An Innovative Knowledge Learning Adaptive Quantum-inspired Algorithm for Trend Ratio-Based Portfolio Construction Model

Quantum-inspired optimization is an emerging computational intelligence technique that achieves quantum advan-tages through inspiration from the mechanisms of quantum properties. Parameter control is vital for optimization methods to address complex problems with extensive solution spaces. However, static parameters lack adaptability, leading to issues like premature convergence, becoming trapped in local optima, and failing to uncover the global optimum. Therefore, we develop a novel adaptive mechanism to dynamically adjust parameters during algorithm execution to strike a remarkable balance between exploration and exploitation. This study first proposes the knowledge-navigated quantum-inspired tabu search (KNQTS) algorithm, which acquires insights from the ongoing process to adjust the environment-sensitive parameters adaptively and non-linearly. Differences between inter-iteration solutions can signal the extent of convergence and guide bidirectional adjustments of a parameter, facilitating the handling of com-plex optimization problems. KNQTS is applied to tackle real-world application portfolio construction problems in the financial technology field. Comprehensive experiments are conducted to demonstrate the substantial potential and robustness of KNQTS in rapidly identifying better solutions with higher probabilities in shorter durations compared to the other state-of-the-art methods. Statistical tests substantiate the significant superiority of KNQTS over these comparative algorithms.