Jimmystats: Targeting Data 2020
This post is mostly a data dump. Because I like advanced stats in HTTV I've been playing with play-by-play data from CollegeFootballData.com and managed to turn over 90,000 rows of this:
…into target data for all receivers in 2020.
…including (broken out so this is optional) pass interference and holding calls drawn.
Expected Points Added
I’ve also calculated Expected Points Added for all the plays, which we get from knowing every down/distance/field position has a value, based on how many points an average offense is likely to score that drive. So for example:
- It’s 3rd and 14 on the Indiana 42 = 1.3 expected points*
- Ronnie Bell makes a 17-yard catch.
- Now it’s 1st and 10 on the Indiana 25 = 4.3 expected points
- Ronnie Bell’s catch equals 4.3 minus 1.3, or 3 EP.
I treated touchdowns as 6.971 points, and adjusted all of these for opponent, based on Bill Connelly's SP+. So for example this play which started at 1st and 10 from the Michigan 30 (1.3 EP) was adjusted down from 5.66 to 5.38 because Minnesota’s defense is terrible.
MICHIGAN 70-YARD HOUSE CALL
— PFF College (@PFF_College) October 25, 2020
pic.twitter.com/QI6SBM0jrW
I also threw out receptions that ended with fumbles and whatnot (so, e.g., Jake Ferguson wouldn’t take a –7.12 penalty for this). I played around with an adjustment based on how many possessions were left but ultimately decided –6.82 EP for Milton’s interception on 1st and 10 from the Indiana 47 down 31-21 at 12:28 in the 4th quarter was sufficiently punitive.
*[Actually 1.27851427694043. You can calculate any game state here, or DM me on Twitter @Misopogon if you would like to use my converter tool.]
[After THE JUMP: Receiver values]
What I like about using EP for receivers is it’s situational, rewarding clutch plays and big plays, but doesn’t damage you too much if your QB takes a shot downfield on 2nd and 1. There’s a fairly significant penalty (about –1.2) for incompletions on most 1st and 10s, but I also found the receivers we generally think of as great had very few negative plays on 1st down.
We can total up the value provided on all targets to a certain receiver last year, or divide it by targets to see how much value a team got from an average pass in that guy’s direction. The former is a better gauge of overall value, e.g. DeVonta Smith scored a 115, while the next-highest guy was at 82. Expected Points Added is cumulative so it’s less valuable when comparing receivers who get vastly different numbers of targets, e.g. when one conference plays half a schedule.
The latter shows you who’s more clutch, and favors go-to guys on 3rd down. Dividing by targets however just pulls up all the tight ends and slot receivers who were good at moving the chains.
Drew Dileo would have been boss at EP/T but I don’t find much here that you can’t tell from passing down target rates.
Michigan and Big Ten Targeting Data
It would have been too much to match up rosters for the entire NCAA, but I know the Big Ten well enough to fix errors and nicknames and such when matching the data to other resources like the NCAA’s own stats or the simple data on cfbstats.com. Here are the numbers for Michigan’s targets last year (gray text = gone):
| Target | Rec | Tar | Yds | YPT | TD | EPA | EPA/G |
|---|---|---|---|---|---|---|---|
| Ronnie Bell | 26 | 43 | 410 | 9.5 | 1 | 23.3 | 3.88 |
| Nick Eubanks | 10 | 12 | 117 | 9.8 | 1 | 11.3 | 2.26 |
| Cornelius Johnson | 16 | 28 | 235 | 8.4 | 3 | 10.4 | 1.73 |
| Roman Wilson | 9 | 14 | 122 | 8.7 | 1 | 7.8 | 1.30 |
| Blake Corum | 5 | 8 | 58 | 7.3 | 0 | 6.8 | 1.14 |
| Giles Jackson | 15 | 23 | 182 | 7.9 | 0 | 4.1 | 0.81 |
| Chris Evans | 9 | 10 | 82 | 8.2 | 0 | 3.3 | 0.54 |
| Ben Mason | 2 | 3 | 17 | 5.7 | 1 | 2.8 | 0.46 |
| AJ Henning | 6 | 8 | 59 | 7.4 | 0 | 1.8 | 0.30 |
| Zach Charbonnet | 6 | 7 | 41 | 5.9 | 0 | 0.8 | 0.15 |
| Mike Sainristil | 7 | 17 | 59 | 3.5 | 2 | 0.4 | 0.07 |
| Hassan Haskins | 0 | 3 | 0 | 0.0 | 0 | -1.6 | -0.27 |
| Erick All | 12 | 25 | 82 | 3.3 | 0 | -4.5 | -0.75 |
Breaking up expected points added by game seemed to solve for most of the issues I was having with differing usages and the varying schedules of last year. Here’s a Viz of the whole Big Ten, showing EPA/Game against the percent of team targets. If this embed doesn’t work I’m going to remove it and make you take the link. It’s worth mouse-ing around—hover over a dot to see the individual, or click any one or the selector at right to see that team:
Unfortunately only one of those five dots in the upper-right quadrant went pro, and Michigan wasn’t playing them this year anyways.
