It's a weird week for me. Florida and Oregon both run fantastically interesting offenses that have a ton of material available. (No doubt it helps that both head coaches are at the forefront of offensive innovation.) UAB ...well, we don't have that luxury with UAB. We have stats and a little bit of video, but that's about it. Basically, last year they ran a very run-heavy attack. This year, with the departure of Joe Webb, there were designs on making it into more of a pass-based offense ...and yet come the beginning of the season, we saw the same running-QB setup. On the other hand, once Bryan Ellis started getting snaps he went nuts through the air. So: they run a spread offense that varies depending on which QB plays, but since David Isabella is injured we'll just see a lot of passing. Well, that was fun. See y'all next week!
Okay, since you're sticking around, I guess I can cover in more depth something I touched on last week - the process involved in developing plays and what we've seen so far from Jim Chaney. I should warn you, though - beyond this jump there be theory.
Let's enter the realm of fantasy for a minute. Imagine you have this fantastic left guard who can block anybody in his way, knock ‘em straight down, take them out of the play and move on to the second and third levels - just as long as his first step doesn't go right whatsoever. His brother is your left tackle, who has the same problem. (It's a weird gene - very recessive.) Now, when choosing which side to run to, there would be absolutely no incentive to run to the right, right? After all, your left side of the offensive line is great as long as they go forward or to the left, but you're down two blockers if you go to your right.
Another example: you have a WR with 4.1 40 speed who can literally catch anything thrown in his direction. He also has a 60" vertical so he can win any jump ball. The only problem is he needs 6 unimpeded steps to get those powers. Given a player with that skill set, your playbook would read roughly "drop back as far as you can and chuck it at this guy, but make sure he gets at least six steps in motion before each play", and it'd be unstoppable.
Obviously it's never that simple. Players have strengths, but they're never that clearly defined - and they never have weaknesses that well-defined, either. When designing an offense, we want to make sure that we're putting the talent we have to the best use we can. Matter of fact, that's our first conclusion: our offense needs to use the players available in the best possible way; we need to maximize the impact of their strengths while minimizing the impact of their weaknesses.
Imagine that you're drawing up a playbook from scratch, and someone erased all your memory of how the game has been played - history, evolution, all gone. (This was the last thing you saw.) But you remember the rules, and with that you pull out the following pieces of information:
- You have to have seven guys on the line of scrimmage.
- You have to have someone to receive the snap from the center.
That's not too bad, is it? So you draw up something that looks like this:
And voila, you have a base formation. There's a lot of stuff you can do with this formation, but there's a lot of stuff you can't do, either (for example, good luck running any pass play to the right side of the formation - or much of any pass play in general, although it is possible. On a related note, Lane Kiffin will deploy this formation within the next couple of weeks for his XP attempts). And it's successful for a while, so you feel pretty good about that. But you're bored, you decide to get creative, and you come up with this thing.
Again, like the previous formation, there are things this does well, and there are other things this doesn't do as well. This formation - we'll call it an I (because nobody's ever used that term before, and that QB-FB-RB lineup looks kind of like an I, doesn't it?) - looks like it should be successful running the ball at first glance to either side of the line. The passing game may not be fantastic, but man, lookit all those blockers.
After using this formation for a while, your WRs revolt until you give them something to do that doesn't involve blocking (man, aren't all those blockers just great?). So you make this up:
Again, there are things this does well and there are things this doesn't do well. Of course, there aren't as many blockers (aw!), but you can throw the ball a lot. You have receivers! That's awesome. Heck, you could probably throw it to your RB if you wanted, but you'd want to move him away from your QB a bit first. And yes, you can run zone-read type plays a whole host of other stuff.
Now, if we gave you your memory back, you'd throw that first formation out as nuts. There just are too many issues with it from a practical standpoint, but those second and third formations? Those look pretty solid, so we keep them.
