I'm excited for the advanced stats community to start regularly plotting standard deviations and error-bars. Yes, it will make it clearer how many charts people make are completely useless, but that's what the community needs right now. Too many averages without variance: not very meaningful

This is a cool chart. Much more useful than TTT. One thing I would be interested in seeing is dot size (or dot color using a colorbar) as total EPA or EPA/db on qualifying plays. It would be a way to account for positive production in addition to sacks taken.

Need to dig into nflfastR more! I was thinking about the AWS NGS one. The E in CPOE is a black box in what is already a noisy metric (because of the binary nature of the counting stat). And most analyses don't subdivide CPOE by target depth, so it's super hard to draw conclusions from the bulk stat.

Fair, but I come from a world (science) where this kind of data can be audited and that's how we validate the methodology. PFF (and other methods like CPOE) have asked us all to accept their methods without letting us look at their data. They provoke this discourse with their lack of transparency

Also, we don't know much about PFF graders but we do know people generally are susceptible to narratives. By midyear we know which offenses are good and bad (which is all EPA/db is), so slight grader bias could bend PFF grades enough for a 0.21-0.24 correlation to emerge

0.23 rsq could mean the top 2-3 performers and bottom 2-3 performers are consistent across datasets. The middle 80% (the slice we actually care about, since these players are harder to accurately assess with conventional metrics or the eye test) could still be an uncorrelated jumble.

Question: wouldn't it make more sense to only count under-center handoffs (Since those are the where run/PA duplicity is supposed to yield results)

Also curious about rushing and PA success rates for each team.

Trump's tarrifs won Carney an election. You're welcome Canada, I guess?

Sorry for the delay this week! Back to regularly scheduled programming tomorrow! Really excited about this one, as the owner of a small but might Ackie monitor named Mushu.

Here's some reading!
www.nature.com/articles/s41...

www.biorxiv.org/content/10.1...

pubs.acs.org/doi/abs/10.1...

Ah shit it's a raccoon. You guessed it. Whole week is a bust.

You can try to guess along at what the creature is throughout the week, or just wait until the reveal! I want to have fun with this, so I’ll switch up the art styles pretty consistently. I plan to do 12 weeks for my first run of this series (“Season 1”), and will start tomorrow, on January 28th!

Very exciting. Have you considered posting scoops, breaking news, etc directly there? With an 11M person userbase, the NFL subreddit feels like a real untapped information avenue for reporters. Sidenote, it's cool to see big people lean into Bluesky.