Deutsche Börse has completed a pilot study into the application of quantum computing to calculate proprietary business risks, in a use case that could enter full production within the next three years, according to independent experts.

A study published today (March 10), co-authored by Carsten Schäfer—the exchange’s risk manager for business risk and IT operational risk and a supervisory board member—and fintech firm JoS Quantum, describes a quantum computation for risk sensitivity analysis, in which DB estimated the business risk it faces from lost revenue, higher costs, changes in the competitive environment, or regulatory initiatives.

All are enterprise risks the exchange operator currently models using classical computing methods, but adding a sensitivity analysis of each input to its model would require days of classical computing calculation time. Such a delay would deliver little business benefit, if the advice was to hedge against shifting interest rate dynamics, for instance.

If the bourse wanted to run a sensitivity analysis on combinations of several inputs, the calculation would require up to 10 years of Monte Carlo simulation. But the researchers found a quantum computer would take less than 30 minutes for the same task.

“The toy example constructed here for the purpose of illustration is a political crisis in a specific country, potentially triggering a credit rating downgrade or a change in the foreign exchange rate—all of which can be interlinked—and the various financial impacts on the group,” says Rory McLaren, technology strategist at DB, in an interview. “We’re mostly interested in the sensitivity analysis, as a small change in a couple of inputs can have a dramatic change on the output.”

Quantum’s huge potential in risk modelling is well documented, with quantum technology allowing complex calculations with large numbers of variables to be run simultaneously in near-real time. Some Dutch banks, for example, believe it can help produce quicker, more accurate results in regulatory stress tests. Goldman Sachs is experimenting with quantum to more accurately gauge the value and risk in a portfio.

Further applications in instrument pricing, credit risk, and operational risk have been explored with quantum technology by other market participants, and McLaren says further use cases are likely to be investigated within DB.

The outputs of the model can be used to inform various business areas in order to allow them to better manage the changing risks

Rory McLaren, Deutsche Börse

“We will evaluate in which areas future use cases could deliver results to gain a better understanding of its applicability,” he says. “There would certainly be a competitive advantage if you can start to use quantum hardware, as opposed to classical, when that hardware is suitably mature.”

Deutsche Börse’s current business risk model—using classical computing methods and on-premise servers—has roughly 400 parameters. It can be calculated in “tens of minutes”, says McLaren: “The outputs of the model can be used to inform various business areas in order to allow them to better manage the changing risks.”

JoS Quantum uses a new quantum algorithm that combines ‘amplitude estimation’—a quantum version of a classical Monte Carlo simulation—and ‘Grover’s algorithm’, a quantum search algorithm, to run the sensitivity analysis. Small bits of the model were executed on IBM’s Vigo machine, accessed via the cloud.

To test the scaling under ‘realistic’ conditions, a toy version of the model used seven risk items and six ‘transition’ probabilities. The transitions represent the likelihood that risk items could trigger further risk events. JoS Quantum’s simulation used about 20 quantum bits, or ‘qubits’, in total.

It’s something that touches on a possible quantum advantage—doing something a classical computer cannot—that hopefully can be realised very soon

Quantum expert

While a bit within a classical computer can be in one of two states—storing either a one or a zero—a qubit can exist in both states simultaneously, a phenomenon known as superposition. This means quantum computers can handle a vast number of calculations. But the future success of the technology hinges on improving its reliability.

Classical bits retain their states—zero or one—because the materials and electronics in classical computers provide stability. In quantum computing, errors can be caused if qubits get too warm, or as a result of stray microwaves and photons, or manufacturing defects. That means they need to be ‘corrected’ to become ‘logical’.

In full production, DB says it would look to model 150 risk items and 250 transition probabilities. For that, the system would need around 200 error-corrected qubits—out of the range of current technology, but still low compared with the number required for other quantum computing applications.

A quantum expert at one bank describes the paper as “rigorous and repeatable, so anyone can verify the result”. He says the exchange could put the model into full production in the second half of this decade.

“Most importantly,” he says, “the authors provide a realistic estimate of how many logical qubits would be needed to execute the whole program on a quantum chip. Maybe that is three years or four years away. It’s something that touches on a possible quantum advantage—doing something a classical computer cannot—that hopefully can be realized very soon.”

IBM announced last year it was aiming to build a computer with 1,000 qubits by 2023, at which point it would be possible to run software that would error-correct qubits, and make the overall system more reliable.

Niklas Hegemann, co-founder and managing director of JoS Quantum, says: “Current technology is in the range of 50 to 100 so-called noisy qubits. But they are not error-corrected, and you need to put a lot of noisy qubits together to form one logical qubit.

“What we are providing here is a framework that can show quantum advantage in the future with these machines,” he adds.

Research into quantum technology’s problem-solving potential is growing. The UK government, through its UK Research and Innovation arm, this month announced investment of £153 million ($212 million), supported by £205 million from industry, to develop quantum technologies that will have a significant impact on financial services. A project involving Standard Chartered Bank was awarded £6.4 million in 2020 to establish an advanced commercial quantum computer in the UK.