Johan Strålfors presents his master's thesis


Calibration of a dynamically weighted Heston and Levy model using the
iterated extended Kalman filter


Abstract

A wide set of financial derivatives is in use today, almost none of
the more complex have readily available two way price quotes. To price
these we need to determine a suitable model for the price process of
the underlying asset. This thesis proposes a dynamically weighted mix
between the finite moment log stable and Heston models, calibrated to
daily option data with the iterated extended Kalman filter. The data
used is a set of vanilla option prices from the S&P 500. The mixed
model outperforms both the Heston and finite moment log stable process
in terms of absolute price deviations and inside spread predictions
but the estimated parameters show less stability in simulation
studies.