@enosarris @MarkPriorRBW Yes, that's his "Slurve". Seems to be the replacement classification for what has been his curveball the past few years. He and Berrios might be the only "slurves" to be retroactively classified on savant. For now in the model I have to treat these as curves.
Stuff+/Location+/Pitching+ now up on FanGraphs! Here are some words about how it works.
(If you scroll all the way to the end, you have permission to yell at me and berate me. Or ask questions, whatever you’d like!)
https://t.co/W6VavFu5rh
@enosarris @LeoMartinic the model doesn't seem to care about the velo if the vertical movement is like this, similar with his horizontal movement profile. not that it's absolute but some reflection of 2020/21 data the model is trained on.
hudson in gray, league avg is the dotted
@enosarris @jimmy__the__kid ashcraft needs specific guardrails since both the cutter and sinker come out at same speeds.
overall, unique characteristics and will therefore break the model but unfortunately can really only be corrected as part of a larger model re-train
@Jacob_817@enosarris@DerekVanRiper@mason_mcrae the current model takes in more spin data than is here but by movement and velo, similar pitchers have around a 97 stuff+. also just want to throw in the small sample caveat
Some words about the nature of the bendy breaking ball. If you’re at all intrigued by the piece, I highly encourage you check out the comments under the post.
It can be difficult to think about the value of horizontal movement, but HAA is a great synthesis of release, movement, and location that gets us a bit closer.
An intro and some visuals here
Kyle Tucker has a team best 144 wRC+ and when everyone’s healthy he’ll likely hit 7th. I take a look at what Dusty Baker is up to and why it just might work.
I love me a good mistake. Some words about the success of backup sliders without diving into the question of sequencing because, well, it’s still a blog post.