Seth Walder @sethwalder.bsky.social · 20hPenalties are an overlooked reason why it makes more sense to go for it on 4th down than people realize. Defensive holding = first down conversion. But if offense commits a penalty they still often get to kick the FG. 0403
Seth Walder @sethwalder.bsky.social · 20h...if teams move closer and closer to optimal then that means they might have more situations they would consider or actually go for it that model might think kick is better 070
Seth Walder @sethwalder.bsky.social · 20hI don't know if we've seen more kicks on screen. But! Here's a way that could happen: the reason it always seemed like it only said go is because any time a team even thought about going for it, that meant it was a go situation because teams were overly conservative. But... 150
Seth Walder @sethwalder.bsky.social · 20hI wouldn't say teams have been failing more on 4th down once we control for distance to gain? 150
Seth Walder @sethwalder.bsky.social · 05/10/2026technically Geno didn't even have enough plays to qualify! but he had a better QBR anyway 100
Seth Walder @sethwalder.bsky.social · 05/10/2026The two worst QBRs of the week thus far were quarterbacks on teams that won: Daniel Jones and Sam Darnold. 4251
Seth Walder @sethwalder.bsky.social · 05/10/2026The five best teams in football, according to ESPN's Football Power Index. Hello, Jaguars! 7284
Seth Walder @sethwalder.bsky.social · 05/10/2026Thought this one was interesting too! Argument for: if you convert opponent scores, a missed PAT results in tie game. Fail, still down 3 on ensuing score. Argument against: make opponent go for 2 to go up 3. Model leaned slightly toward PAT, but it's close. 53% breakeven percentage. 2100
Seth Walder @sethwalder.bsky.social · 05/10/2026there are ones closer that don't get called that should, too. if an NFL QB throws a ball 15 yards away from any receiver we shouldn't give them the benefit of the doubt 3362
Seth Walder @sethwalder.bsky.social · 05/10/2026that's an absurd no call on intentional grounding 57010
Seth Walder @sethwalder.bsky.social · 05/10/2026don't like that spike with 47 seconds left by LV. you can still lose on downs before clock runs out here 1221
Seth Walder @sethwalder.bsky.social · 04/10/2026gotta respect Hockenson for keeping Dolphins +11.5 intact 0210
Seth Walder @sethwalder.bsky.social · 04/10/2026Super close! But our model slightly preferred keeping the three points on the board once we knew FG was good. 0200
Seth Walder @sethwalder.bsky.social · 04/10/2026Have been keeping an eye on them given this: each had been strong in my CB metric. bsky.app/profile/seth... 040
Seth Walder @sethwalder.bsky.social · 04/10/2026Strong coverage numbers for both Bengals corners Dax Hill and DJ Turner II in today's loss to Jaguars, per Next Gen Stats. Hill: 2 receptions allowed for 11 yards (1 TD) (and 1 forced fumble) on Turner: 1 reception for 10 yards. One penalty. 27 coverage snaps for each. 280
Seth Walder @sethwalder.bsky.social · 04/10/2026Abdul Carter recorded 14 pass rush wins in today's win against the Cardinals. That's the most PRWs by a player in a game this season. (ESPN Analytics / NFL Next Gen Stats) 0242
Seth Walder @sethwalder.bsky.social · 04/10/2026Kamari Lassiter allowed eight receptions for 136 yards on 11 targets over 50 coverage snaps, per NFL Next Gen Stats. Plus, three penalties. That's the most receiving yards allowed by a cornerback in a game this season. 090
Seth Walder @sethwalder.bsky.social · 04/10/2026i know what you're thinking: yes, darnold-meter.com is live 2244
Seth Walder @sethwalder.bsky.social · 04/10/2026Good no call on Stafford "late" hit, that would have been a crime if it had been thrown. Defender touched him before he went out of bounds and defender pulled up, too. 1231
Seth Walder @sethwalder.bsky.social · 04/10/2026well then you REALLY need to go for it because kicking lets you lose with two touchdowns allowed 160
Seth Walder @sethwalder.bsky.social · 04/10/2026not important but not sure i've ever seen the LOS blue line be more off than it was on that Bucs 4th down just now 0150
