Using formal verification to develop higher assurance, more maintainable financial software
Abstract
Formal verification — a sibling of artificial intelligence — uses computers to prove that a design — whether hardware, software or a mathematical proof — is correct. As such, it offers higher assurances than testing, which cannot provide guarantees beyond the test data. As formal’s cost comes down, it is spreading beyond its roots in computer hardware and aerospace software into consumer technology and finance. This paper surveys successful use cases outside finance, as well as more recent applications to finance, including by industry leaders such as Bridgewater and Goldman Sachs.
The full article is available to subscribers to the journal.
Author's Biography
Manfred Kerber is a computer scientist, with three decades of experience in automated reasoning and verification. He is CTO and cofounder of fovefi. Before co-founding fovefi, he worked with Rowat, formally verifying auction properties.
Colin Rowat is an economic theorist, with a focus on game theory and financial risk. He is CEO and cofounder of fovefi.
Neels Vosloo has worked in financial risk management for more than 20 years, covering traded market and counterparty credit risk as a model developer in banking and consulting, and as a regulator, where he also contributed to policy development on market risk capital requirements. Neels’s current focus is compliance with regulatory requirements on traded risk capital models, where his interests include the modelling, validation and compliance requirements of the so-called Fundamental Review of the Trading Book (FRTB).
Citation
Kerber, Manfred, Rowat, Colin and Vosloo, Neels (2019, December 1). Using formal verification to develop higher assurance, more maintainable financial software. In the Journal of Risk Management in Financial Institutions, Volume 13, Issue 1. https://doi.org/10.69554/YOVQ1433.
Publications LLP