Things We Can Glean
Ronnie Bell was the only Michigan receiver to make it into the productive quadrant, but he was indeed productive, clumped with Rondale Moore, Chris Autman-Bell, Ramaud Chiaokhiao-Bowman, Bo Melton, and Ty Fryfogle among guys who were getting between a fifth and a fourth of their teams’ targets and still adding about 4 points per game in value. That group is well behind the elites, PSU’s Jahan Dotson, Minnesota’s Rashod Bateman, and OSU’s Garrett Wilson and Chris Olave.
Interesting outliers among Big Ten players were Maryland’s five-star freshman Rakim Jarrett, who only got 14% of team targets but did a ton of damage with them, and a few guys who were over 30% of their teams’ targets but didn’t produce that much, such as Whop Philyor, Wan’Dale Robinson, and Dontay Demus
I can also show you the good ol’ fashioned yards per target leaders in the Big Ten (min 15+ targets, drawn penalties not included):
| Target | Offense | Pos | Tar | YPT | EPA/G |
|---|---|---|---|---|---|
| Chris Autman-Bell | Minnesota | WR | 35 | 12.3 | 3.70 |
| Rakim Jarrett | Maryland | WR | 20 | 11.9 | 4.93 |
| Garrett Wilson | Ohio State | WR | 65 | 11.4 | 6.09 |
| Chris Olave | Ohio State | WR | 57 | 11.2 | 5.82 |
| Jahan Dotson | Penn State | WR | 79 | 10.6 | 5.53 |
| Austin Allen | Nebraska | TE | 25 | 10.0 | 1.25 |
| Brian Cobbs | Maryland | WR | 20 | 9.9 | 1.55 |
| Miles Marshall | Indiana | WR | 36 | 9.9 | 2.38 |
| Ty Fryfogle | Indiana | WR | 69 | 9.8 | 4.03 |
| Ricky White | Michigan State | WR | 24 | 9.7 | 2.31 |
| Ihmir Smith-Marsette | Iowa | WR | 37 | 9.7 | 2.67 |
| Ronnie Bell | Michigan | WR | 43 | 9.5 | 3.88 |
Nick Eubanks isn’t on the above because his 12 targets last year were only 6% of Michigan’s passes, but he was at 9.8 YPT. The frightening thing is all of these guys are coming back.
Speaking of HTTV, when can I give you guys money for that thing?
When Brian posts the dagnab Kickstarter.
What's up with Blake Corum being grayed out in the Michigan table?
Just some stray HTML code. Fixed.
Thanks, Seth. Had me wondering if I missed some major news regarding him.
NGL, was hoping this would be an exhaustive review of all targeting ejections.
Same! Although this is good stuff anyway so it's all good.
I thought the same thing.
The frightening thing is all of these guys are coming back.
Dotson/Clifford, Autman-Bell/Morgan, Jarrett/Tagovailoa, and Fryfogle/Penix could all put up some good numbers. But, Olave & Wilson are breaking in a new first year starter. Although the death star of OSU will probably keep chugging along without skipping a beat. I'll be most curious to see if Clifford can take a step forward. UM desperately needs to find a QB that can keep pace, but Bell should be the guy to help him if he can get him the ball accurately.
I'm concerned with our corner backs and secondary. If we can't contain and stop the deep pass, then this year might be even worst than last year...
OSU loves to rotate a lot of WRs but they didn't last year due to shorted/COVID season. They're planning on resuming rotating WRs this season. Word from C-Bus is that they expect Jaxon Smith-Njiba to be their breakout star. The following season likely will have another WR death machine with JSN, Julian Fleming, Marvin Harrison Jr. and Emeka Egbuka. Don't count on OSU to have a step down at WR as long as Brian Hartline is their WR coach. He's going to be a future OC/HC down the road.
Don't count on OSU to have a step down at WR as long as Brian Hartline is their WR coach. He's going to be a future OC/HC down the road.
Is he a good coach or just a good recruiter? Honest question.
It's both. He was able to develop non-touted WRs into NFL draft picks like McLaurin, Olave. OSU used to have trouble attracting elite WR talents until Ryan Day and Brian Hartline showed up and now they're basically landing top WRs because OSU has shown that they can develop them into NFL draft picks and they're passing the ball a lot more than they used to.
