Vertox

Japan just had its own version of the Hawaii missile scare. At around 12:40, an emergency alert was broadcasted warning of an imminent magnitude 9.9 earthquake off the coast of Hokkaido. A magnitude 9.9 earthquake, if it happened, would be the strongest earthquake ever recorded, releasing approximately 22 times more energy than the 2011 Tohoku Earthquake. Fortunately just minutes later the earthquake alert was withdrawn and confirmed to be a false alarm. But for those few minutes, people across the country held their breath bracing for a potentially catastrophic earthquake that ultimately never came.

Background: I am co-teaching (with Kent Daniel) a grad course at Columbia titled "Quantitative Investing". I can share my slides, starting with the ones on market structure (ch.1 in my 2nd book). Below are a few representative slides. I will share the lecture on linear return models this weekend. At some point also one lecture on very high-dimensional linear models, which are interesting. For now link to first deck is in reply. Enjoy.

About a year ago, Talagrand’s convolution conjecture was announced with a 40+ page proof, and later simplified by Shaposhnikov, with the help of AI, to 7 pages: arxiv.org/abs/2609.11290 Since then AI has improved so much that, after playing with it a little, it rewrote the full proof in 1.5 pages and in my style (I told AI what tools I’m familiar with and asked it to stay within those limits and kept just asking "simplify proof"). As long as a valid proof exists, AI seems increasingly able to simplify it (or even find a simpler alternative proof) and rewrite it much shorter and more cleanly in your style. So I think I agree with @ChrSzegedy that human “desloping” may be a temporary issue and eventually won’t really be needed: teapawt.net/ChrSzegedy/status/2103… Looks like we are converging back to Perelman’s statement: if the proof is correct, no other recognition is needed.