clock menu more-arrow no yes mobile

Filed under:

The five elements of the optimal college football offense

New, 8 comments

Synthesizing a bunch of concepts we discuss a lot but haven’t necessarily put together.

NCAA Football: Rose Bowl-Oklahoma vs Georgia Kirby Lee-USA TODAY Sports

What makes the optimal college football offense in today’s game? That might seem like an impossible question to answer (and it probably is) but there is compelling evidence that there are certain core principles that, individually, create useful advantages for modern offenses.

Perhaps more importantly, combining these principles create synergies that actually lead to more significant advantages, both at the tactical and strategic level. None of the core principles are going to be new to anybody who is familiar with offensive football, but the view of how they fit together and stack up into new advantages might be.

Let me be clear that this isn’t meant to be a referendum on football aesthetics. These conclusions are based on empirical data, logic, and observation, not on stylistic preferences. Personally, I find the diversity in approaches to offense one of the most enjoyable things about college football (and football at lower levels), and I have always found off-beat or novel offenses to be interesting and fun to watch.

For instance, I love watching a pure wishbone or flexbone triple option offense when it is being run well; Air Force was always one of my favorite teams when I was in high school and college because of that offense (and because of Dee Dowis, the greatest pure option QB I ever saw play).

The Five Core Principles

1. Quarterback as a Run Threat

The basic math here is simple. In a traditional offense in which the QB is never (or very seldom) intentionally utilized as a runner, the defense has a built-in numerical advantage. For example, when a QB takes a snap from under center and hands off to a running back, the defense has 11 players available who could make the tackle, while the offense only has nine with which to block.

This can be mitigated by the QB executing something like a bootleg fake to hold a backside defender’s attention briefly, but once that defender identifies that it is not a pass play, he is free to run to the ball.

Establishing the QB as a running threat forces the defense to account for the QB in the running game, whether he keeps the ball or not. That shifts the defense’s advantage from two unblocked men to one. The QB doesn’t have to take on the rushing workload that Tim Tebow did, and he doesn’t have to be the open-field threat that Lamar Jackson was to significantly influence the defense.

Simply demonstrating a willingness to have the QB carry the ball and establishing that those carries can result in large enough gains to maintain efficiency (four or five yards per carry would comfortably do it) ought to force the defense to account for the QB as a runner on every play. If it doesn’t, the offense can run the QB often until the defense adjusts (and the QB can slide, as necessary, to avoid taking a beating).

This might sound like a great deal for the offense, but wait, there is more! When the QB reads a defender, as on a zone-read play, the QB actually can manage to occupy two defenders, which evens out the original two-man advantage for the defense.

While I haven’t found or derived any analytical evidence that illustrates the benefit of QB use in the run game at the college level, it does exist at the NFL level. Josh Hermsmeyer (@friscojosh on Twitter), who maintains and writes for several online publications, has done exhaustive studies of the success rates of various play types versus different defensive fronts in the NFL.

His findings show that over the past two NFL seasons, zone read type plays have been more effective than any other type of run play in terms of success rate (which he defines as the ratio of player generating positive vs. negative Expected Points Added) when facing more than five men in the box. This is despite NFL defenses stopping the run incredibly well.

The general perception is that the zone read was “figured out” by NFL defenses a few years ago, but it still works better than all other run types, on average, at the NFL level.

NCAA Football: Peach Bowl-Auburn vs Central Florida
Not including sacks, UCF’s McKenzie Milton rushed just 97 times last season but gained 683 yards and punished distracted defenses.
Dale Zanine-USA TODAY Sports

2. Optionality

The benefit of using QBs reads in the run game segues nicely into the more general principle of incorporating “optionality” into offensive play and scheme design. In financial theory and in the real world of business, there is a truism that all options have value. This is also true in football.

Reading a defender in the run game, incorporating option routes in the passing game, utilizing pre-snap hot reads based on tags, utilizing RPOs, or simply allowing the QB to change the play at the line of scrimmage based on the defensive look are all examples of optionality on offense. Each of these differ in character, complexity and execution, but they share one common theme: they are designed to punish a defense for its choices by attacking it at a vulnerable spot.

The basic value of the use of options in the running game is well established and has already been covered in this piece, but we’ll return to the value of all types of optionality in part two, when we look at specific synergies.

3. Simplicity

I could get long-winded about the elegance of simplicity, cite Occam’s Razor and a bunch of other wise quotations, but that wouldn’t be very simple. Simplicity is good because it enables players to learn schemes faster and more easily. Research on mastery shows that the less somebody is required to expend on mental processing, the better they can do whatever basic task is in front of them. Players who are thinking about what they are supposed to be doing generally execute their assignments slower and less well than those who are not bogged down in processing too much information. They also make more mistakes.

