Since the global financial crisis a decade ago, banks, asset managers, insurers and regulators have been seeking better ways to manage financial risk. The solution may lie in high-performance computing, say EU-funded researchers who have developed novel financial models and algorithms to improve their financial risk management.
© Bits and Splits - fotolia.com
HPCFinance, funded through the EU’s Marie SkÅ‚odowska-Curie programme, combined the expertise of private sector partners and two experienced researchers with 12 early-stage researchers. Together, they developed a range of high-performance computing solutions, bridging the gap between academic research and the practical needs of the financial industry.
“One partner, Cambridge System Associates, has been enhancing solutions for individual asset liability management based on research results obtained in the project,” says HPCFinance coordinator Juho Kanniainen of Tampere University of Technology in Finland. “These solutions can be used to develop detailed prospective financial plans tailored to an individual's financial goals and obligations.”
Their work fused cutting-edge concepts in financial engineering with supercomputing infrastructure to make the regulatory and capital environment less demanding and complex. Their goal was more reliable and optimised quantitative tools, algorithms and simulations for strategic asset liability management, volatility and financial risk modelling, and contract pricing.
Averting a repeat of the financial crisis
Novel risk-management methods and models should reduce the danger of a repeat of the mispricing of assets, credit crunch and liquidity issues that led to the collapse of banks and triggered the 2007-2008 financial crisis.
“More than ever before, accurate computational models and people able to deploy and operate them are needed to improve the robustness of risk management in the financial sector,” Kanniainen says. “Better computational methods and strategies can enforce risk management under a sound regulatory and capital framework, not just for large financial institutions but also for individuals and households trying to meet their financial goals.”
The researchers explored different high-performance computing platforms, including using graphics processing units, distributed computing via central processing units, and field-programmable gate array technology. This high-performance infrastructure provides the computational resources needed to run data-intensive mathematical models for tasks such as gauging risk and volatility, and monitoring interest rates and asset valuations.
As a training network, HPCFinance had a significant impact on the career development of the 12 early-stage researchers and two post-doc researchers seconded to financial institutions. There they learnt how to apply their knowledge in high-performance computing and have since gone on to find employment in the field.
“The project has reinforced knowledge exchange between participants, who profit not only from cross-fertilisation between sectors, but also from synergies and the pooling of resources between participants with similar research goals,” says Kanniainen. “On the one hand, it has enabled high-performance computing specialists to understand the challenges facing the financial domain, while on the other it has encouraged financial institutions to utilise modern technologies in a realistic way to meet the demands of increasingly complex risk-management processes.”
The close linkage between private sector partners and academia has been important in the exploitation of project results through joint research papers and secondments. The arrangement also contributed to the training of researchers with inter-sector knowledge in high-performance computing for financial applications.