Calibrating Stochastic Volatility Models to High Frequency Data

The practice of calibrating financial derivative models to high frequency data exemplifies a number of modeling, computational and software engineering challenging currently facing the finance industry. This talk illustrates a bias-variance trade-off arising in calibration of stochastic volatility models to high frequency option chain quotations and motivates the need for new modeling and calibration methodology.

Through collaboration with the Old Dominion University (CS) and Xcelerit (a HPC financial company), this talk also addresses the implementation challenges of using multi-core and many-core processors to accelerate robust calibration techniques and describes the many software engineering challenges limiting the ability of financial institutions to adopt accelerator platforms.