Benchmarking Quantum Algorithmic Resilience for CVaR Portfolio Optimization: The Expressibility-Coherence Trade-off

Battle Count: 60

The paper is highly relevant to quantitative finance and risk management, specifically addressing CVaR-based portfolio optimization which is critical for institutional risk management. However, its practical relevance to active quantitative trading is limited because: (1) it demonstrates that current NISQ hardware cannot solve dense financial optimization problems at meaningful scale, (2) the 10-16 asset limit is far below practical portfolio sizes, (3) no quantum advantage over classical methods is demonstrated, and (4) the findings are primarily a negative result showing hardware limitations. The expressibility-coherence trade-off insight is valuable for future quantum finance algorithm design.

Implementation requires: (1) quantum circuit design for both HE-VQNN and WS-QAOA with custom warm-start initialization, (2) classical preprocessing pipeline for the hybrid proxy matrix (VaR threshold computation, tail scenario extraction, covariance matrix computation), (3) QUBO-to-Ising Hamiltonian transformation, (4) transpilation to IBM heavy hex ISA with Sabre routing and peephole optimization, (5) integration with SPSA/Nelder-Mead classical optimizers, (6) access to IBM Quantum hardware (ibm_fez), and (7) exact classical eigensolver for ground truth comparison. The quantum-classical hybrid architecture and hardware-specific transpilation add significant complexity.

The paper uses publicly available NIFTY 50 data, the open-source Qiskit framework, and IBM's public ibm_fez processor. Detailed transpilation metrics and circuit parameters are provided. However, quantum hardware results are inherently stochastic (4096 shots), the exact SPSA/Nelder-Mead hyperparameters are not fully specified, and the hybrid proxy matrix preprocessing pipeline lacks complete code-level detail. The scaling analysis methodology is described but exact reproduction requires access to the same IBM backend at similar calibration states.