2023 YTD MAE
One standout is right there with the line: Stats Tailgate by @BlakeTAtkinson (model added this week).
Also posting strong figures worthy of recognition: @Room44B & CBCR2 from @DawgQuant.
6/7
@professor042 Right now Iโm making do with calculating implied and manually looking up to make sure itโs available. Iโm not sure there are many options. I may just have to figure out the key numbers and then take the closest one to market implied 50/50
Ran out of time making viz, but going to try and post/log picks. Two bets I'd still make: LA Tech inside 13 for the talent gap and SDSU TT over better than 27. Grade on SDSU QB who took over midway last year good, OC has to be better, and Ohio steam has nothing to do w/ defense
Not sure why it smited Michigan, Clemson, Pitt, Louisville. Going to see how it does and tweak it. Right now, not even close to actionable model performance. Shoutout @BudElliott3's summer school, tons of signal there, and of course @ESPN_BillC 's SP+ system.
Gzip is indeed all you need ๐ inspired by the @Sentdex video (on @ZhiyingJ's paper) I compressed @ESPN_BillC burning questions using Gzip and ran a KNeighbors regressor on it. It correctly identified Georgia, Alabama, and Michigan as top teams out of sample. THIS SHOULD NOT WORK