It's both. He was able to develop non-touted WRs into NFL draft picks like McLaurin, Olave.
McLaurin was a 4-star top 250 guy. For comparison's sake he was higher ranked than Giles Jackson. Olave I'll give you, but he seemed like a big time misjudgement on his recruiting profile, based on idk his low weight? But, he seemed pretty skilled from his true freshman year, although I guess in fairness that was also Hartline's first year at OSU.
OSU used to have trouble attracting elite WR talents until Ryan Day and Brian Hartline showed up and now they're basically landing top WRs because OSU has shown that they can develop them into NFL draft picks and they're passing the ball a lot more than they used to.
Same with QBs. To me it seems like it's more a product of Day's offense and Hartline's recruiting. Maybe he's a good coach too, but the attraction of all the guys seems more like Day's offense than anything else. Either way it's working out great for them and they have more 5-stars at WR than they can play at once. They've already had two transfers at WR that were higher ranked recruits than anyone we've got (Williams & Cooper).
The quality of Hartline's coaching ability was evident when transitioning from the worthless hot-head a-hole who coached a WR group that was sloppy & underperforming for years. Hartline immediately came in and the receivers were significantly sharper, better route-runners, etc.
The quality of Hartline's coaching ability was evident when transitioning from the worthless hot-head a-hole who coached a WR group that was sloppy & underperforming for years. Hartline immediately came in and the receivers were significantly sharper, better route-runners, etc.
If memory serves me right Day got to OSU in 2017 and Hartline in 2018. The big offensive production jump happened in 2018 when Haskins became the starter. How do you differentiate the offensive passing/receiving production from either a) Day vs Hartline/Smith or b) running QBs like Barrett vs better passing QBs in Haskins/Fields? It seems just as likely that the improved passing & receiving numbers are due to Day's offense and better throwing QBs as it is Hartline's coaching. Ultimately the reality is we have no idea, but he sure can recruit and you can't argue with the production in the last 3 years.
You can isolate WR's impact without looking at QB. You can see how they run routes, how they get a feel for coverage and how they block out in the perimeter. It's a huge improvement over than what they had in the past.
Does it help to have a good QB throwing the ball to them? Yes, but it's not necessarily the one or the other. OSU has begun to churn out NFL caliber WRs.
Yes, Michigan didn't get much help from QB play but there is a reason why their WRs aren't as highly regarded despite the talent/athleticism. See Nico Collins. His biggest issues are a result of bad WR coaching. If it were that easy to get 4*/5* WR and let them do their thing, Michigan would've had Black, Collins and DPJ drafted at 1st round based on your logic. Saying they have 5* talent is just a lazy take IMO.
You can isolate WR's impact without looking at QB. You can see how they run routes, how they get a feel for coverage and how they block out in the perimeter.
You're saying that as if you've broken down the film. If so, let's see the evidence. If not, then you're just speculating and should be saying "I think..." vs "You can..."
It's a huge improvement over than what they had in the past.
I'm calling your bluff. Prove it.
This is where haloscan would be nice. I want to vote "informative" but not upvote b/c osu. That said...
Informative.
How many expected points does a receiver lose when he drops a game winning pass in the end zone that hits him right in the chest with no defenders around?......just asking for a friend.
A few random thoughts:
A. Actual football content is good. Very, VERY good. Thank you Seth.
B I have no idea what this all means and if I should be happy, sad, pissed or what. Seth's ability to accumulate and present data is far above my tiny brain's ability to process it.
C. But it's football content. So yay!
I'm going to re-read this again. I doubt I'll understand it any better but being that it's football content I'm definitely going to enjoy it.
All you need to know is that OSU has 2 receivers that were way more productive than our current best receiver. So use that info and channel your anger, despondence, apathy accordingly.
To add some nuance, yes Ohio State has good players at WR, but only looking at stats doesn't account for other important factors, such as scheme and personnel packages. It should be expected that pass oriented offenses have more passing stats.
Michigan's offense last year still had a lot of Ben Mason and two tight end personnel. If they ran more three and four receiver sets like OSU, you would expect more targets, yards, and stats for the WRs. Guys still need to execute to keep drives alive, but I wouldn't be surprised if the offensive stats start to look more like OSU this year.
Year one under Gattis was a big transition. Year two was shortened and had a ton of shuffling, but many of those issues shouldn't carry over into year three. If nothing else, at least this year we'll get to see a whole season and get a clearer picture of what this offense is trying to be.
We might not be very good at football.
