The BBVA Research and Patent team has shared the results of a project that ran over the course of last year and which aimed to study the applications of quantum computing in the financial sector. Specifically, they disclosed the results of BBVA-sponsored proofs of concepts (PoCs) that followed six lines of investigation with the objective of identifying those use cases where quantum technology delivers a greater advantage over traditional computing techniques. The PoCs also assessed which solutions available on the market could be used to this end.
As it explores the benefits of using quantum computing in the financial sector, BBVA is following those lines of research, working hand in hand with Spain’s Senior Council for Scientific Research (CSIC), Accenture, Fujitsu, Zapata Computing, and Multiverse.
Lines of research and proofs of concepts
1. Development of quantum algorithms (CSIC)
In addition to hardware improvements, another major challenge facing the business deployment of these technologies is the fact that new algorithms must be adapted to the new computing logic, algorithms that can feed the systems once they are functional and primed to solve concrete tasks. BBVA is working on this front in collaboration with CSIC. Their collaboration on this line of scientific research is producing important advances.
The joint team of researchers has developed algorithms that help select the most relevant variables of a broad set of data, like choosing the assets when building an investment portfolio, for example. These developments have been tested to improve stock index tracking, an investment technique that seeks to replicate the behavior of a stock index by selecting some of the assets within it.
The algorithms that have been developed could be applicable to other fields as well, such as the design of logistics networks or variable filtering in machine learning models. Samuel Fernández Lorenzo, BBVA’s head of quantum algorithm research explains, “With our work we have been able to come closer than ever to using quantum computers or quantum-inspired algorithms for real world applications.”
2. Static Portfolio Optimization (Fujitsu)
The optimization of investment portfolios consists of choosing assets that, when combined, can help customers earn greater returns based on factors such as investor and risk profiles. One way to make this process more efficient is to group a portfolio’s assets into subsets with common risk factors. However, as assets are added to a portfolio — along with the factors that need to be taken into account for their classification — the possible combinations that can be produced multiply exponentially, and in turn so do the number of calculations required to get an optimal outcome.
In collaboration with Fujitsu, and working with BBVA Asset Management, the research team has carried out a proof of concept to determine whether these calculations can be performed more efficiently thanks to quantum technologies. Specifically, the Fujitsu Digital Annealer, a quantum-inspired hardware system that uses traditional algorithms to simulate the technology’s characteristics, was used in the PoC. It was determined that better results could be obtained with this kind of equipment compared to traditional approaches when there are more than 100 assets or factors to be introduced into the calculation.
3. Dynamic portfolio optimization
Dynamic portfolio optimization relies on a significant number of variables in order to determine the best combination of assets. This means that the portfolio’s performance over time, the possible trading fees, and potential impacts on the market price of high volume buying and selling can be calculated. Various tests with different technology providers have been carried out in order to identify how quantum technologies can be used to address this challenge.
Accenture: During this test, a quantum annealing solution from technology provider, D-Wave was used to demonstrate that there are benefits to using quantum techniques over traditional methods when hundreds of assets and/or factors are involved in the calculation. The promising results have convinced the team to continue their investigation of this case with other technologies.
Multiverse: The Spanish startup, Multiverse, collaborated on the second test using two different technological solutions to address the same problem. Both quantum-inspired algorithms and pure quantum hardware from IBM (with associated limitations) were used to perform the tests. The testing is still underway, but the results are promising and will shortly be published in a scientific paper.
4. Credit scoring process optimization
Also in collaboration with Accenture and D-Wave, a proof of concept was undertaken to determine if quantum computing could accelerate the output of credit scoring results compared to existing data analysis systems. The results of this exercise indicate that there could be benefits in cases where there are more variables than are normally used in this type of problem.
5. Currency arbitrage optimization
Currency arbitrage — the looking to profit from buying and selling currencies — is another problem that may lend itself to quantum computing. The window of opportunity for these types of opportunities is very small and powerful processors are required in order to identify and take advantage of this opportunity.
In order to verify if the process efficiency could be improved using quantum technologies, another proof of concept was designed together with Accenture and using the D-Wave technology. The outcome in this case points to possible benefits when the operation is dealing with at least a dozen assets.
6. Derivative valuations and adjustments
Monte Carlo simulations, which use random sampling to simulate the performance trend of different variables, is one of the ways the financial sector calculates the price of derivative products. Derivatives are complex financial products whose values depend on the price performance of other assets. Determining the price of these products is not always straightforward, and in some cases the calculations can be computationally costly.
The BBVA Corporate and Investment Banking (CIB) unit in partnership with the US startup, Zapata Computing, launched a proof of concept to evaluate the use of quantum algorithms applied to the Monte Carlo method in order to determine the price of a derivative instrument with its counterparty risk adjustment. The objective of the test is to analyze if there are benefits to using these techniques, what computing resources would be necessary to yield the improvement, and how do the results scale against the dimensions of the problem.