Hybrid Quantum Solvers Show Promise
Hybrid quantum-classical solvers are demonstrating practical advantages in portfolio optimization. Researchers from IBM claimed in a recent podcast that these systems can achieve 10-15% faster convergence on large-scale trading problems with complex constraints. This progress suggests quantum's near-term impact will be concentrated in solving optimization problems that are intractable for classical methods alone.
- The "hybrid" in "hybrid quantum-classical" refers to a feedback loop where quantum processors handle tasks suited for quantum mechanics, like optimization, while classical computers manage control processes, error correction, and data analysis. This approach is necessary in the current era of Noisy Intermediate-Scale Quantum (NISQ) computers, as it reduces the resource requirements and compensates for noise with classical computation. - Variational Quantum Eigensolver (VQE) and Quantum Approximate Optimization Algorithm (QAOA) are two prominent hybrid algorithms used for portfolio optimization. These algorithms use a quantum computer to prepare and measure quantum states, while a classical computer optimizes the parameters of the quantum circuit in an iterative loop. - Financial institutions are actively partnering with quantum computing firms to explore these applications. For instance, Citi Innovation Labs collaborated with Classiq to improve portfolio optimization using QAOA, and Multiverse Computing worked with BBVA and Bankia to use D-Wave's hybrid solver for identifying investment strategies with high Sharpe ratios. - A recent study by IBM and Vanguard on a bond ETF portfolio construction problem demonstrated that a hybrid workflow using a sampling-based Variational Quantum Algorithm (VQA) on an IBM Quantum Heron r1 processor consistently outperformed a purely classical approach as the problem size increased. - Beyond portfolio optimization, hybrid quantum approaches are being explored for collateral optimization to reduce costs and improve liquidity, as demonstrated by Multiverse Computing with Crédit Agricole CIB and BBVA. Additionally, institutions like Intesa Sanpaolo and IBM are investigating Quantum Machine Learning (QML) for enhanced fraud detection. - The hardware powering these advancements includes superconducting qubits from companies like IBM, Google, and SpinQ, as well as trapped-ion systems known for high precision and long coherence times. These processors are still limited in qubit count and are susceptible to errors from environmental "noise," making the hybrid approach essential. - A significant challenge in implementing these systems is the scarcity of talent with expertise in both quantum physics and finance. Furthermore, integrating these novel systems with existing legacy IT infrastructure in the banking sector is a considerable and costly hurdle. - The long-term threat of "harvest now, decrypt later" attacks, where encrypted financial data is stolen today to be decrypted by future, more powerful quantum computers, is a major driver for financial institutions to invest in quantum-resistant cryptography alongside their exploration of quantum optimization.