But the next QB will fix everything....just like we have thought for the last 15 yrs.
Also I have read several times that one of our strengths is the WR corp. If so, this news does not bode well. Now a competent QB may help alleviate some of those issues, but I think it goes well beyond that at this point. Its gonna be a long season.
Last season 4/5th of the O-Line went to the NFL. No spring practice, messed up summer, fall training camp altered, first year starting QB under center, one of the best WR's opting out. All of that is not a recipe for success. Even with only 6 games of experience from last season and having spring practice, the team should be much better off than they were last season.
Many teams had to deal with issues outside the norm but I wouldn't say the OL was an issue when we had the least sacks in the Big Ten. The problems last year were largely QB play, WR play, cohesive call playing and execution.
Fascinating stuff! Would be interesting to see some historical data for receivers like AC, Desmond, Terrell, and so on.
What it all means to me is Ronnie Bell is a good WR but not nearly an All Conference hopeful. And he is probably already near his ceiling so expectations for a statistical leap are low. After Bell, Michigan has got a crew of electrons with no track record of success and a TE that has issues performing on Saturday.
The 2021 Michigan football team will go exactly as far as this offense can drag the defense. Given what we've seen out of Gattis that probably means how far this passing offense can drag the defense. Which means it may come down to the electrons.
On the negative side, while the Gattis offense has shown sparks of promise it has struggled to knit those sparks into a cohesive full game performance.
On the positive side (at least moving forward--certainly not retrospectively), the 2020 team was awful for reasons that may have been pandemic related. And the electrons were hand picked by Gattis. If he goes down, at least it will be with his guys.
I downvoted to give you a negative one, the charge on an electron. Trying to make "electrons" work for wide receivers is like trying to make "fetch" happen.
I am at +1 so maybe I should edit my original post to call them protons!
I upvoted this comment. Conservation of charge sort of thing.
An atom walks into a bar and says, "I lost an electron."
Bartender asks, "Are you sure?"
The atom says, "I'm POSITIVE!"
So Michigan receivers, individually and as a group are not able to paper over any deficiencies elsewhere in the offense? Hoo boy!
They are the deficiencies in the offense.
I would put QB play in the same bucket. Milton struggled throwing the ball and Cade was healthy for 5 quarters.
I am not disagreeing that the WRs are a concern--at least past Bell.
I am mostly suggesting there may be bigger concerns. We just don't know. In 2019, the QB play was better and we know the WR unit was good. And yet, Gattis still didn't get the expected performance out of the passing game or individual players.
I am not sold that the Gattis offense works yet. I'll need to see that before I start pinpointing individual player weaknesses. Besides QB play in 2020 which was clearly lacking. Being a good coach means putting your players in a position to succeed. Not continually blaming the players when you never succeed. Gattis picked these WR. He went after smaller faster guys. If they are the weakness. It is totally on him.
They are one of the deficiencies in the offense
There, fixed that for you
To some extent. As Gisele had so aptly put after the second Giants SB, the QBs can't throw and catch the ball on the same play; I think there's enough responsibility to go around on the offensive side.
Parts work well, but as soon as the whole needs to perform, everything falls apart.
Paraphrasing: Michigan ran a lot of Mason + 2TE sets -- with unimpressive TEs to boot -- while OSU ran many 3 and 4 WR sets.
The philosophy of the offense is the overall deficiency. It's not modern, it doesn't take advantage of the best players, and when coupled with a shocking amount of predictability, it has a low probability of success.
Man. Seth is the best. Thank you for this great research and content.
So basically, this shows that our receivers/QBs suck AND OSU is return 2 All-American level receivers in addition to the 5*WRs Wilson, Williams, Fleming, Njigba, Scott, and Egbuka, they have from the last 3 recruiting classes. Yep, this is why I'm a UM basketball fan.
Scheme is more important than stars. (I wish there was a scheme the last two years.)
Well, so far the Gattis offense has turned 5-star and 4-star recruits into 3-star players.
Did you correct for really only tries to throw the ball in long yardage situations or down multiple scores and pre-settles for FGs in any end of half situation? I feel like that hampers production a bit.
Dixon and Anthony may get more playing time if Johnson and Sainristil do not make bigger leaps this season. QB play will also be a key!
that was a lot of words and data to tell us a thing we already know, which is that Ronnie Bell is an NFL quality possession receiver.... unfortunately we didn't have a QB (or coaching staff) who seemed to realize that other people could be targeted as well (pre last years stuff, last year is a giant wash of terribleness)...hope Johnson and the rest of the crew make a leap this year... that said, great to see this level of work and analysis
Comments