It seems that the race for speed in finance is driven by sprinting for advantage in providing more frequent trading. Huge computing centers are built to trade in less-than-a-blink-of-an-eye. High-frequency traders use algorithms to zip in and out of markets often changing orders (and even strategies?) within seconds. So far so ….
But the technological arms race in financial markets might also open doors into the human-intelligence business of sophisticated deal types, strategies and trade decisions.
I talk for example about hybrid CPU-GPU architectures that will help us to exploit a deluge of computational information and revitalize complex models and valuation techniques that have been unmasked as unreliable at the begin of the crisis and would be even more in question when adopting the frequent re-calibration paradigm.
We utilize such brand-new computing muscles and offer power quants and risk professionals to use them even under their desk.
Andreas Binder will present the amazing results of Efficient Calibration Of Advanced Volatility Models on the GPU, at the Frankfurt MathFinance Conference, 14-Mar and 15-Mar-11.
But this is not just first views - it is a deliverable result benchmarked on transtec Tesla workstations.
To proof we will show the new UnRisk-Q extension live at the Frankfurt Conference Exhibit together with transtec.
