Sunday, November 2, 2014

UCLA Defensive Gameplan

Up until the UCLA game, our defense had never given up more than 5 ypc against the rush, and had only given up more than 4 ypc twice. We'd only given up 150+ yards one time (to Colorado), and had held everyone else under 120. In the UCLA game we gave up 237 yards at 5.04 ypc. Why the sudden change? It's tempting to say that they're just better than the teams we'd played up to that point, but on the season they aren't that far ahead of UA. As always, it's a lot more complicated than the size, strength, and speed of your opponent.

I've always found UCLA's offense under Noel Mazzone/Brett Hundley to be dreadfully boring, but no one can deny that it's been effective.  It's been said that Hundley is the UCLA offense, and earlier in the year he was getting Heisman consideration, but he's never been as dominant as all that. The UCLA offense has stood out in two main ways in the last three years: (1) in 2014 and 2012, they've had RB's in the top-15 nationally, and (2) Hundley's completion percentage is #2 in the country this year. I don't think that anyone has ever called a mobile QB in a spread offense a “game manager,” so let me be the first: Hundley is the A.J. McCarron of spread offenses. The yards/attempt, total yards, big passing plays, 3rd down QB rating, and passing TD's just haven't been anything to write home about. That's not to say he's not good.  A.J. McCarron had some huge games and won multiple NC's.  Hundley's consistently one of the best QB's in the conference, but he's certainly not being used like a Heisman candidate in UCLA's offense.
Philosophically, UCLA's offense is a lot closer to a RichRod or Urban Meyer offense than to other types of spread out there. The offense starts with a strong zone running game by the RB. Everything else mostly exists in case teams overload to stop those runs, which is why the passing game is more efficient than it is productive. Because of this philosophy, the best way to measure Hundley's true effectiveness is not to look at his own stats, but to look at how effective their zone running game has been. To get into our problems with the run against UCLA, this post will look in depth at zone running. I've introduced the basics of zone schemes elsewhere, so in this post I'll go more in depth into them before building toward understanding the kind of complex problem solving that goes into stopping a rushing attack like UCLA's. Also, I won't look at our offense in this particular post, but we run both zone plays a lot, and if you get the reads and assignments given here you can easily transfer that knowledge into breaking down our rushing offense.

Finally, I should note here that explaining why an offense is difficult to stop doesn't excuse a defense for failing to stop it. The point of a post like this is that the solutions that need to be found are more complicated than they might seem, and that there's a lot more to it than just putting more men in the box or having a good idea for what play is coming.  The better the DC, the better they are at this kind of complex problem solving.
 

History
But first, a few words on the development of the zone running game. Zone runs gained popularity in their modern form in the NFL in the mid-90's with Terrell Davis and the Denver Broncos. The concepts existed long before then, but the zone play really took the league by storm starting from that point. From there these schemes have entered the college game in at least two ways. One way is through college staffs with NFL connections implementing the scheme with their college teams. A classic example is Carroll's decision to make USC into a zone team about halfway through his time there. Alabama also runs a ton of zone, which is no surprise given Saban's links to the NFL.

Additionally, spread teams started using zone plays in combination with a QB read. In the mid-90's, as zone running was becoming more trendy in the NFL, Rich Rodriguez was trying to run it at Grambling State. His team was struggling, his QB came up with the idea of reading the DE instead of trying to block him, and the “zone read” was born. The influence goes from college to the pros as well as from the pros to college, though, as evidenced by the best free source you could ever want on zone blocking: (https://www.youtube.com/watch?v=d3fRunvq4hA&list=UUeHB5VbZYLMBbiAsEEyw4LQ). Here you see Alex Gibbs, then the Falcons' OL coach, leading a clinic with Urban Meyer's Florida staff. It's like 8 hours long, and it's awesome. Why would a well-established NFL line coach meet with a college staff? As you'll see if you watch the video, he's intrigued by the QB run game off of zone blocking. If Vick hadn't gotten into dogfighting, Gibbs very well could have brought the zone read to the NFL half a decade before the likes of RGIII, Wilson, and Kaepernick. The point of all of this is that there's a tendency to see zone running as a “spread” play and Stanford's power run game as “pro-style,” but that's not the way it is at all (and in fact I've seen Stanford run a lot of zone this year).  Zone running is an essentially pro-style concept that's continued to develop in both college and the pros through dialogue between spread and pro-style coaches. It's no surprise that Carroll, who made the decision to convert USC into a pro-style, zone running team, was also one of the first coaches to bring the college-style “read” element into the NFL.

