Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Seth Walder Profile picture
Seth Walder
@SethWalder
ESPN Sports Analytics Writer, primarily covering the NFL. DMs open.
Sep 17, 2020 • 6 tweets • 3 min read
Week 1 Completed Air Yards Over Expectation (CAYOE) passing leaders.

CAYOE is basically @NextGenStats CPOE*Air Yards, with passes behind the LOS excluded.

E.g. Imagine a 10-yard pass with a 60% chance of completion. If complete, that’s a +4 CAYOE play. If incomplete, it’s a -6. Image @NextGenStats Why do it? Because while CPOE accounts for the difficulty of the depth of target, it doesn’t account for the value

E.g. if Wilson and Brees are both +10s in CPOE, completing 10 % points over expectation is a lot more valuable on Wilson passes (deep) than on Brees passes (short)
Sep 16, 2020 • 6 tweets • 2 min read
How often did each team use motion in Week 1? First column is motion at the snap, second column is all pre-snap motion.

Shoutout to our video trackers who make this stat possible. ImageImage So we've shown before that motion at the snap is an edge for the offense.

And crazily, the top 12 teams here all won in Week 1.

But guess what: EPA/P league-wide was *not* higher on motion at snap plays in Week 1.
Sep 2, 2020 • 5 tweets • 2 min read
I like to look at which players I was highest/lowest on relative to the group in NFL Rank. These are the 20 non-QBs I was highest on relative to everyone else. Image I know Linsley is #1 but I really consider myself an Elgton Jenkins stan
Aug 23, 2020 • 7 tweets • 5 min read
Re: Earl Thomas, here's a quick plot of 2019 safeties' target rate and CPOE allowed, via @NextGenStats and inspired by recent @reinhurdler plot. Image @NextGenStats @reinhurdler I’ll say: a low target rate for a safety might be heavily influenced by role (say, centerfield) as opposed to a corner with a low target rate (meaning QBs avoid).
Aug 13, 2020 • 10 tweets • 3 min read
Some more Tyler Lockett love. So I’ve written before about how he dominates CAYOE (completed air yards over expectation). But the question always lingers: is that because Lockett is good? Or because Russell Wilson is throwing to him?

espn.com/nfl/insider/st… And to make it tougher: Only Wilson has thrown to Lockett in three year span we have CAYOE.

But Wilson has also thrown to other receivers (who have played with other quarterbacks).
Aug 11, 2020 • 6 tweets • 2 min read
Building on the team coverage clustering, here’s a depiction of the clusters (made a slight fix, four clusters now) in a principal component chart to visualize.

This is 2019, regular season. Image The idea here is that since we can’t see all 7 variables at once, we create these principal components, which are made up of combinations of those original variables. Between the two variables on this plot (PC1/PC2) over 50% of the variance in coverage profiles is captured.
Aug 9, 2020 • 5 tweets • 2 min read
After a couple of recent hires, here’s the updated NFL analytics staffers list.

As always, this is to the best of my knowledge based on not only what teams list but also conversations with staffers themselves.

It is also uses a fairly narrow definition of analytics. Image Changes:

-Jacqueline Davidson (TB) was hired
-Matt Ploenzke (SF) was hired
-Dawson Friedland (GB) was hired
Aug 4, 2020 • 5 tweets • 2 min read
Another half-baked project, please take only as such.

K-means clustering of the rate teams used defensive coverages in 2019. ImageImageImageImage How did I decide k? Umm, eyeball. I know there are methods to choose k. But I have no idea what I’m doing, and this is half-baked, like I said. So I chose 5.
Jul 30, 2020 • 13 tweets • 3 min read
Want to share more half-baked/unfinished projects. So take this as that.

Going to share this chart here and then I'll thread how I got there. Then we can find out if I did something wrong, too!

Also: Tyler Lockett remains a legend. Image So we know that there are non-linear relationships between air yards and receiver separation and air yards to sticks (AY relative to first down marker) and receiver separation.



