Explore University College London’s Bold Plan to Automate Financial Regulation and Save the City a Fortune

Professor Philip Treleaven and his financial computer science team at University College London have impressive form when it comes to executing grandiose projects. About 25 years ago, UCL successfully built the first insider dealing detection system for the London Stock Exchange. The first time the system ran it identified seven people that looked like they were engaged in insider dealing; five of them were prosecuted, and the new innovation was splashed across the front page of the Financial Times. About 10 years later, Treleaven's team collaborated with Deutsche Bank on the first automated fixed income trading system. The time has come for another big ticket UCL-led innovation: the automation of financial regulation. The shift into what could be called core "regtech" (a term coined by Treleaven) is conceived as an assemblage of technology-enabled components. There are five distinct parts: an intelligent regulatory adviser as a front end to the regulatory handbook; automated monitoring of online and social media to detect consumer and market abuse; automated reporting using online compliance communication and big data analytics; regulatory policy modelling using smart contract technology to codify regulations and assess impact before deployment; and automated regulation employing blockchain technology to automate monitoring and compliance. 'Hello Alexa, How Do I Register a Bank?' From a firm's perspective, the first major pain points is registering, which Professor Treleaven pointed out takes an inordinate amount of time to actually do, is hugely expensive and involves being asked a lot of stupid questions. "So we are trying to build a front end to the FCA [Financial Conduct Authority] Handbook that basically will interact with the registrant and try and fill in the forms for them; and then you can just give this to a specialist lawyer who can say: 'this is correct; this is wrong'," he said. The FCA has seen its workload double without really increasing its staff: previously the regulator monitored 25,000 large and medium sized firms, and now has to supervise an additional 21,000 small firms. An obvious way to supervise this ecosystem, with its potentially dodgy individuals, is to monitor social media for signs of financial abuse, said Treleaven. regtech ucl fintech Signage is seen on one of the main campus buildings of University College London (UCL), part of the University of London, Britain, April 24, 2017. REUTERS/TOBY MELVILLE Treleaven lamented the sort of political knee-jerk regulation Europe is famous for, which usually consists of implementing lists of things with the result that systemic risk actually increases thanks to all the added complexity. "One of things you should be able to do with automation is actually to simulate regulations before you deploy them," he said. "The other great challenge, which isn't really discussed, is using some sort of behaviour analytics to try and encourage people to behave well. Asia's Regtech Revolution Treleaven said that while the FCA has done a very good job to date in providing thought leadership in the area, it's becoming clear that the Monetary Authority of Singapore (MAS) and the Hong Kong SFC now see regtech as a key economic driver and are earnestly pumping money into it. He said: "Britain needs to pump more funding into this area. You've got the Asians who are saying, this can give us economic advantage so we will pump huge amounts of money into experimentation and certainly MAS in Singapore are really gung-ho about what they are doing." Treleaven added that the Americans are uncharacteristically on the back foot with regtech, while the European Union has written some papers on it but doesn't seem to really understand what's going on, and isn't doing any experimentation to speak of. MAS, by contrast, should be seen as a blueprint for how to encourage innovation; multi-tiered and distributed across the ecosystem and not just confined to a sandbox. "We can learn from these guys and it's a good thing to copy, but it's also quite difficult because most of the politicians are quite removed from innovation and technology. You have a lot of companies setting up accelerators and investment funds and I have been putting students into companies, which is quite low risk. And also suggesting we look at some way of injecting an R&D lab into some of these companies supported by the universities, rather than vice versa." Code Is Law The UCL team has carefully considered various approaches to coding up smart contracts that might be used to capture and automate compliance monitoring. To be clear, a "smart contract" is term of art that generally refers to the automation of some business process that is verified and enforced by its participation on a shared ledger. To date, the most advanced version of the concept is on Ethereum, where the public chain's native currency, ether, locks in and disburses value from contracts. UCL in association with R3 and Barclays Investment Bank is part of important work around so-called "Smart Contract Templates", which combine natural language and code in machine readable formats, and which are being tested with standardised financial instruments such as interest rate swaps. There are further plans on the horizon at UCL to develop a new financial smart contracts language. Related Stories London Leads the Way in Regtech Innovation Hedge Funds Opt for Open Source and AI Goes 'Fintech' Is "The Wisdom of the Crowd" the Future of Governance? Big Data in Markets Is Getting the 'Fintech' Treatment Professor Treleaven sees no need to create a new language to encode and automate regulations, however. Out of the traditional languages, there's a choice between using functional, declarative programming languages like Haskell, which are concise and amenable to mathematical analysis and verification, versus object-oriented languages like Java and Python. The latter are more popular with developers for general programming, but because they use statements that tell the computer what to do and change a program's state, this introduces semantic complexity and potential side effects. Treleaven hopes to harness the best of both worlds, however, by using a declarative subset of Python; a "multi-paradigm" language. "It turns out programming in a declarative subset of Python brings quite a few benefits," he said. "From a mathematical viewpoint, this includes formal provability, modularity, composability, and ease of debugging and testing, whereas pragmatically, the benefits consist of the wealth of associated code and seamless analytics."