Analysis of QB option reads vs Army, 1st half
One of the prevailing themes out of the Army game was Shea refuses to pull the ball on reads and it's killing our offense. I myself bought into this hypothesis, and actually went even farther to suggest reads were simply not a part of the offense in the second half.
Still, watching live, it's extremely difficult to determine the correctness of reads, and I could not say for certain if handing the ball off almost exclusively was a Shea thing or something Army was dictating, so I went through and looked at each play in detail.
Generally my criteria for a correct read is a fairly low bar of was the threat of a QB keep enough to take that defender out of stopping the RB within a reasonable distance (looping back and tackling 10 yards downfield in pursuit doesn't count). I also tried to see how many defenders that left Army in the box, once you remove the optioned guy, vs how many blockers.
I had intended to make gifs of all of these, but Fox seems to be extremely aggressive in removing IP from youtube and I tried to make the gifs from 3 separate videos before each was taken down. I finally gave up and just took screenshots from my YTTV recording. If you want to follow along with moving video, the down and distance and game clock are listed for each play.
Play 1, 1st and 10 on M27, 12:54 1st Qtr
RPO with Shea reading the OLB circled. He drops into coverage, so give is the correct read here. Michigan has 5 vs 5 in the box and the RB gets an easy 1st down. The optioned player gets a hand on the RB 9 yards downfield.
Verdict: Correct give read
Play 2, 1st and 15 on M46, 11:19 1st Qtr
Army has a scrape exchange on between the OLB and MLB on the left anticipating zone read, but I'm pretty sure Shea is again reading the circled safety for an RPO. The safety again drops into coverage to buzz the slant so give is the correct read here. The scraping MLB runs himself out of the play leaving M again with 6 blockers for 6 defenders, but unfortunately Ruiz and Onwenu bork their combo block and leave the other MLB free to tackle. Turner trucks him and gets 11 yards anyway.
Verdict: Correct give read
Play 3, 1st and 10 on M30, 4:53 1st Qtr
This is a standard zone read on the circled DE, who's crashing hard. He ends up tackling the RB at the LOS, so this is clearly a bad read
Verdict: Incorrect give read
Play 4, 3rd and 1 on M39, 4:00 1st Qtr
Michigan has a mesh point but Shea's staring off into space and there's no read. With the down and distance, likely M just wants to block the front and have the RB fall forward for the yard. Charbonnet meets an unblocked LB in the hole and does just that.
Verdict: Not an actual read
Play 5, 2nd and 10 on M42, 3:30 1st Qtr
The read is on the DE whose first two steps are straight down the LoS. This should be a pull. At the last second, the DE anticipates pull and stops his momentum, leaving him caught behind a mess of bodies. McKeon loses his block pretty comprehensively, but Charbonnet mansomely grinds out 4 yards anyway.
Verdict: Incorrect give read
Play 6, 1st and G on O9, 1:15 1st Qtr
Both eligible receivers on the right of the play pull the other direction. The CB is left with no coverage responsibilites and is the read. He stays home, discouraging the pull. Michigan has 7v7 in the box with an OLB shaded out over the slot just outside and this should go well. Black oles a crack block on the OLB though and that guy tackles after a gain of 4.
Verdict: Correct give read
Play 7, 1st and 10 on M6, 12:02 2nd Qtr
Standard zone read on oLB on the end of the line, who tries to play both sides of the read by slowly shuffling. Give is marginally the correct read here. The OLB has to fight through a ton of traffic to be in a position to help 7 yards downfield. Minus the optioned OLB, there's 6v6 in the box with no safety over the top and the play deserved to go a long, long way except Onwenu released into the same LB as Bredeson, leaving the other LB free to tackle. Still a gain of 4.
Verdict: Correct give read
Play 8, 2nd and 6 on M10, 11:32 2nd Qtr
Another read on the OLB on the end of the line with a TE arcing around. OLB plays it pretty well and makes the read difficult. Shea gives and while the OLB does make the tackle, he has to give 5 yards to loop back to the play and the RB drags him for another 2, picking up the 1st down. For what it's worth M has 7v7 in the box, even if you account for the TE taking the "optioned" guy. Tough call, I'm going to go with incorrect read here with the TE pulling around there's a good chance the keep gets more than the give, but I don't know what M's rules are in this situation.