Jul 24, 2020 • 6 tweets • 1 min read
Updated NFL analytics staffers list. Image Changes:

-Had been missing Taimoor Chatoor (LV)
-Sarah Bailey (LAR) was promoted
-Jeff Scott (WAS) was promoted into a scouting role, removed from list
Jul 16, 2020 • 4 tweets • 2 min read
We've got a ways to go evaluating corners quantitatively and they are also notoriously variant from year to year. That being said, it's fascinating that Jalen Ramsey is No. 2 here when in 2019:

-His CPOE allowed was +7.9%
-18% target rate (avg: 17%)
-46% coverage success (50%) Those averages are among players with at least 300 coverage snaps that played at outside corner at least 80% of the time.

And based on when those players are the nearest defender to receiver at time of ball arrival, which may not always mean primary coverage responsibility.
Jul 16, 2020 • 7 tweets • 3 min read
hey nerds you know what this was? a red/blue confusion!!!! Image so let's try this again.

tanner morgan sure does love that middle of the field!

blue = high Image
Jul 12, 2020 • 4 tweets • 1 min read
Justin Tucker If we multiply @NextGenStats' FG +/- by attempts, we get field goals over expectation. Over the past four seasons (as far back as stat goes) Justin Tucker doubles-up the next-best player.

Justin Tucker: 24
Matt Prater: 12
Wil Lutz: 12
Matt Bryant: 12
Josh Lambo: 11
Jul 9, 2020 • 4 tweets • 1 min read
We now have motion data back to 2017 (previously 2019). Here teams over the last three seasons are grouped by the rate at which they had a man in motion *at the snap*. Image espn.com/nfl/story/_/id…
Jul 8, 2020 • 6 tweets • 2 min read
How often do teams target certain route types relative to league average?

Rather than specific routes for this I'm using route groupings like vertical (includes post, go, corner, seam, deep fade) and outside-short (includes swing, flat, short fade, speed out). Image This is the difference between the % of each team’s targets to a particular route group and % of the league’s targets to that same route group.

Red = above average
Light gray = below average
Jun 23, 2020 • 5 tweets • 2 min read
Copying an idea from @HockeyGraphs and @hayyyshayyy: I’m compiling a list of NFL analytics staffers that I'm making public and intend to maintain.

I've checked the list with a bunch of staffers -- but not someone with every team -- so please consider this a draft/first pass. Image @HockeyGraphs @hayyyshayyy There’s a lot of gray area in building this list. Had to draw some lines somewhere, so my four general parameters were:

1. The list focuses on staffers that perform data analysis and build analytics tools

2. Sports science is excluded
Apr 23, 2020 • 8 tweets • 1 min read
If Tua somehow slipped to them, which teams in range 10-15 *would* take him and which teams *should* take him? OK let's try to answer this question for each team with a poll.

If Tua fell to 10, would/should the Browns select him?
Apr 15, 2020 • 4 tweets • 2 min read
Our new NFL Draft Predictor can tell us the chance a player is still on the board at a certain spot. The model is based on industry mock drafts, team needs and Scouts Inc. grades. I wrote about some storylines with it today (cont.) espn.com/nfl/insider/st… But if folks have questions about other players and their likelihood of being available at a certain spot, fire away and I'll answer some for a bit.
Apr 10, 2020 • 9 tweets • 2 min read
I wonder if Detroit could pull off a dream scenario where Miami trades up to 3 for Tua and then Jacksonville trades up to 5 for Herbert and the Lions end up with 9, 18 and 20. ^^obviously was not thinking big enough before Image
Mar 19, 2020 • 5 tweets • 3 min read
Not my area of expertise so legitimately asking: why would a Gurley trade happen instead of cut?

Rams would surely have to give up comp to cover a negative asset, but it’s an asymmetric contract situation, I think. Rams could cut and make 2020 salary and 2021 roster bonus go away, but trading team would have to pay that salary and let 2021 roster bonus vest.

Rams save ~7.5M by trading instead of cutting, but new team would have to take on $10.5M of money the Rams wouldn't spend if LAR cut
Mar 6, 2020 • 12 tweets • 3 min read
Double team rate as an edge rusher (x) by pass rush win rate as an edge rusher (y) in 2019 for players who were in the last season of their contract.

Lower qualifying threshold than usual to get a few more players in. Image Double team rate as a defensive tackle (x) by pass rush win rate as a defensive tackle (y) in 2019 for players who were in the last season of their contract.

Lower qualifying threshold than usual to get a few more players in. Image