The beta stochastic volatility model

Local stochastic volatility models combine perfect calibration at time zero with realistic price dynamics. But traditional methods tend to underestimate the forward skew, and mis-price exotics such as cliquets as a result. Piotr Karasinski and Artur Sepp introduce a new model that uses the sensitivity of the at-the-money implied volatility to spot as a key input to ensure forward skew is better fit and exotics are correctly priced

The traditional approach to stochastic volatility (SV) modelling begins with the specification of an SV process, typically on the grounds of its analytical tractability (see, for example, Heston, 1993). Then, after a closed-form solution for vanilla options has been derived and implemented, the parameters of the SV process are calibrated to the implied volatility surface of vanilla options using complicated non-linear optimisation methods. The drawback of this approach for business applications