Collaborate to innovate: Quants

Collaborate to innovate: Quants

Back in 2014, I was involved in an implementation project with a large Asian bank. The goal was to integrate their in-house pricing library with our structured product solution. We set the whole front-to-back workflow including limits and Value at Risk (VAR), and designed a risk simulation tool.

The main challenge we faced was how to address performance, as many scenarios are required to perform VAR and risk calculations. We used a GRID computing approach to parallelize as much as we could, but rapidly came up against network and machine limits that had to be overcome. While we achieved what was necessary within this use case, we started to think about how we could help banks be more efficient, as the number of scenarios they need to run is increasing exponentially with today’s regulatory demands.

According to Moore’s Law, CPU power almost doubles every 18 months. Where it has been respected in the last decades by increasing clock speeds and reducing transistor size, the trend in recent years is now to implement more cores – with CPU having dozens of cores, while GPU has thousands of them. Using multi-core in software development is key, but optimizing the code to benefit from it is complex.

That’s why we created a platform for parallel processing. It allows Quants to focus on their pricing algorithms, while the platform ensures optimized performance. Behind the scenes, the platform does the heavy lifting to convert code into various low-level languages, to take care of memory management and data alignment to maximize the performance, regardless of the target device (CPU and GPU).

Our internal Quants have been working with the plaform for a while now and we have seen impressive performance on standard, derivative and structured products. But now is the time to bring this technology to Quants, academics and Fintechs outside of our organization. How?

The Financial Model Builder is a cloud-based framework running on Microsoft Azure. It allows Quants to develop, test, debug and execute numerical algorithms and pricing models on our platform technology.

It was presented at the QuantMinds event in Lisbon in May 2018.

What’s included?

  • an online Integrated Development Environment (IDE) to develop, test, debug and execute your financial library
  • contextual tutorials, dynamic documentation and videos
  • collaboration using GitHub 
  • REST end points to call your financial library from FusionFabric Application Builder or any REST client such as the Excel plugin provided
  • an API management directory to control and obtain metrics from the REST calls
  • access to powerful CPU devices via Microsoft Azure 

Combined with the FusionFabric Application Builder, you can build your own application to perform pricing, with different scenarios, and with a great user experience to visualize the results.

You can have a look to the portfolio valuation tutorial to get an idea.

The Financial Model Builder has been tested by Fintechs, including GMS and Efficiency MC. It has also been used as part of a study course at University College of London. 

User feedback

“UCL Graduate students have gained tremendous exposure to business and Quant teams at major global FIs and risk management solution providers like Finastra. Students have had hands-on experience pricing and modelling risk in a constantly evolving regulatory environment. Their first hand understanding often becomes apparent in interviews when students describe use cases they have worked on. 
We are proud and cherish these win/win partnerships.”
- Donald Lawrence, Computational Finance Director at UCL.

Now we’re welcoming you to take a look at our vision.

Get in touch here:

Laurent Chollet is Product Manager at Finastra, working on the platform. He joined the company in 2006, having previously worked for Société Générale bank. He holds a Master’s degree in Engineering and the Financial Risk Manager (FRM) certification. During his free-time, Laurent practices Freediving!