Meta-heuristic algorithms distinguish themselves from conventional optimization methods owing to their intrinsic adaptability and straightforward implementation. Among them, the sine cosine algorithm (SCA) is lauded for its ability to transition seamlessly between exploration and exploitation phases throughout the optimization process. However, there exists potential for enhancing the balance that SCA maintains between exploration and exploitation. To augment the proficiency in global optimization of SCA, an innovative strategy—nSCA—that integrates the roulette wheel selection (RWS) with opposition-based learning was formulated. The robustness of nSCA was rigorously evaluated against leading-edge methods such as the genetic algorithm (GA), particle swarm optimization, moth-flame optimization, ant lion optimization, and multi-verse optimizer, as well as the foundational SCA. This evaluation included benchmarks set by both CEC 2019 and CEC 2021 test functions. Additionally, the performance of nSCA was confirmed through numerous practical optimization problems, emphasizing its effectiveness in applied settings. In all evaluations, nSCA consistently showcased superior performance compared to its evolutionary algorithm counterparts, delivering top-tier solutions for both benchmark functions and real-world optimization challenges. Given this compelling evidence, one can posit that nSCA serves as a strong candidate for addressing intricate optimization challenges found in real-world contexts, regardless of whether they are of a discrete or continuous nature.