Why do we keep those formations? Again, implicitly we know they're good, but why do we know they're good? They're good due to versatility; I can design a run or a pass play out of both of those formations. More importantly, I can design many run plays and many pass plays out of those formations. Granted, I don't have an infinite set of possibilities - for example, I can't have four guys run streak routes in an I formation and I can't run power with a lead blocker in a 4-wide shotgun approach.
So which formations do I choose? Well, that depends on a whole host of things, not the least of which are offensive philosophy and available talent. Offensive philosophy is a whole different animal entirely (in short, what are you trying to do to be successful), and for our purposes we want to be able to do everything. In order to do everything, we need to find ways to maximize our available talent, but remember that we can't do everything from every formation. There's a balance that needs to be struck, and that balance is our second conclusion: the formations we choose need to maximize our available talent's odds of success while minimizing the things we cannot functionally perform.
Once we have talent and formations, we can design plays. But we have a crucial limiting factor opposing our play design:
What we see responding to our talent and our formations forms the third area of impact on any given play. We'll exaggerate yet again. Let's say that every time I initially line up with a fullback, the defense will stop the play I call - whatever play that is - for a five-yard loss. If I initially line up with 3 WRs, the defense will stop the play for no gain. Anything other than that is in my favor, but there aren't any guaranteed wins like we saw with the previous examples. This in turn gives me options, but I have no reason to ever set up with a fullback or 3 WRs.
Those of you who are eagle-eyed readers noted the presence of one key word in the above paragraph - initially. There's no rule preventing me from using motion to get a third WR or bringing a tight end into the backfield to be a lead blocker, but there are things I clearly cannot do. This, of course, is the goal of defense, but as burgeoning offensive geniuses, it's our goal to figure out a way around them. Will we be able to always solve the defense? There are ways, but that's yet another topic (and to be honest, one we don't really see here, at least not in obvious senses). The point isn't necessarily to always solve the defense - it's to solve the defense enough.
That sounds strange, but strictly speaking any set of talent in any formation can score a touchdown on any play. Yet we rarely see offenses actually do that, but instead they are successful enough. After all, a constant four yards a play will always produce a touchdown. We need to be able to beat the defense consistently enough and strong enough to maintain drives and score points. This is our third conclusion: our ability to call successful plays is dictated by the defenses we see, but we have an ability to figure out ways to exploit these defenses. By exploiting these defenses, we can maintain drives, score points, and win.
Now we have talent, formation, and defense. Let's add it all up.
(We're almost done with the
naval-gazing philosophy, I promise.) It's easy to play rock-paper-scissors against Bart Simpson - he always picks rock. If only calling plays were that easy. Playcalling is the synthesis of talent, formations, and defense; I pick a play that best maximizes talent and formation while minimizing the defense's ability to cope with said play. On the surface, this sounds like an offense composed entirely of constraint plays would be perfect - after all, the defense would never be able to cope with an offense of nothing but constraint plays.
What that does mean? It means that I have a set of plays that I enjoy running because they always are successful enough. For Tennessee, this is the zone running game with power. For Oregon, this was the zone read. For Florida, ...well, I'll have to get back to you on that. These plays are what I'll use most of the time, because they - you guessed it - maximize my resources.
So when do constraint plays come into play? Well, let's say I always call power to the right on 1st and 10. Eventually, anytime I get a first down, the defense would stack 10 in the box and destroy this play. So while I like this play a lot, it probably isn't the best idea to use all the time. I need to mix it up. At the highest levels, this is true for every play I choose. As a result, I need to be variant in my play choices at all points in the game. How variant is a function of what I can get away with (and there's a fine line between variant and random, which is explained further in the advanced reading section) and what my offense is capable of. After all, there's no incentive for me to call all-streaks out of a 4-wide set when I only have 3 players who can run any kind of streak route. I'm not maximizing my available talent. In addition, there's no incentive for me to be 1500% more unpredictable than I need to be; once I pass into the unpredictable realm, I've succeeded. This isn't Spinal Tap; I don't need my unpredictable amp to go to 11.
