Portfolio Optimisation with Adiabatic Quantum Computing
Join the first of a new series of Finance Industry specific quantum computing talks!
Alexei Kondratyev, Managing Director, Financial Markets at Standard Chartered Bank, is going to be speaking about his work on portfolio optimisation using adiabatic quantum computing. Alexei thinks there are many possible use cases for quantum computing in finance and that existing quantum annealers are already powerful enough to execute meaningful research projects
At the moment his research is focused on portfolio optimisation. Standard Chartered partners with NASA-USRA on investigating capabilities of the D-Wave 2000Q quantum annealer. They presented their first results at the QuantMininds conference in Lisbon in May 2018 and the final results at the Adiabatic Quantum Computing Conference in California at the beginnig of July 2018 – Standard Chartered is a one of the major sponsors of this conference.
The following two short videos, filmed at the QuantMinds conference, are of Alexei being interviewed on the subject:- https://www.youtube.com/watch?v=BtosVtB7ZFI (4:23) https://www.youtube.com/watch?v=dQCbhnDnmvg (1:55)
Alexei has already attended some of our talks and so it will be great to have him kick off this special series.
We look at several problems of industrial value that could be represented with less than 100 binary variables coupled in fully connected QUBO (quadratic unconstrained binary optimisation) form. We employ state-of-the-art programming techniques to evaluate the performance of the D-Wave 2000Q quantum annealer on these problems and the possible outlook of overcoming the current limitations on the use of the D-Wave architecture for optimisation problems requiring heavy embedding.
We extend the mean-variance portfolio optimisation approach and solve sample portfolio optimisation problems with adiabatic quantum computing and with classical benchmark based on Genetic Algorithm. Our aim is to use the principles of the Modern Portfolio Theory as our starting point while allowing for more general dependence structure. The quadratic form of objective function makes it suitable for being solved on the D-Wave machine, which operates on QUBO problems. The QUBO formulation of optimisation problem also allows us to express discretionary portfolio manager preferences in the form of QUBO coefficients. We also explore questions of algorithm convergence and problem scalability.