Seth Walder @sethwalder.bsky.social · 04/10/2026Eagles needed just a 13% chance of converting to justify going for it, so this was the only decision. Value of being up 13 vs. 10 is very small and would have given Rams better field position than a 4th down failure, too. Correct call by Philly. 8243
Seth Walder @sethwalder.bsky.social · 04/10/2026That's a ball don't lie PAT miss by Jason Sanders, because Jets should have been going for 2 down 11 (down 8 scenario + FG / down 4 scenario + TD) 0200
Seth Walder @sethwalder.bsky.social · 04/10/2026Our model easily agreed with all three Packers 4th down go decisions on that touchdown drive. 0190
Seth Walder @sethwalder.bsky.social · 04/10/2026We've been overdue for a Solak show and a daily show. Love that this is happening. 0220
Seth Walder @sethwalder.bsky.social · 04/10/2026I also happen to love what I saw from Daniels in the preseason. So those two factors together make me awfully bullish on him. 0130
Seth Walder @sethwalder.bsky.social · 04/10/2026Because of selection bias, I think knowing nothing else our expectations for Jalon Daniels should be higher than a rookie who was, say, a 5th-round pick. A UDFA gets no benefit of the doubt, must clearly outperform others to take their job. So the fact that Daniels earned backup job says a lot. 2261
Seth Walder @sethwalder.bsky.social · 04/10/2026Week 4 best bets, alongside @benjaminsolak.bsky.social and Matt Bowen! www.espn.com/espn/betting...espn.comNFL Week 4 betting guide: Analysis and top plays for Sunday's gamesOur experts dive into the odds and player props for Week 4 and offer their betting tips and favorite plays. 021
Seth Walder @sethwalder.bsky.social · 02/10/2026no, I have it for the past few years! Here's 2025! bsky.app/profile/seth... 110
Seth Walder @sethwalder.bsky.social · 02/10/2026I point this out for transparency, and to show how tricky something like this is! A given play can have factors unseen or misinterpreted by the model. But in aggregate over a season, I think it’s valuable. And the fact that Phase 1 is so stable for players from year-to-year reinforces that as well. 050
Seth Walder @sethwalder.bsky.social · 02/10/2026If the model thought this were zone, the ghost defender would have played differently than in the GIF above 120
Seth Walder @sethwalder.bsky.social · 02/10/2026Also, of course, there can be data issues. Here’s play the model really didn’t like from Witherspoon. But it also has been told that this is Cover-1. But it doesn’t look like man to me, and I think that explains Witherspoon’s positioning. 100
Seth Walder @sethwalder.bsky.social · 02/10/2026Those are obvious. But it’s not always so clear. Devon Witherspoon at the bottom of the rankings really stands out. Here’s a play the model did not like from Witherspoon. It dings him for the extra cushion he gives Wilson, boosting the xComp. 100
Seth Walder @sethwalder.bsky.social · 02/10/2026The Puka play was incomplete! So "regular" advanced CB metrics like yards per coverage snap, EPA allowed, CPOE, etc. would all pick up that as a positive play for Lenoir. But the player-tracking picks up that it wasn't 110
Seth Walder @sethwalder.bsky.social · 02/10/2026Here are two plays the Phase 1 model significantly dinged cornerbacks for. Deommodore Lenoir beat by Puka Nacua and Charvarius Ward against Chris Moore. Again, these both happen to be targeted but they don’t have to be. 120
Seth Walder @sethwalder.bsky.social · 02/10/2026Here’s another Hill play the model loved. You can see how the gap between Hill and the ghost isn’t large. But the difference of a yard or even the angle a corner is facing or moving can have significant impact on completion probability. 110
Seth Walder @sethwalder.bsky.social · 02/10/2026Dax Hill leads all corners in Phase 1. Here’s an example play the metric loved. This play he happens to be targeted, but Phase 1 looks at positioning regardless of target or not. On the dots, you can see Hill’s positioning (blue) is better than the ghost defender (red). 110
Seth Walder @sethwalder.bsky.social · 02/10/2026I do want to stress how fickle a complex player-tracking metric like this is, and that three weeks of play is a small sample. Still, certainly worth looking at I believe. 330
