Erik Lindgren presenterar sitt examensarbete

Calibration of Heston's Stocastic Volatility Model Using the Extended Kalman Filter
	
Abstract

Exotic options are derivatives with extra features that make them more
complex than regular vanilla options. Usually, these derivatives are
not quoted in the market, like stocks or currencies, but are traded
over the counter. To accurately price exotic options, a pricing model
that describes the dynamics of the underlying asset(s) is
needed. After choosing an appropriate model, the calibration problem
is about fitting this model to market data. It is desirable to find a
calibration method that is insensitive to small departures from the
model assumptions and that produces stable estimates. This paper
evaluates the Extended Kalman Filter as a calibration tool for Heston
s stochastic volatility model. Calibration is performed against both
simulated, OMX and S&P 500 index call option prices. A least squares
calibration method is used as a benchmark. The Extended Kalman Filter
outperforms the least squares estimator in terms of speed and
stability in the simulation study, but fails to find any converging
parameters when calibrating against OMX index options. Further, the
least squares and Extended Kalman Filter estimates do not coincide for
the OMX data. The more liquid S&P 500 index options produce more
realistic calibration results and show a higher correspondence between
the least squares and filter estimates for nearly all parameters in
the Heston model.