In essence, I need to be just variant enough in my playcalling that I have an advantage against the defense. That may work much of the time, but we fall into traps, run the same play 6 times, and the defense responds. Once the defense begins to expect me to do something and cheats up a guy in predicted response? That is when I use my constraint plays. So now we have our fourth and final conclusion: our ability to call plays needs to be just variant enough at all points in the game that the defense cannot successfully determine a pattern, and we need to have a subset of plays designed to exploit the defense if they do figure this out.
I just skipped all that: what the heck does this mean? English, please!
Here, let's list the conclusions again.
Talent: we need to maximize the impact of their strengths while minimizing the impact of their weaknesses.
Formations: the formations we choose need to maximize our available talent's odds of success while minimizing the things we cannot functionally perform.
Defense: our ability to call successful plays is dictated by the defenses we see, but we have an ability to figure out ways to exploit these defenses. By exploiting these defenses, we can maintain drives, score points, and win.
Playcalling: our ability to call plays needs to be just variant enough at all points in the game that the defense cannot successfully determine a pattern, and we need to have a subset of plays designed to exploit the defense if they do figure this out.
Jim Chaney, come on down! Let's list what we know based on this team so far:
- Tauren Poole is the most talented offensive player on this team.
- We haven't seen a ton of formational variance all season.
For fun, count the number of times you see a 3-WR formation out of shotgun, a straight-up I formation, or an I formation with an unbalanced line this Saturday. Compare that number to the total number of plays Tennessee runs and see what the percentages turn out to be. I'll bet they make up a good chunk of the total plays, and that's not a bad thing necessarily. For one, they put our available talent in the best position to succeed; Matt Simms may not be able to make enough reads to successfully go four-wide constantly, and blockers for Tauren Poole can only be a good thing. (Man, I love blockers. Sorry, the Men In Black mind-zapper is still wearing off.) Besides, going four-wide means we leave Luke Stocker off the field, and that's not a good thing either.
So that covers talent and formations. We can't do a ton about the defense save extensive scouting (although as we've seen in previous weeks, plays can succeed to some extent regardless of formation and defense thanks to talent), so we'll acknowledge it plays a big role and move on. Remember, our base plays should be effective against any defense, and we have seen enough in the way of constraint plays to know that they exist.
That leaves play calling. So far, Tennessee has run 102 rushing plays to 90 passes, which feels about right. On aggregate, you can't really get much in the way of trends from that. But let's pull a random subset - say, third down. What does that look like?
Run plays: 9
Pass plays: 31
Sacks / other: 2
That ...well, that's not really balanced, is it? Part of the reason we've been seeing an abysmal success rate on third down is a fairly obvious preference to go for the third down pass - even on third and short. If you'd like more of a breakdown against Florida, check it out. The balance against Oregon was decent, but UT-Martin was pass-happy again.
Pulling those plays out of the aforementioned 102 / 90 run / pass balance yields 91 rushes and 59 passes on 1st, 2nd, and 4th down. That is a little bit predictable too, isn't it? Granted, it's not so predictable that we can bank on it - if the breakdown was 110 rushes to 5 passes, then that balance would be pretty bankable. (And that would be why Georgia Tech would be incredibly dangerous if Josh Nesbitt could hit water throwing a vertical route on a raft in the Gulf of Mexico.) Also, remember that we want to maximize talent, so putting the ball in the hands of Poole and Oku may be happening a bit much, but it's an understandable slant.
This can be alleviated pretty easily, though - simply pass more on first and second down and run more on third down. I'm not advocating a full conversion, of course, but if you pass enough on first and second down that it's a legitimate concern and if you run enough on third down that it's a legitimate concern, then that's all you need. And while we're pretty close to that right now on first and second down, we're not at all on third down. Let's give running on third down more of a fair shake. After all, we're only converting on 21% of our third downs, so hey, why not. There's no better place to try this out on than a team that should be completely overmatched on defense. So let's see if it happens. Simple changes can often produce huge results.
If you'd like to read about game theory, have at it. We spent a while skirting around the edges of game theory, but if you'd like to read about it in more detail, this would be the place.
Smart Football has already written about run-pass balance and game theory, too.