Verdict: Incorrect give read (marginal)
Play 9, 1st and 10 on M17, 11:06 2nd Qtr
RPO play with a read on the OLB, he stays in coverage so give is the right read, although he's shuffling and making this a close call. That guy is removed from the run play except to clean up a cut back 5 yards downfield. M had 6v6 in the box but the DE slants hard under Mayfield and gets a tackle at the line. Hayes gets called for the hold opposite him. I'd love to see Charbonnet read the slant and cut this back because the backside edge was wide, wide open.
Verdict: Correct give read
Play 10, 1st and 18 on M9, 10:47 2nd Qtr
Read on the shuffling DE with a jet motion to freeze the OLB on that side. DE makes it pretty close, but give is the right call and the DE has no shot at a tackle. Michigan actually has 7 blockers in the box (including jet WR) and only 6 defenders not counting the optioned defender, but Hayes ends up coming off his combo and blocking the optioned guy instead of climbing to the LB. Either way, M has a hat on a hat all the way to the safeties and Charbonnet grinds out a couple for a gain of 11.
Verdict: Correct give read
Play 11: 1st and 10 M34, 9:41 2nd Qtr
RPO on the walked up safety. He doesn't exactly come downhill, but his starting position is close to the line and he's shuffling down after the snap. IMO, this should be a throw, but it's pretty close. M is down 6v7 in the box not even including him, and he's able to rally to the ball at 4 yards. Doesn't matter as BVS fumbles (ugh).
Verdict: Incorrect give read
Play 12: 3rd and 1 on M32, 1:58 2nd Qtr
I don't think this is an actual read despite the mesh. Everyone on the line is blocked and the receivers aren't going out in real routes so it can't be RPO either. Given down and distance, this is probably just a designed handoff. Safety screaming down gives M 6v7 disadvantage, and Mayfield can't step in front of a blitzing LB.
Verdict: Not an actual read
Play 13: 4th and 1 on M32, 1:36 2nd Qtr
Same thing, no one read, no routes. Straight up goal line with everyone bunched at the line, M is 9v10, but middle of the line lurches forward to (barely) pick it up.
Verdict: Not an actual read
Play 14: 3rd and 2 on M45, 0:46 2nd Qtr
Read is on OLB on the left end of the line. He's shuffling down slowly, but mostly forcing the give. He's not even close to being able to touch the RB, FWIW. M has 6v5 as Army is conceding the 1st down and playing prevent coverage given the time left. Big hole and easy conversion, probably more if Collins doesn't whiff his block.
Verdict: Correct give read
Play 15: 3rd and 4 on O27, 0:07 2nd Qtr
Army has a nasty defense called and comes with a extremely well disguised CB blitz, switching the S on the WR. The end that's being read crashes hard inducing the pull, which Shea obliges. Unfortunately, he should be aware of the CB coming behind him (remember, it's a zone read, not a player read). So, his one pull of the day is dead to rights and Shea makes it maximally painful by throwing it OoB short of the LoS drawing an intentional grounding.
Verdict: Incorrect keep read.
Summary: Michigan Ran 15 plays in the 1st half with a mesh point, of which 12 were actual reads.
Of the 12 actual reads, 4 RPO and 8 zone read.
Shea made the correct read overall on 7 of the plays, 3 RPO (75%) and 4 zone read (50%) although to be fair at least one was a pretty close call. Only one of the incorrect give reads were so bad that it resulted in the optioned player tackling the RB close to the LoS.
Of the 4 incorrect zone reads, 3 were gives and his lone keep was a bad read.
When the read was done correctly, overwhelmingly Michigan had at least the same number of blockers as (non-optioned) defenders in the box, and in some cases even gained a free blocker. Many of these were unfortunately not paid off because of one or two key blocks being missed, or Army slanting hard and blowing up the gaps. Even still Michigan still managed decent chunks.
Next time, I'll take a look at the second half plays. I'll also editorialize a bit more on what I thought the issues were that prevented us from being more successful, but I'll hold my thoughts until I've looked at the full game in detail.
September 12th, 2019 at 12:46 AM ^
Great analysis, thank you for posting. Maybe we've all been a bit too hard on the general offense, because they fumbled away so many first half possessions.
On the other hand, you can't really call a full-time 2 TE unit Speed in Space, and I predict the second half carries will look more 25% correct/75% incorrect than 50/50. I'm looking forward to seeing what you think though.