Zone Blocking
So what are these schemes? At its heart, a zone play is called “zone” because the entire offensive line steps in the direction of the play and blocks whoever shows up in front of them. On these plays there isn't one specific hole that the RB is supposed to head to, but instead he has a series of reads that takes him to openings in the defense. We'll start out looking at outside zone:
I've drawn this from a normal I-formation against an Under front. This is the idealized version of this play. This is outside zone to the left, which means that the TE and all of the OL will step to the left, but who do they block? Before the snap, the OL are asking themselves if they are “covered” or “uncovered.” A “covered” lineman has a defender on the LOS to the playside of him (to his left in this diagram). An “uncovered” lineman has no one on the LOS to his left. So, in the diagram above, the TE, LT, C, and RT are covered, while the LG and RG are uncovered. As a lineman, if you're covered, you're trying to “reach” the guy who's covering you. This means that you want to come across them to seal them off from the outside. If you're uncovered, you're double-teaming with the guy next to you. To see this, we can look at the LT and LG in the diagram. The LT is covered by the DE, and the LG is uncovered. The LT is going to try to “reach” the DE and drive him back inside to the LG. They'll double-team that DE until the MLB threatens to come downhill. If the MLB works outside, the LT will come off the double-team and block him while the LG stays on the DE. If the MLB flies inside and downhill, the LG will come off the double team and block him while the LT continues to reach the DE. We'll say that they're combo-blocking “from the DE to the MLB.” Similarly, the C and RG are combo-blocking from the DT to the WLB. Communicating and executing these combo blocks well is one of the hardest parts of teaching the zone play for the OL. If the two OL don't read the LB in the same way, there's a risk that both of them (or neither of them) will come off the DE, which is obviously a problem. Even if they read it correctly, on a bad zone team (or a team that's getting whipped by the front-7) guys can start to come off the double team early, leaving their partner high and dry against the DL, or too late, letting the LB run through for a TFL. It takes a lot of work to get good at this part of the scheme, but this system also gives the scheme its versatility. Note that I've said nothing about opening up a specific hole. The defense can play whatever front they want and (theoretically) can run any stunt they want. The OL always blocks in accordance with these covered/uncovered principals and things sort themselves out from there, giving zone plays an advantage over runs that try to open up one specific hole; when a run (like power or ISO) tries to open one specific hole, it becomes vulnerable to fronts that line up with a DL in that hole or that run a stunt into it.

This covered/uncovered system works its way down the OL until you get to the backside edge, where the FB kicks out the backside edge player to prevent him from crashing down the line. Remember this role of the FB for later when we talk about the zone read.

That's what the OL is trying to do ideally. If everything goes according to plan, then the RB will aim outside the TE and take the ball around the edge. This is how big gains (15+ yarders) happen, because if the SLB/DE gets reached and the MLB gets cut off, the RB's not going down until the safety gets to him.

This idealized version hardly ever happens, though. It's hard to pull off a reach block on the edge, especially because defenses will widen their edge defenders against teams that like to run zone. Add to that the fact that the blocker has to get all the way across the face of his defender to pin him inside, and it just takes a significant advantage of quickness and lateral agility to pull off. Most coaching time for the RB goes into what happens if the edge defender doesn't get reached:

Let's talk about RB reads in the context of this diagram. As mentioned above, the RB's aiming point is outside the TE. That's where he'll aim on every single outside zone play. It's important that the RB not attack the LOS immediately, because once he goes downhill the LB's will come downhill. He needs to keep them flowing laterally for a few steps so that the offensive linemen making their combo-blocks have time to get out to the LB's before they shoot their gap. The longer it takes the LB's to come downhill, the longer the offensive linemen can stay on their combo-blocks and get the level 1 defenders under control. Getting back to reads, the RB aims outside the TE, and his first read is the edge defender. If that defender gets reached, he continues outside for the big gain. If that defender doesn't get reached, then the TE will just keep driving him outside to the sidelines, and the RB will look to his second read, which will be the next down lineman inside. He reads him just like he read the edge defender. If the SLB works hard outside and the DE gets reached, then there will be a big hole in the C-gap between them. That's what I've drawn in the diagram above.

