Stock Exchange Trading Optimization Algorithm Improved Based on Information Technology Investment Analysis for Numerical Optimization and Practical Problems

Stock Exchange Trading Optimization (SETO) is a population-based metaheuristic with a compact market-inspired search framework, but its performance may be limited by uneven random initialization, coarse Relative Strength Index-based search-state identification, and excessive concentration of trading resources. To address these limitations, this study proposes an improved SETO, termed ISETO, by reformulating concepts from information technology investment analysis into population-search mechanisms. First, an IT investment portfolio screening initialization strategy jointly considers solution quality and spatial diversity to improve initial population coverage. Second, a discounted-return and downside-risk dual-state search strategy evaluates the magnitude, temporal relevance, and risk of historical improvements to adaptively coordinate exploration and exploitation. Third, an entropy-regularized real-option capital reallocation strategy redistributes search resources according to solution quality, capital efficiency, technical debt, and population diversity, while reinvesting abandoned candidates into underexplored regions to mitigate stagnation. The dominant computational complexity remains approximately O(TmaxND). ISETO is evaluated on 22 benchmark functions from the IEEE CEC2020 and CEC2022 suites under 10- and 20-dimensional settings and compared with nine representative metaheuristic algorithms. Under the conventional iteration-based setting, ISETO achieves the best or tied-best mean results on all CEC2020 functions at both dimensions, all 12 CEC2022 functions at 10 dimensions, and 11 of the 12 CEC2022 functions at 20 dimensions. Additional experiments under an identical objective-function evaluation budget confirm that these advantages are not simply caused by additional function evaluations. Holm-adjusted statistical tests and Cliff’s delta effect sizes further support the significance and practical magnitude of the observed improvements. Ablation experiments verify that all three proposed mechanisms contribute positively, while parameter sensitivity and population-diversity analyses demonstrate that ISETO is relatively robust to parameter variation and better able to preserve exploration capability during evolution. The additional mechanisms introduce only a moderate runtime overhead relative to the original SETO. ISETO is further applied to a three-dimensional robotic-arm path-planning problem. Over 30 independent runs, it obtains a best objective value of 290.69 and a mean value of 291.91. Its best result is 5.56% lower than the best competing result, while its mean result improves upon the original SETO by approximately 6.02%. These results demonstrate that ISETO provides improved convergence accuracy, robustness, and search diversity while retaining competitive computational efficiency, and has promising potential for constrained engineering optimization.