September 12th, 2019 at 1:24 AM ^
That last play (15)-- the corners to the bottom of the screen are lined up 10 and 12 yards off of the receivers :( Am I missing something here? I hope at some point they add the opportunity to audible to a hot read here. Are quarterbacks taught to look at the entire field on called zone reads? or is he so focused on the relevant players to the play call?
Also, don't really understand the play call to run a zone read here. The previous play was an incomplete pass so the clock is stopped, there's 8 seconds left with 1 timeout left so it looks like they wanted to run a 'safe' play, call a timeout and kick a field goal? I'd much prefer more aggressiveness here as you're already looking at a 43 yarder which is makeable.
September 12th, 2019 at 3:05 AM ^
There is a similar play in the 2nd half when it was 3rd and 3 and the DBs were playing 10yds off, including on the slot receiver. Have some hand signal or something to get a hot read and pick up the gimme first down. Instead we plowed straight ahead and failed to convert on 4th down.
September 12th, 2019 at 8:38 AM ^
I agree, but I will say that IF Shea is having a hard time with the reads, the worst thing to do is probably add more too good responsibilities at this point. Maybe in the second year in the system you add hot reads? Later in the year of her looks better? But not when it seems like he's already struggling a little.
September 12th, 2019 at 9:37 AM ^
Michigan did spurn some easy yards here and there as DBs were playing off, but this one I don't have a problem with. 0:08 is probably pretty tight for a shot in the endzone from that distance. They're just looking for a safe play to get what they could, call a TO and kick a hopefully easier FG. The actual outcome wasn't good, but a zone read is pretty safe in that situation as long as you're QB doesn't lose his mind.
September 12th, 2019 at 9:34 AM ^
Great stuff. What would be par for the course on the QB percentage of correct reads, and is the zone read easier or more difficult than the RPO?
September 12th, 2019 at 10:14 AM ^
Somebody more knowledgeable than me should answer this, but a ZR and an RPO are essentially the same thing. They are both reading an area of the field, or a particular player, to see what the defense does. That 'what' determines what the QB does with the ball. In the case of a ZR, he either hands it off to a RB, or keeps and runs himself. For the RPO, he either hands it off (or runs himself) or throws it. But both plays depend on what the defense does. Read them correctly, and the ball goes where Michigan has more players. Get it wrong, the ball is destined for a disadvantageous outcome.
I think the difficulty of any read depends on how well the defense can disguise what its doing, and whether anyone on the offense misses a block. It would be interesting to see some percentages of success/failure about how often reads are correct. But that probably depends on lot on who's playing - both the teams involved and the individual players.
September 12th, 2019 at 10:57 AM ^
My correct/incorrect binary grading is probably way to simplistic. In reality, defenses are very good at making reads difficult and most reads are defensible either way. A shuffling DE that tracks down a RB at 4 is probably a bad give read when a keep likely gets 6, but it's not a slam dunk one way or the other way either. I could easily have graded all the reads mostly correct except the one crashing guy tackles at the LoS, and the pull into a blitzing CB. The other 3 are judgment calls on comparing success of the give vs the hypothetical success of a keep.
September 12th, 2019 at 10:07 AM ^
Thanks for taking the time to do this. I'm going to disagree with you on play 8. To me thats a correct give. I could be wrong on this, but wouldn't a keep be the correct play when the optioned guy crashes down giving Shea the ability to beat him around the edge? That leaves Eubanks 1 on 1 with the DB thats at the 18 yd line. This would have the potential for a huge gain. Since the optioned guy stayed there, if Shea keeps that means Eubanks must block the optioned guy. Let's say he does this successfully, IMO the DB at the 18 yd line has plenty of room to come up and make the play. Sure, Shea might pick up some yards, maybe even get the first down (not likely IMO, hes playing hurt), but I'd much rather have Shea not taking unnecessary hits, especially when the yardage potential is roughly the same as the give.
Interested to see you do the second half cause that's where the offense really went bad.
September 12th, 2019 at 10:29 AM ^
I struggled with that one, it could easily have gone in the "correct" bucket. The conservative, standard approach is to have the TE kick out the shuffling end if he's trying to play both sides so he has no shot on stopping the give. The keep is riskier and requires the TE getting around and sealing him, but it's a lot of yards if it works. The end plays it really close here, but IMO, it's close enough to try to arc. The TE only needs to harass him a little and can still get out for the safety.I don't know what options Michigan has here for Shea, so hard to say if he made the right call in the context of the play design.
As I'm going through the 2nd half, I think I'm going to need to separate the 50/50 or 60/40 calls from the clearly incorrect, play killing ones.
