TOPSIS-RAD is a proposed improvement to the traditional TOPSIS decision-making algorithm that incorporates decision-maker-defined reference points to prevent ranking misalignments and sensitivity to outliers. The method introduces Vetoed Performance Levels (VPL) to exclude non-viable alternatives and Desired Performance Levels (DPL) to anchor rankings in explicit aspirations rather than dataset extremes.
TOPSIS-RAD addresses fundamental limitations in multi-criteria decision analysis that have practical implications across sectors relying on algorithmic ranking systems. Traditional TOPSIS derives its reference points from observed data, creating vulnerability to outlier distortion and rank reversals when datasets shift—a critical weakness in domains requiring stable, reproducible rankings.
The proposed framework introduces two mechanisms: VPL acts as a pre-normalization filter to remove non-viable alternatives before they influence the ranking frontier, while DPL establishes explicit performance caps aligned with decision-maker aspirations rather than dataset maximums. This dual-constraint approach grounds the algorithm in human preferences rather than statistical extremes, enhancing alignment between computational outputs and stakeholder requirements.
While primarily academic in nature, this methodology holds relevance for blockchain and AI applications requiring transparent, reproducible decision-making. Portfolio ranking systems, token selection algorithms, and governance voting mechanisms could benefit from incorporating explicit preference boundaries. The framework's emphasis on decision-maker-defined parameters rather than data-derived extremes addresses growing concerns about algorithmic bias and the need for explainable AI systems.
Future implementations might integrate TOPSIS-RAD into decentralized governance protocols or index selection mechanisms where stability and explicit goal-alignment matter. The research represents incremental academic advancement rather than breakthrough innovation, but contributes to building more robust decision-support infrastructure for complex multi-criteria evaluations in emerging technology sectors.
- →TOPSIS-RAD prevents outlier performances from distorting ranking boundaries through decision-maker-defined reference levels.
- →Vetoed Performance Levels exclude non-viable alternatives before normalization, stabilizing the ranking frontier.
- →Desired Performance Levels anchor the ideal solution in explicit aspirations rather than dataset extremes.
- →The method preserves TOPSIS's distance-based structure while grounding rankings in human-specified boundaries.
- →Framework has potential applications in blockchain governance, portfolio ranking, and algorithmic decision systems.