Advanced Computing in the Age of AI|Friday, June 5, 2020
  • Subscribe to EnterpriseAI Weekly Updates:  Subscribe by email

GPU-Driven Near-time Valuation Coming to Derivatives Market – Microsoft 

Microsoft Azure and an AI startup named Riskfuel – along with Nvidia GPUs – have teamed to deliver deep learning based near-time valuations of over-the-counter (OTC) derivatives, a capability long established in exchange traded stocks, treasuries and currencies but new to derivatives, according to Microsoft. This could be welcomed by traders in this $500 trillion market who typically rely on overnight batch processing for analysis of the complex pricing factors that impact valuations.

The cloud-based capability, adopted by Scotiabank, runs Riskfuel’s accelerated derivatives deep learning model on Microsoft’s Azure ND40rs_v2 (NDv2-Series) Virtual Machine instances utilizing NVIDIA V100 Tensor Core GPUs. Ian Finder, Microsoft senior program manager, accelerated HPC infrastructure, said in a blog the system “offers 20 million times faster valuation of derivatives” than convention models run on CPU-based on-prem clusters.

“In times of increased volatility,” Finder said, referring to the COVID-19 pandemic, “traders and their managers need to know the impacts of market conditions on a given instrument as the day unfolds to be able to take appropriate action. Reports reflecting the conditions at the previous close of business are only valuable in calm markets and even then, firms with access to fast valuation and risk sensitivity calculations have a substantial edge in the marketplace.”

He said that unlike exchange-traded instruments, “where values can be observed each time the instrument trades, values for OTC derivatives (interest rate swaps, credit default swaps and structured products) need to be computed using complex financial models.” This is usually done with traditional Monte Carlo calculations than run a “probabilistic sweep through a range of scenarios and resultant outcomes or finite-difference analysis,” said Finder.

“Banks spend tens of millions of dollars annually to calculate the values of their OTC derivatives portfolios in large, nightly batches,” he said, “…run on on-premise clusters of conventional, CPU-bound workers — delivering a set of results good for a given day.”

“…as the influence of machine learning extends into production workloads, a compelling pattern is emerging across scenarios and industries reliant on traditional simulation. Once computed, the output of traditional simulation can be used to train DNN models that can then be evaluated in near real-time with the introduction of GPU acceleration,” Finder said.

Comparison of traditional model running versus Riskfuel model (source: Microsoft)

The first stage of the trial consisted of generating 100,000,000 samples for training data, said Finder, by running the traditional model repeatedly with inputs to be approximated by the Riskfuel model. Azure engineers and Riskful then measured the performance of a Riskfuel model on an Azure ND40rs_v2 instance against traditional CPU-driven methods. The study found 20M+ times performance improvement, Finder reported. For portfolios with 32,768 trades, the throughput on the Azure ND40rs_v2 is 915,000,000 valuations/second, he said, compared with 32 valuations/second using the traditional model running on CPU-based VMs.

“These results clearly demonstrate the potential of supplanting traditional on-premises high-performance computing (HPC) simulation workloads with a hybrid approach: using traditional methods in the cloud as a methodology to produce datasets used to train DNNs that can then evaluate the same set of functions in near real-time,” said Finder.

The Azure ND40rs_v2 has eight NVIDIA V100 Tensor Core GPUs, each with 32 GB of GPU memory, and with NVLink high-speed interconnects, delivering one petaFLOPS of FP16 compute, according to Microsoft. Finder said the derivative valuation modeling instances leverage the system’s floating point performance “to achieve the highest batch-oriented performance for inference steps, as well as the greatest possible throughput for model training.”

“By migrating to the cloud, we are able to spin up extra VMs if something requires some additional scenario analysis,” said Karin Bergeron, managing director and head of XVA trading at Scotiabank, which has implemented Riskfuel models in its derivatives platform. “Previously we didn’t have access to this sort of compute on demand. And obviously the performance improvements are very welcome.”

Riskfuel’s founder and CEO Ryan Ferguson said the valuation model could strongly impact on derivatives market.

“The current market volatility demonstrates the need for real-time valuation and risk management for OTC derivatives,” he said. “The era of the nightly batch is ending. And it’s not just the blazing fast inferencing of the Azure ND40rs_v2 Virtual Machine that we value so much, but also the model training tasks as well. On this fast GPU instance, we have reduced our training time from 48 hours to under four.”

Add a Comment

Do NOT follow this link or you will be banned from the site!
Share This