Game-theoretic Statistics and Sequential Anytime-Valid Inference: Game-theoretic Statistics and Sequential Anytime-Valid Inference (SAVI): A Martingale Theory of Evidence

Aaditya Ramdas

International Conference on Machine Learning 2025 · Tutorial

Aaditya Ramdas’s tutorial at ICML 2025 introduced attendees to the rapidly evolving field of **game-theoretic statistics** and **Sequential Anytime-Valid Inference (SAVI)**, presenting it as a foundational shift in how we approach statistical inference. Ramdas, a distinguished statistician and recipient of numerous prestigious awards, posits that this framework, rooted in a **martingale theory of evidence**, offers a robust, flexible, and universal alternative to classical statistical methods, particularly addressing the pervasive issues associated with **p-values**. The talk highlighted how concepts from gambling and betting provide a powerful lens through which to unify and re-envision hypothesis testing and estimation, enabling valid and efficient inference even in complex, adaptive, and sequential data analysis scenarios.

AI review

Ramdas delivers a rigorous and well-organized tutorial on game-theoretic statistics and Sequential Anytime-Valid Inference, grounding the framework in martingale theory and making a credible case that e-values and e-processes are the right objects for sequential hypothesis testing. The theoretical backbone is sound — Ville's inequality, the universality theorem for e-processes, Kelly/log-optimality, the reverse information projection, and the time-uniform CLT are all presented with appropriate precision. The talk does not pretend to be a paper with original proofs; it is a tutorial, and…