A Workout in Computational Finance

19 Building Large Software Systems for the Financial Industry

The bigger part of this book has focused on numerical methods for either the valuation of financial instruments (trees, finite differences, finite elements, Monte Carlo, and Fourier techniques) or for the calibration of financial models (inverse problems and regularization, optimization). Building a large software system for the financial industry requires, from the authors’ point of view, combining these methods with data management techniques in order to obtain a robust system able to run continuously for years. The design and development of the UnRisk FACTORY, in which the authors have been involved, was driven by commercial users from the finance industry interested in a more server-oriented and persistent solution than the valuation and calibration library mentioned above. On the following pages, we will highlight key aspects of such a solution.

Database

The financial database comprises the central hub of the application. Foremost, it needs to store the mathematical transcriptions of the termsheets of all instruments that are valuated on a regular basis. This may be necessary for either risk management and control, or because legislation forces the institution to carry out such a regular valuation; typically, both reasons are relevant. For a bond paying fixed rate coupons, this transcription will contain the currency, maturity date, coupon rate and coupon frequency, day count conventions and, for credit or limit ...