The Best Player Does Not Play

Your coaching board¶

My seventh-grade football coach, Coach L, was a real football guy. I mean why else would anyone want to coach a bunch of undersized seventh graders in the middle of nowhere Pennsylvania? Especially during a season where we went 0–8. Also, this isn't really relevant but our games were on Tuesdays???

One of my clearest memories from those days was halftime of our seventh game. We still hadn’t scored a touchdown that season, and Coach L snapped a clipboard in half, followed by a pretty impressive Gatorade cooler launch. Such passion.

NFL coaches are notoriously passionate, too. After the 2024 season, Mike Macdonald said, “I literally haven't seen my son when the sun's been out since he's been born.” ¹ His son had been born the previous month.

But as Coach L and that winless season showed me, hours and passion alone don’t explain what makes a great coach.

So What does?

Think about that, then go ahead and adjust the board below.

McVay and Macdonald make opponents guess. Ben Johnson’s offense gives us a way to examine how a coach can help his quarterback. Shanahan shows how one play sets up the next. And fourth down tests whether a coach will take a calculated risk. Those are some of the decisions behind my coaching tiers, which I’ll reveal at the end.

Skip the analysis and see how I tier the coaches →

Making the defense guess¶

Not gonna lie, I was a little uncomfortable at that halftime when Coach L went ballistic. At the time, I thought he was a bit of a lunatic.

Making the defense uncomfortable is part of what makes Sean McVay so good. That’s a big part of why I have him near the top of my coaching tiers. Take his 2018 Rams squad. They used the same personnel package on more than 90% of their offensive snaps. The defense knew what kinds of players it would face. The Rams made the Super Bowl anyway. ²

That package was “11 personnel”: one running back, one tight end and three wide receivers. The personnel package tells you who’s on the field, but leaves plenty of room for what they might do.

McVay’s Rams kept the same personnel on back-to-back snaps 84% of the time in our 2022–2025 sample, compared with 58% league-wide. ³

McVay likes to repeat the same personnel groupings on consecutive snaps, sometimes into a third, fourth, fifth or sixth look. Inside those stretches, he runs the fourth-most play variety in the league. Same personnel, plenty of different play calls. ³

To see that variety in a more recent season, we need to look at a different package. By 2025, McVay was leaning into 13 personnel: one running back, three tight ends and one wide receiver.

Here are five examples of plays from 13 personnel. ⁴

Across the full 2025 season, including the playoffs, the Rams used 13 personnel on 30% of their offensive plays, nearly five times the median among qualifying teams. They averaged +0.195 EPA per play from it. Only Buffalo was more efficient, but the Bills used the package on 63 plays. McVay’s Rams used it on 388. ³

Making the offense guess¶

Our starter, Scuba Steve, had a hell of an arm but couldn’t remember the play call from the sideline. So I’d relay it to everyone in the huddle as the scrappy blocking wide receiver. Great guy, by the way. Shout out Scuba Steve.

When he broke his wrist in a four-wheeler accident, someone needed to step up. It made sense that the guy already calling the play would be the one taking the snaps.

Getting the call into the huddle was one job. Figuring out what the defense was about to do was another, and I was horrible at it. As soon as the ball was snapped, I felt like I’d been thrown into a blender.

Mike Macdonald’s Seattle defense put opposing quarterbacks in a similar blender in 2025. His defense shows them one picture before the snap, then changes it at the last minute when they have less time to adjust. They have to process what’s happening faster than they’re comfortable with.

One example of this deception is simulated pressure. A defender who looks ready to rush drops into coverage while someone else comes after the quarterback. Four defenders still rush, but the offensive line has to account for a different four than it expected.

In our sample, Macdonald’s simulated pressures produced a free rusher nearly twice as often as the league average. ⁵

This is a good example of how a coach’s scheme can make life easier for his defenders and a lot harder for the quarterback.

Making life easier for the quarterback¶

Speaking of life being hard for a quarterback, I took over as field general of our winless middle school team at ninety pounds soaking wet. I could barely see over the line. Coach L and I already had that to work around, and then there was the line itself. An absolute sieve.

I ended up in the dirt a lot more often than you’d want a quarterback built like me to be. In the last game of the season, I broke my ribs on a sack. That effectively ended my football career.

