Year: 2026
Type: preprint
Abstract: We study renewable-energy portfolio selection using quantum approximate optimization algorithms (QAOA) with penalty-based X, cardinality-preserving XY, and sector-preserving XY mixers, benchmarked against exact and mixed-integer classical solvers. On 18 small primary problems, the mixers recovered optimal portfolios in 6, 13, and 15 cases, respectively; classical methods solved all 18 faster. The penalty-X baseline sampled almost uniformly among feasible portfolios, and a 72-run penalty sweep offered no reliable improvement. A second exploratory study used eight listed issuers across three dated problems; corrected 64-shot readouts recovered optima in 5 of 18 X settings and 16 of 18 XY settings. Financial backtests remained retrospective. Between May and October 2026, the baseline strategy returned −3.41%, while renewable/SPY blends returned +2.50% and +5.46%, below SPY’s +8.41%. Improved quantum sampling did not establish a speed advantage, independent validation, or investment alpha.
Author: Ishanth Guntko
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7223262615