The data sciences team for one of the nation’s largest investment brokerage companies came to Corios because they were struggling with numerous inefficiencies that were limiting their overall profitability.
Our client sought to dramatically reduce the cycle time needed to develop new models and to refresh their scores for customers, so that financial advisors could receive daily updates on the best treatments for their customers and grow their most profitable relationships.
The brokerage utilizes over 50 treatment response scoring models to predict customer behavior and direct financial advisors on the most profitable recommendations for their clients. Their platform wasn’t as efficient as desired, requiring months to acquire data and days to execute model scoring routines; and it failed to capture the effects of a massive array of information on customer positions, trades, and online transactions.
Corios worked with the data sciences team to design and implement massively-parallel model training and model scoring routines for three of the data science team’s models. Utilizing the SAS® High Performance Analytics platform on a large database application, Corios dramatically modernized the analytics development and deployment process in a few months.
Corios’ solution enabled our client to execute computations on billions of transactions in a matter of minutes, and increased model performance 10-20% by incorporating detailed transaction information. With their new model deployment framework, the brokerage can now score their models within a nightly window, allowing them to update financial advisors every 24 hours instead of waiting for weeks or months for the next best customer treatment.
Want to dig deeper into how model deployment affects the bottom line of your business? We wrote a whole paper on it!