this :-) As for 3 if you're referring to numerical efficiency then try in general variance reduction methods; among those applicable in most contexts are stratified sampling or (its advanced high-dimensional version) Quasi Monte Carlo, they can be very tricky though.
Jun
8
comment Covariance estimation
Yes, that's it.
Mar
20
comment Alternatives to CDSs for default term structure?
Thanks, probably an edit gone wrong... (we can delete these comments anyway.)
Mar
20
revised Alternatives to CDSs for default term structure?
edited title
Mar
20
revised Alternatives to CDSs for default term structure?
added 27 characters in body; edited title
Mar
18
asked Alternatives to CDSs for default term structure?
Nov
13
comment Mutivariate t markets
Added some more context, hope it clarifies my issue. Moving to copulae afaik complicates all tasks such as affine transforms, PCA&c, conditioning, fast simulation, stable calibration, which are most practical with elliptical densities (isn't the t copula defined&treated in terms of the multivariate density anyway?). But I'm certainly missing some cool shortcuts, I'm not up to date.
Nov
13
revised Mutivariate t markets
Added more context.
Nov
12
comment Mutivariate t markets
I agree with your considerations, the different multivariate t-s do have many limits, but also some advantages, especially in dimension reduction one might want an elliptical density anyway, or in fast simulation. However the point is to clear up issues in the simpler setting first, since the mixed one is only more involved. One has to understand what are the issues with a certain dependence, be it in form of a copula or not. There are always better models. By the way, I've seen few robust copula calibrations...
Nov
12
asked Mutivariate t markets
Sep
28
awarded  Yearling
Sep
24
awarded  Autobiographer
Aug
22
comment Comparison of multicurve calibration methods
Thanks a lot, I definitely underappreciated some of these points. Nevertheless what do you think of the bootstrap drawbacks?
Jul
28
revised How to deal with extreme cases in normal random numbers generation?
deleted 1 character in body
Jul
28
revised How to deal with extreme cases in normal random numbers generation?
added 9 characters in body
Jul
28
comment How to deal with extreme cases in normal random numbers generation?
Btw what is your $\mathcal U(0,1)$ generator?
Jul
28
answered How to deal with extreme cases in normal random numbers generation?
Jul
21
revised Comparison of multicurve calibration methods
added issue
Jul
21
comment Comparison of multicurve calibration methods
I don't even get the comment: it is bootstrapping that fits datapoints perfectly giving awkward curves... and no-arbitrage conditions are rarely satisfied by either approach.
Jul
21
asked Comparison of multicurve calibration methods