Abstract
<jats:p>Predicting enantioselectivity in asymmetric (organo)catalysis remains a longstanding challenge, particularly in systems where stereocontrol is governed by weak, highly flexible, and substrate-dependent noncovalent interactions. In such regimes, simplified transition-state-based DFT analysis can become unreliable due to the presence of multiple low-energy conformations, competing binding modes, and alternative reaction pathways. We introduce a previously unexplored chiral imidazolium/aryloxide betaine organocatalyst platform for asymmetric 1,3-dipolar cycloadditions of CH-acidic imines with maleimides that operates with exceptional efficiency, reaching turnover numbers of up to 3000 while maintaining high enantiocontrol. To overcome the limitations of transition-state-guided catalyst optimization and enable predictive control beyond the limits of mechanistic intuition, we combined combinatorial catalystsubstrate screening with descriptor-based statistical modeling. Analysis of the resulting dataset reveals that no single catalyst is universally optimal across substrate space; instead, enantioselectivity is governed by a catalyst-substrate matching effect. The resulting models identify the molecular features controlling selectivity and enable prospective prediction of enantioselectivity for previously unseen catalysts, structurally distinct substrate classes, and new imine motifs. These findings establish highly productive metal-free azomethine ylide cycloadditions and demonstrate how interaction-aware data-driven models can provide predictive control in complex asymmetric catalytic systems.</jats:p>