From here the process simply continues down the line until the RB finds a crease. If, at any point, the RB sees that the hole that he's aiming for is closed for any reason, he just keeps working back across the formation:

The RB keeps reading down linemen, but this cut-back philosophy theoretically works just as well even if a LB fills the hole. Here I've drawn the WLB overpursuing. The RB will see his seam closing and will cut back to his next read, cutting into the space vacated by the WLB. This all sounds very methodical. In practice, if you can get your RB to look to his third read, you're doing a fantastic job. Oftentimes it'll be Read 1, Read 2, instinctive cut to daylight. The big thing for the RB is that he can't keep running laterally indefinitely if nothing's there, or he'll get tackled for a loss. At some point early in the down he has to at least get back to the LOS, because zero yards is a lot better than -2. You also want your RB to be a good cutback runner. When you hear people talk about young RB's “trusting their speed too much” and always going for the edge, it's often in the context of outside zone. If you see a RB run toward the edge, hesitate/cut/dance, then try to get outside again, don't even bother watching, because you're almost certainly looking at a TFL. This happens when the RB's first read tells him that the edge isn't open, but he tries (late) to outrun the defense instead of cutting back.  It's not so much about "trusting his speed" as it is not trusting his reads.  This was a big problem for us on these plays last year, and Lasco's improvement in this area has been huge for us.

Inside zone is the same in a lot of ways. For the OL, the blocking assignments are basically the same, but the technique is different. Now, instead of trying to reach their defender, they're drive blocking him and trying to get a push north and south. The big difference between outside zone and inside zone comes in the RB reads. In inside zone, instead of aiming outside the TE and reading the edge defender first, he'll aim at the playside guard and read his block first:
Now for the minor differences: If you're a covered lineman, you're still going to block the guy that's covering you. The big difference is for the uncovered linemen. Since this play is aiming at the playside guard, uncovered linemen need to make the blocks that keep defenders away from that part of the line. So, the TE doesn't want to reach the DE here, he just wants to get motion on him and keep him outside.  Also, because the TE's angle is naturally good due to alignment, we don't need to double that DE, so the uncovered LT will release straight to the SLB instead of comboing up to him. The LG's block is the RB's first read. The LG is still trying to hook the DT inside, much like in outside zone. If he does so, the RB will go outside him (the leftmost arrow in the diagram). If that outside gap gets filled, the RB will go inside of the guard (the middle RB arrow in the diagram). If both of those gaps are filled for whatever reason, he'll look for the cut-back (the right-most RB arrow in the diagram). The C is uncovered but needs to keep the DT to his right away from the RB's aiming point, so he and the RG combo from the DT to the MLB. The RT works out to the WLB, and the FB kicks out the backside DE. Once again, remember the FB's backside responsibility for when we talk about the zone read.

Defending UCLA (or not, as the case may be)
Now that we know how zone plays work, we can talk about defending them, and the way that the threat of other offensive plays impacts our ability to defend zone runs. The UCLA game is a great case study in this, so I'll have plenty of video from that game to illustrate this section. Above I drew these plays out of a pro-style look. How does that adjust for UCLA's version of the spread? The most obvious difference is that they'll add the QB read to the zone play. You can do this on either inside or outside zone, but we'll focus on inside. The other difference is that they run it out of spread formations with either 3 or 4 WR's. This combination creates a few additional problems in defending the zone play. Here's the 4 WR version:
Let's look at a video of UCLA running this:
What exactly is going on here? Remember the OL assignments and RB reads that we talked about above for IZ. The playside in this video (and the diagram) is to the right. That's where the RB will aim initially. On that side, the RG is uncovered, and there isn't even a LB immediately threatening him (since the WLB is aligned outside), so he combo-blocks with the RT from the DE up to the WLB. The C blocks the DT covering him straight up. On the back-side the LG is uncovered. Because this is inside zone, and they want to open up inside running lanes, his assignment here is to help the LT get the DT blocked. They're combo-blocking from the DT up to the MLB.  If you really want to understand this play, go back and watch this combo-block specifically, because that's where it all happens.  Since this is a run to the right, it might look like they have an awful angle to keep that MLB from getting to the playside, but remember that if he flows hard to the playside that will give the RB a cutback read. The combo-block of the LT and LG doesn't need to stop the MLB from getting to the right side of the diagram, therefore. They just need to stop him from getting back to the left side of the diagram in the event that the RB decides to cutback, and they have a great angle for this. The RB's aiming point is for the RG. There's no space outside the RG on this particular play. Because the RG and RT are double teaming the DE, there is no longer a gap between them. Inside the RG, Perkins sees Trevor Kelly taking up an awful lot of space, so he cuts it back. Nickerson actually does a decent job of shooting the backside A-gap (to the left of the C) and the DT is occupying the gap outside the LG, so Perkins cuts it all the way back inside the DE for a 6 yard gain.  In the diagrams it might not look like it's easy to cut the ball back all the way across the formation, but this video shows just how fast a RB can cut it back and get downfield.

