Anytime-Valid Inference Overview
Inference that remains valid at arbitrary stopping times (preserving type-I error even under optional stopping): e-process / confidence sequence. Roadmap Track 3. Applies directly to real-time monitoring of identification-validity drift.
Overview
Game-theoretic statistics that resolves the “peeking” problem of fixed-sample hypothesis testing. The mathematical foundation for real-time monitoring of identification-validity drift.
Target atomic notes
- e-process ✓ (includes e-value)
- Confidence Sequence ✓ — time-uniform (Howard et al. 2021)
- Safe Anytime-Valid Inference ✓ — SAVI (Ramdas et al. 2023)
- Anytime-Valid OPE ✓ — Waudby-Smith et al. 2024
References
- Study Roadmap §Track 3 (Howard 2021 → Ramdas 2023 → Waudby-Smith 2024 → Ramdas-Wang 2025)
- MOC-DecisionMaking