Despite all the good things going for it, the pricing tool is still quite far from being a trading system because it lacks a few crucial features. We discussed these shortcomings earlier. We may be able to add on the missing features and call the pricing tool a full-fledged trading platform, but it may not be a wise course of action. A far wiser modus operandi would be to start from scratch, by talking to the business units in order to understand and document their requirements and by designing a robust and scalable architecture to support them.

Many of the ideas and concepts in the pricing tool architecture are useful, and we should keep them in mind as guiding principles. In particular, a user interface that stays agnostic to the intricacies of the structure it is displaying is a powerful idea. Such script-driven GUIs can greatly improve time-to-market and make release cycles short and painless. Another powerful concept is the clean separation of the pricing logic so that the end user can augment the program without touching its core components. This feature, if implemented well, also cuts down the time-to-market – one of the basic motivations behind embarking on an in-house trading solution.

For the sake of completeness and at the risk of repeating ourselves, we will go over the missing features of the pricing tool once more. The most basic missing feature is user authentication and security. Then comes the ability to manage the trade life cycle and maintain ...

Get Principles of Quantitative Development now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.