Hypothesis Testing with E-values (Ramdas & Wang 2025) - Weekend Book Review

English Podcast starts at 00:00:00Bengali Podcast Starts at 00:40:14Hindi Podcast Starts at 01:04:30ReferenceAaditya Ramdas and Ruodu Wang (2025), "Hypothesis Testing with E-values", Foundations and Trends® in Statistics: Vol. 1: No. 1-2, pp 1-390. http://dx.doi.org/10.1561/3600000002Open Access book-link https://sas.uwaterloo.ca/~wang/files/e-book-final.pdf‌Youtube channel link https://www.youtube.com/@weekendresearcherConnect on linkedinhttps://www.linkedin.com/in/mayukhpsm/Welcome to Revise and Resubmit, Weekend Book Review. 🎙️📚Today I am opening a fresh chapter on evidence itself with Hypothesis Testing with E-values by Aaditya Ramdas and Ruodu Wang, published on 10 Sep 2025 by now publishers inc. ✨ Ramdas is an Associate Professor at Carnegie Mellon in Statistics and Data Science and the Machine Learning Department, a CMU PhD who sharpened his craft at Berkeley with mentors Michael Jordan and Martin Wainwright, and an IIT Bombay computer science grad with an All India Rank of 47. Wang anchors the University of Waterloo’s Department of Statistics and Actuarial Science, roaming the terrain of quantitative risk management across actuarial science, financial engineering, operations research, probability, statistics, and economic theory. 🧠🌍This monograph is humble in tone, united in purpose, and rich in structure. Three parts. Sixteen chapters. First, Fundamental Concepts with four clear chapters that lay the groundwork. Then, Core Ideas with five chapters on universal inference, log-optimality, e-processes, operations on e-values, and e-values in multiple testing. Finally, Advanced Topics with seven chapters that push into the frontier and pull together results from a flurry of modern papers, plus many insights not published elsewhere. 📘🔍If p-values measure surprise, e-values measure support you can multiply, carry forward, and spend wisely. They generalize likelihood ratios and behave cleanly under the null and alternative. They shine in sequential inference and post-hoc testing. They help you merge results and control false discovery rates. You meet the numeraire, you meet the reverse information projection, and you chase log-optimality like a north star. I felt the cadence as I read: define, prove, build, test. It reads like a course you can teach tomorrow and a toolkit you can use today. 🚀🧪🎯Thank you to the authors, Aaditya Ramdas and Ruodu Wang, and to now publishers inc. 🙏 If you enjoy these brainy page-turners, subscribe to the podcast channel on Spotify and to our YouTube channel Weekend Researcher. We are also available on Amazon Prime Music and Apple Podcast. 🔔🎧📲 So, if your evidence could compound with every look, what would you test first? 🤔