Pre-registration (frozen 7 Nov 2026): replication of the AnoOnly gain for DeepSAD and PReNet (arXiv 2305.18798), as published and against a DeepSAD with tuned eta

Frozen protocol, decision code, job design, harness, planted-effect table and kernel for a private pre-registered replication, deposited before any confirmatory run. Clauses: the mean AUCROC gain of AnoOnly over DeepSAD across ten datasets has a 95% bootstrap lower end above 9.8 points (half the paper's 19.6) with at least 8 of 10 datasets improving, in an as-published arm and in an arm where DeepSAD's own loss weight eta is tuned by cross-validation; and at least 7 of 10 datasets improve for PReNet (as-published arm). Outcome wording: Replicated / Did not replicate / Inconclusive under protocol X.