Is there a difference between quant finance 11 years ago and now? Some say, it became more boring, because the simplification of instruments moved emphasis away from modeling to data management. I always disagreed and the week in London confirmed: modeling, and correct model solving matters more than ever.
The core event in London was our Workout in Computational Finance on Thursday, but we had a quite tight schedule meeting partners, members of focus groups and distinguished quant finance experts. It was an exciting week, where we discussed about pure, versus applied mathematica, mathematical creativity, new ways of white-box/black-box learning, the tech offensive that has disrupted quant finance technology vendors and forced them to reinvent and retool their products.
I was privileged to bring together two dozens distinguished delegates from famous institutions with our top speakers Andreas and Michael for the 5 hours workout.
They discussed the pros and cons of numerical methods for mean reverting including trees (only good as fire wood), finite differences, finite elements and Monte Carlo techniques and explained stabilization techniques for convection dominated PDEs.
In the model calibration session they explained why the ill-posedness of calibration is a principle hurdle and showed how stabilization techniques help obtaining stable model parameters.
When Monte Carlo is the only choice, speed maybe the problem. Quasi Monte Carlo or variance reduction techniques speed up.
But having accurate, fast and robust pricing and calibration engines is perhaps not enough. Proper risk management based on VaR, Expected Shortfall measures and the introduction of CVA/FVA/DVA introduce much more complexity in the valuation space. In the workout an out look was given how a sound implementation looks like (including symbolic and massive parallelization).
The spirited panel and individual discussions made it clear to me: mathematical quality still matters, as well as their implementation. And we hope to have also shown how applied mathematics can be fun.