This is an area where there is a significant difference from the NFL, where the players are full-time professionals, and college (or high school), where practice time is significantly limited and players have to balance academic pursuits with studying the playbook, watching film, etc.

NFL playbooks are thick, dense tomes containing huge amounts of complex verbiage and other esoterica. If a college football playbook looks like the Houston phone book (remember phone books?), the offense is likely hurting the program in one way or another.

When college football schemes are so complex that younger players take a long time to pick them up, that hurts the effective depth of the team. It can also have an opportunity cost. Consider a team that lands an elite talent at wide receiver or running back but is unable to effectively utilize him in his freshman year because the schematic complexity is too demanding. If the player leaves after three seasons for the NFL, the program has lost up to a third of his potential contribution.

Finally, and most importantly, excessively complex offenses impose huge burdens on the quarterback position. The need to prepare a single player as thoroughly as possible to shoulder the burden of executing a very complex system forces practice repetitions and coaching attention to be heavily focused on the starting QB.

That means the backup and other QBs on the depth chart receive relatively little attention. Therefore, the more complex an offense is, the more fragile it is and the costlier an injury could be, because of what it would demand of a comparatively unprepared player. The same opportunity cost concept applies to quarterbacks over their career as it does to other position players as well.

Goodyear Cotton Bowl Classic - Michigan State v Baylor
The Baylor offenses of the mid-2010s were powered by simplicity, spacing, and tempo.
Photo by Ronald Martinez/Getty Images

4. Space

The idea of stretching the defense horizontally and vertically is foundational to offensive scheme design and as old as the hills. Eleven defenders can only cover so much ground, after all. And there have always been only three ways to bypass defenders with the ball: go around, over or through them. But the evolution of football tactics has made the “spread” offenses of yesteryear, like the flexbone, seem extremely compact by today’s standards.

At the risk of triggering old-school football purists, there is ample evidence, both logical and statistical, that the more spread out horizontally the defense is pre-snap, the more efficient the offense is likely to be. What might be surprising, is that this is more due to the impact of horizontal space on the running game than the passing game.

The traditionalist view is that putting three or four wide receivers on the field and lining them up in some sort of spread formation might be good for the passing game but is inherently detrimental to a power running game. There are both convincing logical reasons and empirical evidence that this is the opposite of how it really works for the running game.

First, from the simple logical side, let us consider the example of an offense lined up with 21 personnel (two running backs, one tight end, and two receivers), with the TE in-line, the backs in an I-formation and the receivers split out wide on either side. This is a traditional and generic formation, still often seen on Saturdays in the fall, at least in some places (ahem, Madison).

Florida Atlantic v Wisconsin
Wisconsin still loves itself some 21 personnel.
Photo by Dylan Buell/Getty Images

Against that personnel grouping and formation, the defense is likely to line up with a single high safety, both corners out wide to cover the WR, and the other safety lurking somewhere near the edge of the box. Depending on how the defenders with alley responsibility are aligned, that means there are a minimum of 7 and possibly 8 defenders “in the box” to defend the run. This creates several real issues.

The first issue is that it is just a mess to block properly, almost regardless of the type of play called. Anything attacking A, B, or C gaps is going to be very crowded at the point of attack, and depending on the defensive alignment, there is always a possibility that the blocking assignments could get confused because the permutations of assignments get complicated quickly. But let’s assume that the play is a Power (see example below and note that this is just a small section for a single play) and that none of the blockers are confused over their assignment.

In order for the play to really work, seven blocks have to be executed properly and at least three have to be executed essentially more or less perfectly

Sample Power O Blocking Schemes, From Kansas State 2002 Playbook

So, let’s run a little thought experiment. Keep in mind this isn’t meant to be exact, it is a simplified example, but it should illustrate a key point quite clearly. To keep the math simple, let’s assume seven blocks need to be executed right and that if the critical blocks are executed at all, they are executed perfectly.

If we assign the blockers each an 90% chance of success on their blocks, the joint probability of all seven being executed properly is just under 48% (0.907 = 0.478).

Now let’s consider an inside zone run out of 10 personnel (one running back, and four wide receivers) with the WRs split wide. This personnel grouping and formation forces the defense to either take a risk in the passing game by lining up with a single high safety or to concede some toughness against the run by lining up with a two-high look. If it lines up with two high safeties, it can have five men in the box (see simple example below). That means if we assign the same 90% probability on a per block basis, the joint probability of a properly blocked play is 59% (0.895 = 0.59).

Inside Zone from 2x2 Spread, 10 Personnel

That is a quite significant difference (a nearly 25% increase in blocking success) in favor of the spread personnel and formation (and less loaded box), assuming the same per-block probability. Clever readers might suggest that if the defense wants to overwhelm the offense’s ability to block such a play it could just drop the strong safety down into the box, giving the defense 6 players against the offense’s five linemen. While that is certainly true, it ignores tactics the offense could employ to neutralize that advantage (see earlier discussion of QB use in the run game and wait for further discussion of this topic when we get to synergies).

