Banca Carige Integrates a MATLAB Based Valuation Library with Its Enterprise Pricing and Risk Platform
“We consider MATLAB to be the best choice for mathematical modeling and numerical computation. MATLAB gives us the unmatched flexibility to embed our algorithms in automated finance systems as add-ins for Excel or components in a .NET framework.”
Challenge
Solution
Results
- Transparent, flexible pricing model deployed
- Products priced in real time, not weekly
- Integration with existing systems streamlined
Gruppo Banca Carige is among the largest banks in Italy, with more than 1100 branches and 2 million customers. In addition to the parent company, Banca Carige, the group includes five other banks, two insurance companies, and an asset management company. Within Banca Carige, the Financial Administration team is responsible for valuation and risk analysis, including pricing the instruments and derivatives in the bank’s own portfolios and those of its clients.
A renewed focus on transparency in the financial services industry led this team to replace a proprietary solution for pricing instruments with a library, called MatFin, developed in-house using MATLAB®.
“Moving to MATLAB gave us the freedom to develop and improve our own algorithms and compare their results with industry benchmarks,” says Paolo Raviola, project manager, Banca Carige. “From our MATLAB code we compiled a .NET component that we integrated with our enterprise structured products platform, enabling us to automate processes and reduce manual steps.”
Challenge
Previously, Banca Carige used a third-party add-in for Microsoft ® Excel ® to price derivatives, bonds, and other complex instruments. This add-in required the team to perform numerous steps manually in a time-consuming,
error-prone process. In addition, it was inflexible and could not be modified. “No third-party supplier can provide a solution that meets the specific needs of every bank,” says Raviola. “If we requested a change or a new feature, we had no way of knowing when or even if it would be implemented, or at what cost.”
The team wanted to use in-house expertise to develop their own pricing libraries. Once they had tested the libraries, the team needed to integrate them with the Misys Kondor Structured Pricing (KSP) system, which was used throughout the bank. “The natural solution was to extend Misys KSP with pricing models that we developed ourselves, but to achieve that integration we needed a way to create .NET components,” says Simone Ligato, senior financial analyst at Banca Carige.
Solution
Banca Carige used MATLAB to build and deploy pricing models for a wide range of financial instruments, including interest rate swaps, interest rate options, equity and index baskets, and inflation linked options.
Using MATLAB, Financial Toolbox™, and Financial Instruments Toolbox™ the team developed their MatFin libraries by applying a variety of methods and financial models, including Black-Scholes, Monte Carlo, and Cox-Ross-Rubinstein.
Date conversion and cash flow analysis were performed using Financial Toolbox. The libraries include algorithms that price cash flows from a set of zero curves using Financial Instruments Toolbox capabilities. The team also developed MATLAB algorithms for bootstrapping interest rate curves in accordance with the Bloomberg® specification for discount factors.
To verify their pricing models, the team compared the results produced by their libraries with market prices from Bloomberg and several financial counterparties.
Using MATLAB Compiler™, they packaged their MATLAB algorithms as a Microsoft Excel add-in to help validate their new libraries and prototypes.
Finally, they used MATLAB Compiler SDK™ to deploy their algorithms as a .NET component, which they integrated with Misys KSP.
The MATLAB pricing libraries are in production at Banca Carige, and the team is currently developing additional libraries for other derivatives products.
Results
Transparent, flexible pricing model deployed. “Because we have developed our own pricing models in MATLAB, we know exactly how the results are produced, and we can make improvements at any time,” says Pier Giuseppe Giribone, Ph.D. “Our previous system was a black box that required programming expertise to use yet did not allow us to make modifications.”
Products priced in real time, not weekly. “By integrating MATLAB models with our enterprise systems as .NET components, we have automated several processes that previously required manual steps,” says Raviola. “As a result, we can now price our products daily instead of only once a week, and we can deliver timelier results to several analysts and managers throughout the bank.”
Integration with existing systems streamlined. “With MATLAB we can develop advanced algorithms and create Excel addins and .NET components, as well as Java™ classes and C++ code, from those algorithms,” says Raviola. “This unique capability makes it easy to incorporate complex financial computations into our existing banking systems.”