Anyway, I didn’t feel like my line or my coach was doing me many favors. So when a coach makes his quarterback’s job easier, it gets my attention. Macdonald’s job is to make the quarterback uncomfortable. For an offensive coach like Ben Johnson, the job is to keep him off the dirt.

In 2024, Caleb Williams was sacked on nearly one in ten dropbacks. After Ben Johnson arrived, that fell to about one in thirty, even though defenders were pressuring him just as often. ⁶

You might be thinking Caleb Williams decided he was done getting sacked (don’t blame him) and started getting the ball out quicker. But his throwaway rate barely changed, from 7.5% of attempts in 2024 to 7.2% in 2025. His average time to throw actually increased, from 2.76 seconds to 3.01. Fewer sacks didn’t come with more throwaways or a faster average release. ⁶

One change under Ben Johnson was how often the Bears used play action: from 17% to 35% of dropbacks. That’s a schematic choice that gives the defense another thing to read before it can react to the pass. ⁶

And it’s a sharp schematic change. We have good evidence that, league-wide, play-action passes generate more EPA per dropback than regular dropbacks. ⁷

I can already hear Coach L: “That’s just more reason to run the damn ball early.” He was an establish-the-run guy. Why would the defense bite on the fake otherwise? You need a little razzle before the dazzle, right?

But how much establishing does it take? In our sample, play-action passes were more productive than regular dropbacks even before the offense had called its first run. ⁷

Play sequencing¶

You may not need to establish the run for play action to work, but the order of your calls can still matter. Kyle Shanahan brings us back to what we saw with McVay: different plays from familiar personnel.

Sequencing is another reason holding personnel constant can be useful. Keep the same group on the field, and you get to see how the defense responds: who covers whom, where the help comes from, which matchups might be worth trying again. That response can inform the next call. Coach L’s establish-the-run idea had that intuition behind it: show them something, then give them something else.

Here’s how Shanahan’s offense performed on its first three consecutive snaps in 21 personnel, compared with league plays in similar situations. ⁸

To help address survivorship bias, we compare plays at the same personnel-streak depth and similar down, distance and drive position. That accounts for some of the advantage of a drive staying alive, though coaches choosing to stick with a successful package can still affect the results. ⁸

My argument is that keeping the personnel constant gives a coach a more consistent picture to learn from. He can watch the defense react, then use the next call to test what he just saw. That’s a plausible source of an advantage, and the kind of coaching I value. I’d like to dig into this more in another article, using tracking data to examine how the defense’s response changes from one play to the next.

Giving your team a chance¶

There’s scheme, there’s play calling, and there’s in-game decision making. Fourth-down discourse has been run into the ground, but it still belongs here.

As for Coach L, he would have punted on every fourth down he could. With me at quarterback, I totally get it.

The question is whether that caution gives your team the best chance to win. We use Ben Baldwin’s fourth-down model as a benchmark. It weighs going for it against kicking or punting. A coach can disagree with the recommendation, but I want a better reason than a feeling that taking the points is safer. Safety for who? You? Your job security? ⁹

Let’s see what coaches do when the model says go versus when it says kick.

When the recommendation is to kick or punt, coaches follow it 99% of the time. When it says go, about half still send out the kicking unit. These are clear recommendations: at least a 1.5-percentage-point estimated win-probability advantage. ⁹

That’s what bothers me. The disagreement is overwhelmingly in one direction, especially from guys who are supposed to be a bunch of alphas. To me, it looks a little cowardly: go for it and fail, and you have to answer for the turnover on downs. I get it. But don’t act tough.

Speaking of tough guys, let’s take a look at the fourth-down analytics wall of shame.

Reid and Saleh repeat a common argument: analytics always says go. But we just saw 8,530 clear recommendations to kick or punt. Come on, guys. If you’re scared, just say you’re scared. ⁹

Side Quest coaching tiers¶

Its pretty simple... good coaches make life easier for their players, harder for their opponent, and maximize their team’s win probability by taking calculated risks.

As for Coach L, looking back, he didn’t really have the Jimmys and Joes. Maybe he didn’t have the X’s and O’s either. But he gave enough of a damn to show up to practice every day and coach a hopeless group of undersized middle schoolers. He gave us all a place to show up. And for that, Coach L is definitely S tier.

And without further ado, here are my 2026 coaching tiers.

Enter your email to see the tiers →

Would you move anyone on the board you made at the top? Go back and make your case.

the-best-player-does-not-play.mp4_□×

Watch the full video

Play

Open in YouTube