Pre-snap this really doesn't look like it should go for 6 yards. Remember: if X = the number of offensive blockers and Y = the number of defenders you need to stop the run (or, put another way, the number of gaps you need to defend), Y = X + 1. They have five blockers and we have six defenders in the box here, so according to conventional wisdom, we have enough guys. Looking back at the last diagram, from right to left, the DE should have the C-gap (if that term is unfamiliar, check out this post), the WLB should fold into the B-gap when he reads run, the DT should take the A-gap to the center's right, the MLB should take the other A-gap, the DT should take the other B-gap, and the DE should have the left C-gap. You've also got a SLB as a D-gap player/bonus run defender on outside runs to the wide side of the field.

The QB read element of this play throws conventional wisdom out the window, however. Watch the video one more time. This run is getting cut-back into the backside C-gap. Why doesn't the DE make the play? Because he's slow-playing the QB keeper (or “keeping contain,” to borrow the cliché about zone read defense). In the early years of the zone read in college, defenses tried to make this work. They tried to make that DE the C-gap player and ask him to play either the QB run or the RB on the cut-back into that gap. More recently the NFL made the same mistake, because they thought that the zone read was a gimmick, and that more disciplined and athletic defenders would be able to shut it down. Wrong. The DE can't move in two directions at once. If he stays put, he's going to be late to the cut-back. Remember when we looked at the pro-style diagrams and I told you to keep the FB's backside responsibility in mind? By reading that DE and forcing him to stay put to defend the QB, the offense uses the QB to do that FB's job. It's as if the offense has one more blocker than they actually do. So, when the QB is a run threat, we have to modify the equation above to read Y = X + 2. If you truly want to match numbers on the backside edge, you need both a cut-back defender AND a QB defender, much as you would in defending any other kind of option. This also explains why you can't just hit the QB on every play. If the DE hits the QB, he takes himself out of the play and leaves the cutback even more wide open than it was in the last video. Against UCLA, we opted to have our DE sit on the QB. Here's a video showing Jonathan Johnson (who's come on very strong over the UCLA, UO, and OSU games, by the way) on his QB responsibility, who was by far our best DE against this play:
 
Great play, right? The bummer is that he was also the guy dragging down Perkins after a 6 yard gain in the previous video. It's just exceedingly difficult to do both tasks well, even if you're a smaller, more athletic DE like Johnson.

OK, so you need an extra defender. Once again, let's look at the last diagram. Didn't we say that the SLB is there as a bonus defender? Isn't he precisely the kind of guy who could be a QB player, freeing up the end to take the RB off-tackle? This is something teams (including ours) will try to use to get a 7th defender against the run. Enter the bubble screen, our Achilles' heel from the UCLA game:
If the SLB plays hard outside here, we'll be able to defend the screen, but we won't have a QB defender. If he comes inside to be a QB defender, we won't have the numbers to defend the screen. UCLA's going to put a bubble on just about every run play they call:
Look at Hundley's eyes in that video, looking straight at the nickel back. UCLA wasn't calling all of those bubble screens and outs, they were reading them on the field. That's why they always looked like great calls.  This is also why OSU didn't run as many bubble screens as UCLA even though they were successful when called.  OSU calls the bubble screen, so they need to guess right on what the defense will be.  Teams like UCLA, Oregon, UA, and USC read the bubble screen, and can throw it whenever it's there.

One way to swing numbers in your favor while still defending the screen would be to change up the coverage. Cover-1 (Man with a FS deep) isn't going to solve your numbers problem in the box. You'll need four defenders to cover the 4 WR's man-on-man + 1 safety deep = 5 dedicated pass defenders, still leaving you with a 6-man box. Your best bet might be Cover-3, and we ran some of that. The problem is that UCLA would just use an H-back to give themselves a sixth blocker:

And the video:
Once again, watch it while thinking about assignments and reads. I've drawn the diagram flipped from the video because it's easier for me to just draw all these plays from the same hash, so in the video the playside is to the left, but in the diagram the playside is to the right (everything's mirrored). I'll talk about assignments in terms of “playside” and “backside” players so that my discussion is translatable to both the video and the diagram. The playside tackle is covered by the DE and blocks him. The playside guard has a LB threatening him, so instead of combo-blocking with the tackle he tracks straight to that LB. The C is covered by the NT and blocks him, and the backside guard and tackle combo from the DT to the LB (once again, the crucial block to watch). The RB is initially aiming for the playside guard but, once again, Kelly does a great job of occupying the A-gap. In general, we defended the playside very well in this game, so Perkins hardly ever took it. This means, of course, that he cut it back all game, and we defended that decidedly worse. Our LB gets blocked out by the backside OG combo-ing up to him, the H-back blocks our seventh man in the box, and UCLA picks up a big 4th and 3.