September 12th, 2019 at 10:49 AM ^
That’s true, I didn’t think about Eubanks pulling off a combo block.
September 12th, 2019 at 10:33 AM ^
This is excellent, thank you.
Maybe one thing to think about with your editorial - did Gattis get beat by a good DC, should he have done something differently?
September 12th, 2019 at 11:25 AM ^
This is great and a very interesting analysis. However, after reading Seth's piece from yesterday and hearing the various interviews, it only deepens my confusion and concern. What I am most interested to see are the two 4th and 2 calls (provided they were reads of this sort.)
I can live with the focus on the run in the second half. Provided there was a reason to think it would work based on the defensive alignment and play call. And those two 4th down calls are the most critical. Neither was particularly close to being converted.
Unfortunately, it seems as though Army was continually prepared for this play and Michigan still continued to call it. The success of this play decreased dramatically the more they called it and it never hit big (regardless of whose fault that is.) You would hope that all of this was to set up the defense for something good when you needed it. Which is those two 4th down calls. I would prefer to punish the defense whenever you can predict what they are going to do rather than backing yourself into a corner keeping your ace card hidden but OK.
The strategy, as stated by Gattis, hinges on getting those two 4th downs. If you don't dial up a counter for those two plays and just keep calling the same thing the defense is planning on stopping (like on the 3rd down late in the first) and has decreasing effectiveness then what are we doing here? I doubt anyone thought based on previous results that a straight ZR arc (or whatever the terminology is) had a very high probability of being successful. If Harbaugh asks "do you have a good call" and Gattis says "yeah, the same play as the previous 10 plays." Then what are we doing here?
Add in the idea that Michigan had advantages at the point of attack with blocking (the play "worked") which Gattis confirmed with his stuff about "some 6 man blocking fronts" and Ruiz confirmed by saying the OL blocked really well and Army didn't do anything unexpected. And one really wonders what we're doing here. Everything works, everyone does their job, on a critical play and yet the interior OL of future NFL players can't push Army for 2 yards.
What are we doing here? Why does Michigan keep calling a play that either they can't execute or doesn't work? Do they only have one play?
September 12th, 2019 at 1:51 PM ^
The first failed 4th was just comprehensibly poorly executed. There was no real cleverness by either team, and no read, just man on man push forward or don't. 3 separate M players, and the ones on the side the play was going, blocked no one at all due to miscommunication.
The second was blown up by a perfectly timed (possibly too early) CB blitz. The read was on this guy, who screams at the mesh point. I think the give is the correct call here as he ends up hitting McCaffery and doesn't impede the RB at all. Unfortunately, Eubanks stops his pull trying to decide whether to redirect for the CB and instead blocks no one. His assignment, the OLB, is free to tackle at the line and then the safety who was coming downhill the whole time arrives to clean up.
For 2 yards, I don't necessarily hate the playcall, although I think they need to consider going under center if they're going to just run it straight up. Michigan snapped it from center on their first goal line TD and it was successful, so you know it's in their playbook still.
September 12th, 2019 at 12:17 PM ^
Great work and easy to follow with the screenshots. In real time, my eye isn't trained well enough to see the entire field (this is from years of casual ball-watching football fandom). This is extremely helpful. Keep up the good work!
September 12th, 2019 at 12:18 PM ^
Someone pointed this out in the neck sharpies yesterday and I get that these screen shots are a few seconds after the snap so the DBs could have back-pedaled, but on Plays 11, 12, 14, and 15 there are WRs without an Army player within 10 yards of them. Do other teams have checks at the line where the QB can call a quick pass to the WR if the other team is playing off coverage?
September 12th, 2019 at 3:32 PM ^
Since Gattis is the box, seeing the whole field, why can't he communicate down to the field, have the folks sending in the play call the hot read, and have Shea execute? It'snot like we're playing too fast to notice this.....
September 13th, 2019 at 9:08 AM ^
Is it possible that the team has never practiced that? If the team has never practiced it, can they run it? I've never played so I don't know how this works. Borges was resistant to bubble screens week after week after week, but I am hoping Gattis just didn't have a WR screen installed, and that he'll have changed that before we play again.
September 12th, 2019 at 1:13 PM ^
Outstanding Analysis. This is why I LOVE this community! Bravo.
September 12th, 2019 at 1:30 PM ^
This is well done. Thanks for posting.
