UBS found no advantage in quantum computing—ex data chief

UBS abandoned a multi-year effort to utilize quantum computing for trading after concluding that it offered no significant advantage over existing technologies, according to the person who led the project.

“We ran Monte Carlo simulations against market data, we ran volatility curves, and what we were trying to do was simply use the technology to get an exponential ‘back out’—something that would give us a competitive advantage,” said Lee Fulmer, who was chief data officer and head of the innovation lab at UBS until March.

The aim was to speed up the bank’s models, but Fulmer said UBS’s experiments failed to produce any meaningful results: “In investment banking you live and die by microseconds. The end result of all of that effort was that we found we weren’t getting a substantive uplift.”

The end result of all of that effort was that we found we weren’t getting a substantive uplift

Lee Fulmer, formerly of UBS

Fulmer was speaking at a conference hosted by The Economist on May 17.

Quantum computers can complete calculations much faster than classical computers by exploiting the properties of subatomic particles, such as electrons and photons. A bit, or unit of information, in a digital computer can be in one of two states: either a one or a zero. A quantum bit, or a qubit, can exist in both states simultaneously, meaning the computers can theoretically handle a far higher number of calculations.

Proponents of the technology claim it can turbo-charge existing financial models, including those based on machine learning. But Fulmer, who is currently independent chair of the Bank of England’s reporting and data standards transformation board, and previously worked at JP Morgan and IBM, said he is yet to find a convincing use case in finance despite “playing with quantum computers now for nearly 15 years now”.

It is not for a lack of trying. UBS went to great lengths to test the uses of quantum computing—even conducting experiments with UK spy agency GCHQ in 2019. Another project, carried out when Donald Trump was president, involved feeding Twitter data into a quantum computer to analyze “how that was affecting markets and commodities to predict literally in microseconds the correlation to volatility”.. The results were disappointing. “We weren’t able to get any substantial uplift over what our existing algos were able to do,” said Fulmer.

His conclusion is that financial models are “linear” and will not benefit from the capabilities of quantum computing. “That’s the way we look at problems. And actually, as human beings, we are binary—we are not wired to be able to think about two states at the same time,” Fulmer said. “I think until we get to a phase where we have true business problems that can be bifurcated in real-time, we’re not going to be able to ask the right questions.”

Other banks that are in the early stages of experimenting with quantum computing are more bullish on its potential. A recent JP Morgan study found the technology could nearly double the effectiveness of its machine learning based hedging strategies. Researchers at IBM and Barclays have claimed it can dramatically speed up securities settlement cycles. ABN, ING and Rabobank have trialed the use of quantum computing in regulatory stress tests, while Standard Chartered teamed up with US space agency Nasa to explore uses in portfolio allocation, and exchange operator Deutsche Börse is testing whether it can be used to model business risks.

Asked directly if his view was that quantum computing had no real value in finance, Fulmer responded: “Yes. If we look at how quantum computing started to evolve out of the ‘90s, alongside blockchain, alongside VR [virtual reality], we have the same sorts of challenges with those. They are technical solutions for business problems that don’t exist.”