New Medium Post: Part 2 of the FG blocking series as the #NFL season starts. This time focusing on edge FG blocking challenges and how 1 play turned the edge block on its head. Thanks again to @StatsbyLopez and the NGS team for '22 #bigdatabowl data. Link👇https://t.co/3nPj9npASQ
@mike_petriello An alternative hypothesis would be that there’s something flawed about the success rate calculation within the play itself (if the flaw hid some player variance in talent etc…)
What does the 3D axis look like (already on here)? How does that relate to the pitch’s movement? What does the axis of a comparable normal gyro look like? How would you get from A->B Are there other effects etc…
@tangotiger Claiming what it’s “mostly” about is a bit too strong imo. It’s difficult to disentangle, but faster pitches are much harder to time up. If you take it to the extreme in either direction (sound barrier pitches from far and soft-toss from close), timing would seem to be the key.
@tangotiger I’m a bit confused by what you mean. The reason why we wouldn’t say the equivalent is >200 mph is because reaction time is *not* all that matters.
@sunshinevvn@alicealeph0@srbrown70 Yup. Looking at old slacks it says 80% top-1 which is hard to believe, but very interesting if so. ofc splitting by id is important. Fortunately I don’t build stuff models (and I never will!) but this discussion comes up every few months and I haven’t seen much public research
@sunshinevvn@alicealeph0 @SamskiNYC Yes I tried this a year ago as part of a convo with @srbrown70 and a simple stuff-like classifier for FF was pretty accurate. Don’t remember exactly but maybe 50%(?) top-1 identified in a moderate sample of P. The stuff space is quite large for unique pitches to stand out.
@CarlosACollazo It’s a bit tricky though because excellent college prospects typically quickly become MLBers while elite HS prospects stick on lists for longer. So the implication of the first sentence doesn’t necessarily follow.