Client: A leading global financial product and data company.
Client wanted to completely revamp their performance attribution platform to meet
the demands of the market. Specifically, they wanted to
1. Make the platform highly scalable to meet performance goals with increasing
2. Build a market leading visualization capability to slice and dice performance
3. Add new performance attribution models to the platform.
The older performance attribution platform suffered from scalability issues as it
recomputed results for client jobs every day. However, most client jobs only
extended the attribution period by a single day. Thus, it was possible to store
previous results, perform incremental calculations and then link with previous day
results to generate full period results.
A result storage capability was developed to store partial results. The results were
stored in a generic fashion in an Oracle database to support results from all
attribution models. A fast linking engine was built to link partial results on the fly to
generate full period results.
Spotfire was selected as the visualization platform after a thorough analysis of
various BI tools. A custom extension was built in Spotfire to dynamically query the
new linking engine for full period results. Thus, a new visualization capability was
built leveraging partial stored results.
The new visualization capability received significant industry attention and became a
primary selling point for the product. Also, the introduction of storage capabilities
led to reduction of job run time by 80-90% and thus resolved the scalability issues
with the older platform.
Java, .NET, Spotfire, GXT, MS SQL Server, Oracle