If you can't get a numbers advantage by pulling a safety in to help against the run (as in Cover-3), your only other coverage option is to go Cover-0 (straight man coverage with no safety deep). Then you only need to pull out four defenders to cover 4 WR's, leaving 7 in the box as run defenders. This is what Pendergast did against Oregon in 2010. Hardly any college DC's will show the fortitude needed to play their DB's in Cover-0 the whole game, though.

In another post, I wrote about our gameplan for stopping UA's zone read/screen combinations. If you check out the diagrams and video in that post, you'll see that our gameplan for that game was much more aggressive about stopping certain parts of these plays. Look at the diagrams in that post, and note how we employed stunts to shut down the playside and scraped LB's around the edge as QB players on the backside. As I mentioned in that post, this made us vulnerable to the cut-back because, in order to work outside for the QB, the LB had to leave his spot behind the DL on the backside. If the DL gets beat in that scenario, there's no one behind him to make the tackle. While Wilson hardly ever burned us on the cut-back, Perkins was happy to take it all day. My guess is that we knew he was that kind of runner going into the game, which accounts for the difference in our gameplan. Against UA we didn't think Wilson could kill us with the cut-back, so we only nominally covered it and sold out to stop the playside run and QB keeper. The result was a bad running day for the Wildcats. Against UCLA we felt that we had to cover everything.

There are a few ways that our gameplan would've worked better without any schematic changes. The first is just about how well our individual players executed their assignments.
Once you start talking X's and O's everything turns into a numbers game.  If you have more guys than they do, you'll stop the run.  If you don't, you won't.  That's obviously not how things go on the field, though, and at some point you'll just need one of your guys to beat one of their guys.  Check out the guy labelled MLB in this diagram (it won't always be the MLB with this assignment, though it often was). If that guy can attack the line of scrimmage instead of making it easy to combo up to him, he can cause a lot of damage. First, remember the rules of the combo block: Double-team the down lineman until the LB threatens, then whichever OL is in the best position to block that LB will leave the double team. If the MLB can trigger quickly and aggressively attack the LOS, he forces the OG to leave the double team early, making it easier for the DT to get penetration in the B-gap. The more penetration and disruption these two can cause, the flatter the RB's cut-back has to be to the line of scrimmage, making it take longer for the run to get donwhill and funneling it toward the DE who's been slow-playing the QB keeper.  Now all of the sudden, that extra defender doesn't seem so important. When that LB is slow to get downhill, on the other hand, our DT's are getting pushed back several yards by the double-team, giving the RB a great angle to get downhill inside of the DE. We could also ask our DT's to hold up better against the double team, but it's probably easier to ask our LB's to impact to play more at this point. On this strategy it's not about getting more bodies in the box, but messing up blocking angles so that you can win more of your match-ups.

Another way to improve your match-ups would be to stunt any combination of the DE, DT, and LB on the backside. You could, for example, pinch the DE and DT into the A and B gaps and bring the MLB around as the QB player. If you do that and your DE and DT aren't able to get across their blockers' faces, however, you're left without a LB behind them (because he's scraping outside to the QB). Still, by doing that you're testing the OL's ability to quickly pick up and execute assignment switches. In this case, that more aggressive gameplan on the backside might have paid off. We didn't go that route, though, and the result was kind of interesting. On the one hand, by failing to attack on the backside, we had our worst game of the season up to that point against the rush. On the other hand, by limiting the big play we held Perkins to his worst game in their last five, so there's both a glass-half-full and a glass-half-empty way of looking at it.

At any rate, even leaving the scheme the same, the problems I've discussed in this last section are exactly the kinds of things that improve with more experience. You don't have to speed up the LB's trigger time and instincts that much to make a major difference. If a good MLB emerges for us, it'll help our D a ton.

For discussion, see the BI thread here: 

No comments:

Post a Comment