Pricing Partners announced today that Simplex, a new optimisation technique, has been implemented in the Price-it Library to enhance its calibration power. This algorithm allows time dependent Heston Model parameters, accelerates calibration procedure and improves calibration accuracy. With this upgraded numerical method, Pricing Partners enriches Price-it Library and undoubtedly enhances its strong authority on cutting edge model expertise.
A calibration procedure is a necessary step during which market data (usually implied BS volatility) is used as input to determine the parameters of a financial model. The optimum research for those parameters enables the model to fit the best the input market data. However, calibration can be time consuming and depends on the size of the input data, on the number of parameters to calibrate and also on the calibration algorithm itself. Therefore, Pricing Partners tested several methods before picking this Simplex one. Simplex is a powerful method that optimises calibration procedure and effectiveness.
Compared with Levenberg Marquardt, the local and initial value sensitive algorithm, Simplex, a global algorithm, is the only robust and easy method, which is irrelevant to initial input value. On the other hand, a model parameter can be either time dependent or constant. Unlike commonly used calibration methods on the market today, which only provide constant parameters, Simplex produces time dependent parameters. Simplex is suitable for the time dependent Heston Model, a model known for its competitive accuracy and speed. With this outstanding market fit algorithm, virtually any type of underlying derivative can be calibrated precisely by Pricing Partners.
Zaizhi Wang, financial engineer at Pricing Partners, comments: “As a great advantage in Price-it solution, one can use either constant or time dependent parameters for a Heston model. Our newly implemented Simplex enables us to calibrate the time dependent parameters for a stochastic volatility models with a good fit on the entire volatility surface. Users can even choose constant or time dependent parameters. They can also easily retrieve calibration error and quickly measure the overall fitting error for the entire smile surface. I consider the Simplex algorithm robust and reliable in calibration.”
Pricing Partners CEO, Eric Benhamou, adds: “Calibration is key to valuation. Improving the calibration enables us to provide more accurate models and provide unmatched precision in our valuation. Time dependent stochastic volatility models are currently the latest cutting edge models and generate smile and forward smile very consistent to the market. Proving powerful calibration sets therefore gives us a clear advantage to our competitors and positions us as a leading independent valuation and analytics provider”