wilmott.com

I am looking for data to test against using the Burgard-Kjaer nonlinear PDE model. Maybe MC as baseline case.

I am looking for data to test against using the Burgard-Kjaer nonlinear PDE model. Maybe MC as baseline case.

Hi,
for a start: Maybe just download the ore engine from quaternion. Its free and quantlib based. There are some examples in there. So this could be a benchmark.
I havent done this yet but maybe will be doing this in the future.

Hi Bernd,
Thanks. Two of my MSc students did some work on CVA using stochastic meshes and I wanted to compare against the Burgard-Kjaer PDE (as in Green XVA book). There was not enough to get them up to speed on PDE. The idea was to use Boost odeiint.
I downloaded Quaternion. I am wondering how the API works on top of Quantlib.

Hi Bernd,
Thanks. Two of my MSc students did some work on CVA using stochastic meshes and I wanted to compare against the Burgard-Kjaer PDE (as in Green XVA book). There was not enough to get them up to speed on PDE. The idea was to use Boost odeiint.
I downloaded Quaternion. I am wondering how the API works on top of Quantlib.

Hi Daniel,
sounds like interesting work.
i am not the best ore expert but let me try:
The ore is generating future market data (curves for example) based on an MC Simulation. Then for each scenario it calls all the (QuantLib-) pricers needed for the netting set you are calculating the CVA on. Finally it collects the results and does some simple calclations on the szenario npvs to get to the cva.
Before you can run it you have to set up some xml files for the market data and for the instruments in the netting set. Some general parameters for the netting set have to be set also.
Choosing boost:odeint is nice. I only wonder if the performance will be good enough.

My notes how to build ORE from source
https://letyourmoneygrow.com/2018/06/25 ... thout-git/