September 12th, 2019 at 2:45 PM ^
This is great stuff especially for those of us that have never coached but want to understand more about what is going on...I have only one question, and it's on play 2...if I look at it from my unschooled football mind, it appears to me this could go either way...that Turner trucked the guy and got 11 is a good result, but was it the right call? Shea is facing the top of the screen on the read, Hayes and Bredeson double teaming one guy. It would seem Hayes could slide off to the linebacker, Eubanks has the safety OR there is enough blocking for the wideout to blow by his coverage. Could this have been a pull and either run or pass?
September 12th, 2019 at 3:52 PM ^
I think play two is simply a pull and pass or give based on the one guy circled and whether he steps down or back into coverage. It's possible this is a read on the MLB scraping outside, but it looks like Shea isn't tracking him. Also, the QB fake throwing motion instead of a fake run after the give is another clue this is RPO.
Either way, the OL is zone blocking and Bredeson and Hayes actually have a guy each (can't see it in the screenshot). I'm not 100% sure who screws up here. Onwenu starts doubling with Mayfield and then moves to help Ruiz, so 3 guys only block 2 when someone should be climbing to the LB. My guess is this is an Onwenu issue.
September 12th, 2019 at 3:03 PM ^
Good stuff. A question on the no read plays: if there is no one outside to read, doesn’t that scream KEEP? To the fan watching at home, those plays are the most frustrating.
September 12th, 2019 at 3:59 PM ^
The play is blocked up for an inside run with the edge blocks set up as kickouts and the edge defender set up with outside leverage to keep contain. It's basically like a read with the optioned guy 100% staying home to contain, because in this case there's a blocker keeping him from crashing.
September 13th, 2019 at 11:43 AM ^
Fine, but in the screen caps you used, the EMLOS is being blocked. If Shea can't beat a blocked guy to the edge, the whole #SpeedInSpace thing is a sham. If the tackle can just hinder the end a little, Shea should be able to turn it up field and then all we need is one block from a WR. I'd rather depend on 2 blocks working than 6.
September 13th, 2019 at 4:04 PM ^
He's blocking him at the wrong angle. The only way an tackle can hinder a defender on a kickout from chasing outside is to hold him.
September 18th, 2019 at 2:19 PM ^
Have you ever tried to run with someone pushing you in the shoulder?
September 12th, 2019 at 3:28 PM ^
Thanks for doing this, reshp1 !
Looks like Shea did a bit better job than I thought, and the blocking did a bit worse than I thought.
September 12th, 2019 at 3:41 PM ^
This was a great and very eye-opening read. Thanks for putting this together!
September 12th, 2019 at 4:34 PM ^
Zach missed a hole on one of the 4th downs which should have netted the yardage. He pushed left into the pile vs shooting straight up the tight opening. Should have been worth 2-3 minimum.
September 12th, 2019 at 9:53 PM ^
Brian - great UFR, but I disagree on these:
- Play 3 - this is a scrape and Bell is way out of position. Keeper would be a TFL.
- Play 5 - unless we are in triple option pitch to Bell, this is correct read
- Play 8 - wow, look at that hole up the gut - total give
Otherwise I agree ... the issue with this offense is not Shea on run read options. The issue is pass pro, fumbles, penalties, way too many drops and a few missed throws (but not many). I think if we clean that shit up we are tough to stop on offense.
Go Blue !!!
Edit - fans need to back off booing Shea and the team. McCaffery is NO WAY near ready. Looks like a Stork who can't pass nearly as good as Shea. Both games he got away with throws that should have been picked off ... on like 5 attempts. Not good.
September 13th, 2019 at 10:24 AM ^
Great work! What seems fairly obvious about Josh Gattis is that he's still pretty green behind the ears as an every down play caller. Having more film time, analyzing every snap and each decision he's made in the two games thus far, and then finding areas to improve upon are equally important for him as they are for the players. How good/great this offense can be has yet to be determined, but I think we can all agree that we should be past the worst of it.
September 16th, 2019 at 6:39 PM ^
September 23rd, 2019 at 9:03 PM ^
So..... are we past the worst of it, now? I feel dumb for thinking Army was as bad as it would be this year. Ouch
September 13th, 2019 at 11:41 AM ^
Thanks. Maybe because these are NOT moving--and not crapped up with ads I have to struggle to remove--I see not just what you're saying but how the offense is supposed to work better than I ever have after a first review. Will read again. . .
Comments