Evaluative Crowding in Financial Markets: Shared Objective Structures and Strategy Convergence

This article examines a recurring empirical pattern in financial markets: strategies constructed on heterogeneous data and models nevertheless converge in positioning and unwind in a highly synchronized manner under stress. Episodes such as the August 2007 quant crisis, the March 2020 cross-asset dislocation, and the 2022 UK liability-driven investment (LDI) episode exhibit this structure. Existing explanations based on liquidity, leverage, and funding constraints account for the amplification of shocks but do not fully explain why positions are aligned ex ante. The article advances a complementary mechanism—evaluative crowding—through which convergence arises from shared transformation rather than shared inputs. Agents optimize under evaluative structures ℓ∈L, comprising risk operators, admissibility constraints, and objective functions. When these structures are partially aligned across institutions—through common risk models, constraints, and implementation infrastructure—heterogeneous signals are mapped through a common operator Φℓ, inducing clustering in admissible strategies without requiring shared information or coordination. A tractable game-theoretic model characterises the structural implications of shared evaluative mappings. When evaluative operators exhibit low effective dimensionality—a standard feature of factor-based risk models—optimal portfolios concentrate in a restricted subspace. Within this framework, alignment arises as a property of shared transformations rather than shared positions, and a formal distinction emerges between first-order failures (performance deterioration within a fixed evaluative structure) and second-order failures (misalignment of the evaluative structure itself). When evaluative structures are fixed over relevant horizons, perturbations to the mapping from signals to strategies generate correlated adjustments across agents, producing synchronized responses under stress. The article derives empirically testable implications, including convergence conditional on evaluative structure, discrete shifts in evaluative frameworks around crisis events, break-type adjustment dynamics, and regime-dependent persistence in positioning. These implications distinguish evaluative crowding from factor-based and ownership-based mechanisms under appropriate controls. A stylized simulation, matched to an analytic benchmark, demonstrates that, in the stylised setting, dispersion compression and directional alignment are fully determined by the spectral properties of the evaluative operator under isotropic inputs. The results imply that systemic fragility depends not only on balance sheet constraints and market liquidity, but also on the structure of evaluation governing decision-making.