The point here is simply that spacing helps to increase the odds of success by reducing the opportunities for something to go wrong.

The best part of this little logic exercise is that it is borne out in the numbers. Again, I will refer to Hermsmeyer’s work with the data from the NFL’s past two seasons. His analysis found that the number of men in the box was a far more significant factor in terms of rushing success than the number of extra blocking surfaces that the offense could utilize. In fact, the men in the box completely swamped the number of extra blockers available as a predictor of play success.

I would sum the preceding discussion up in a very simple way: “Space never splits a double team to make a tackle”.

I don’t want to neglect the issue of vertical space either. Forcing the defense to defend vertically is somewhat a function of formation and personnel but is also heavily influenced by play-calling. Throwing downfield enough to force the defense to respect it as a threat is essential to seeing those two high safety looks that make rushing from spread formations and personnel so attractive. Like the QB running threat, this doesn’t mean it has to happen frequently, but it needs to happen often enough to be respected as a real threat.

5. Pace

The final principle, pace, probably deserves another trigger warning for traditionalist football fans (and Bret Bielema). To be clear, when I refer to pace, I am not simply referring to constantly running a warp speed hurry-up no huddle (HUNH) all the time, though some programs certainly have embraced that with mixed results. Instead, I am suggesting that playing with enhanced pace can have tactical advantages at times for all teams and can have strategic advantages for some programs. In addition, practicing with pace has strategic advantage for all teams.

I’ll deal with pace first as a tactical weapon, because it is probably the simplest and least controversial element here. All teams should be ready, willing, and able to shift to HUNH, even outside of the traditional two-minute drill. Varying pace of play allows the offense to dictate to the defense, by preventing substitutions when the defense is caught wrong-footed with personnel and also by following up especially successful plays with another play quickly, before the defense can react and re-group.

Also, if the defense is unsure how quickly the offense will come to the line of scrimmage and snap the ball, it handcuffs the defensive coordinator and players on the field in terms of how many adjustments they can make.

The bigger-picture strategic advantage for playing with at least above average tempo for certain teams is a topic that has received little attention but should be intuitively obvious upon examination. Bill Connelly mentioned it a couple weeks back on Podcast Ain’t Played Nobody when he was talking about Alabama, which used to play quite slowly but has quickened the pace in recent seasons.

Simply put, if a team has a meaningful per-play advantage in net efficiency, it should seek to lengthen games (within reason) in order to reduce the variance of game outcomes.

NCAA Football: Alabama at Mississippi State
Even Alabama can adopt some tempo when it needs to.
Matt Bush-USA TODAY Sports

This is a practical application of a fairly simple statistical concept that is too dull to go into great detail about here but can be properly understood by analogy. In football, a team with this advantage is like the house in Las Vegas. Over time, the casinos end up winning because the games are tilted in their favor. A team with a marginal advantage due to scheme or talent is like a blackjack table, while a team like Alabama is like the sucker bets on the craps table.

It might be simpler to just ask how many games Alabama would have lost over the last decade if football games were five quarters instead of four. And the answer is fewer than they did.

The converse is also true. A team with a significant disadvantage on a per play basis should try to shorten the game while also trying to increase variance in any other way it can (which would probably include being extremely aggressive on fourth down, blitzing a lot on defense, running trick plays, and utilizing onside kicks).

Finally, embracing practicing with enhanced pace is a simple way to speed development of players and improve execution. Assuming the teaching element doesn’t suffer (or suffer too much), increasing the number of reps is an all-around win for the program.

The two most common objections to the use of pace and the HUNH are that “this isn’t real football” and that operating high pace offense hurts a team’s own defense so much that it isn’t worth it.

The first objection is a stylistic and aesthetic one, and those considerations aren’t part of this discussion. The fact an individual fan or coach may not like something doesn’t mean that it isn’t effective.

The second one is a piece of conventional wisdom that simply has no empirical basis. One might think it is completely obvious that a team’s defense is not going to execute as well and will be vulnerable to breakdowns in an extremely long game. And while could be true, it ignores the fact that the offense also tires, potentially leading to the same issues.

As a result, we see a lot more variance in play outcomes near the end of games with a lot of snaps, but that variance does not unilaterally benefit the offense (see graphics below, courtesy of Bill Connelly).

If this isn’t sufficiently convincing, have a look at a pretty exhaustive study (with a very large sample set) performed by Ben Baldwin on NFL data that examines the relationship between defensive rest and offensive success (TL;DR: there isn’t one).