Constructs synthetic C51+IQN distributions matching observed val-collapse failure mode. Verifies argmax(E[Q]) deterministically picks Flat/Hold while Thompson sampling produces ≥30% Long+Short over 10000 seeds.
Constructs synthetic C51+IQN distributions matching observed val-collapse failure mode. Verifies argmax(E[Q]) deterministically picks Flat/Hold while Thompson sampling produces ≥30% Long+Short over 10000 seeds.