News & Events
Blog

Comparing Quantlab and Matlab for financial calculations

All through my university years and for some years thereafter, working as a financial consultant, I have pushed the limits of what my Matlab could do for me. Such great relationships can only be superseded by having your own baby.

Author
Robert Thoren
Published
January 20, 2005
This is some text inside of a div block.

Let me start this comparison on a personal note by admitting that I am Matlab’s greatestfan. All through my university years and for some years thereafter, working as afinancial consultant, I have pushed the limits of what my Matlab could do for me. Suchgreat relationships can only be superseded by having your own baby. Quantlab is such abrainchild, conceived by several brilliant financial engineers and programmers atAlgorithmica Research together with some key customers (not including myself amongthe brilliant). The new baby talks my language, has a financial engineer’s brain, andlimbs that fit seamlessly into historical and realtime information, and I get all thiswithout compromising too much on what I love in Matlab. Now, let me get down to business.

The overriding design goal for the Quantlab development environment is to empower the financial engineer with out-of-the-box building blocks on which to test new ideas. Point being – the user should not have to invent the wheel over and over again just to handle user interface interaction and basic financial math, and the result must be a usable application, not only to the developer, but to his non-engineering colleagues at the trading and sales desks. Graphs and tables should be formatted in a way we have become accustomed to in Bill Gates world, i.e. through visual click-and-change.

This document will deal with a comparison of presenting a simple interest rate curve based on a Nelson-Siegel model as a graphic plot. Why take this example? In many applied research projects, the task of building sufficiently smooth interest rate curves from which to extract spot and forward yields is not the main objective. It is merely a necessary building block for more applied analysis. I found a piece of representative and excellent written Matlab code on the web coming from an academic researcher having just such a problem. The only alteration to his Matlab code that I have made has been to reduce the spacing for readability and the financial input data in the example has been updated to present time. In the next part I will try to show the difference in using a general-purpose high-level development environment and one specifically targeted to get the job done.

[download complete text as pdf]