Data science in the cloud massively speeds up risk calculations

Data science in the cloud massively speeds up risk calculations

A new economic capital risk model implementation developed by DiUS helped a large financial institution dramatically improve its ability to model scenarios and calculate outputs to better support future capital strategy

To deliver a faster and cost-effective economic capital risk model execution, a new implementation was designed to leverage the power of parallel processing and the elasticity of Amazon Web Services (AWS) cloud computing.  DiUS optimised and re-implemented the risk model algorithm so that it ran in a massively parallel way using AWS’s hosted Hadoop service, Elastic MapReduce (EMR) and leveraged AWS EC2 spot instances to reduce the cost of model execution.  An interface was created so that risk analysts can easily set up and initiate model executions and collect results.

The new model implementation markedly reduced the time taken to process 600 billion calculations required to fully model 10,000 economic scenarios from an estimated 208 days to just 8.5 hours. The risk team is now able to run the full set of scenarios overnight and come in the next day to drill down to investigate the drivers behind variations in the results, as well as passing them on to senior management for more timely business decision making. The new model implementation also provides a scalable platform to support the company’s future capital strategy.

Watch a video presentation of this Big Data case study.

DiUS is an AWS Advanced Consulting Partner.

Get in touch

Full Name



  Send me a copy