SciCal: SV/SVJ pricing model calibration

SciComp has applied its expertise in numerical methods to produce SciCal, a full-featured standalone calibration module for stochastic volatility (SV) and stochastic volatility + jumps (SVJ) models. SciCal provides a very fast and stable global optimization that avoids the instability pitfalls of standard calibration methods. SciCal is a powerful tool to calibrate your SV and SVJ models, whether SciFinance-generated or programmed traditionally.

Pricing for the forward volatility skew * SV and SVJ

Choosing the correct model for accurate pricing in the presence of volatility skew has always been a challenge. It's even more difficult for the growing volume of cliquet structures and other options with forward starting features, whose price depends critically on forward volatility skews as well as today's skew. Unlike local models, stochastic volatility (SV) reproduces today's skew while preserving forward skew, with just a few adjustable parameters. These desirable features are making SV a standard for pricing skew-dependent structures. Adding jumps (SVJ) further improves the fit at the shortest expirations.

SV and SVJ * SciCal easily handles both

SciCal can quickly calibrate SV or SVJ models to your market data, from a handful to hundreds of options. Like most calibrations, SciCal searches for the parameters that minimize the model/market mismatch – the fit energy. What makes SciCal so different is the way it searches.

Hate guessing? * SciCal doesn't ask you to

Standard calibration algorithms start from your initial guess for model parameters and slide down the fit-energy 'hill' until reaching a minimum. Unfortunately, there are typically many local minima, and your initial guess may not take you to the global one. This leads to instability. Calibrated model parameters can jump suddenly as the market moves. Running an exotics book based on unstable model parameters is a risky proposition. SciCal requires no guessing.

Global method * SciCal looks 'everywhere'...fast

SciCal looks throughout parameter space, randomly, and will climb out of local minima and up a hill to find a deeper minimum on the other side. Stochastic global search algorithms such as this do not have a reputation for speed. But the modern variant used in SciCal is very fast and has been highly tuned to the SV/SVJ calibration problem. Using SciCal, an SV model can be calibrated to several hundred options in about 30 seconds. An SVJ calibration runs in under a minute.

Accessible to all * SciCal works for quantitative analysts, traders and risk managers

All can benefit from SciCal, which runs in Excel or as a standalone executable for Windows or Unix. SciCal allows expert users to adjust important search parameters for complete control over the calibration process, or SciCal can make optimal choices for time/accuracy/stability tradeoffs.

Portable and price-savvy * SciCal is the expert solution

This expert solution can start helping you today and is much less expensive than an in-house development. We've developed SciCal with practitioner input from many banks so you benefit from the shared experience of users. SciCal is portable, not dependent on resources that may not be there tomorrow. Versions that handle other volatility models and include exotics in the calibration suite are under development now.

The one-two punch

If you are building, converting, or testing models requiring SV and SVJ calibration, leverage the power of SciCal together with SciFinance to implement your state-of-the-art models in one tenth the time. This complete solution allows you to always stay ahead of the curve.

Need more information on SciCal? Contact us >>