What’s an NFL Win Worth These Days?

What does it take to win in the NFL in terms of advanced analytics?

I used to play center on the offensive line of my elementary school’s football team, and while I didn’t last in football long (I’m sized more like Darren Sproles than Matt Birk) one game in particular sticks out in my memory.

We were down at the half against the toughest team in the league. Coach gave us the game plan for the second half and went back to the sideline. Before the first snap, I said to the other players in the huddle, “Guys, remember: it’s not about whether we win or lose today. Heck, it’s not even how we play the game. Always remember: this is about looking good.” We all laughed and a weight was lifted off our shoulders as we lined up. I smiled as I looked the nose tackle right in the eye and snapped the ball.

I barely had time to blink before I got thrown backward into my quarterback. We were annihilated; no underdog victory story that day.

At the peewee level, there’s little on the line when you lose a football game. In the pros, though, there are billions of dollars riding on teams’ successes, fans’ rooting interest is much higher, and the caliber of play is much better from any NFL center than my sack-allowing turnstile ways.

It’s hard to say how much an NFL win is "worth," but that's what we're going to try to figure out. What is the value of an NFL win in terms of advanced analytics?

Playing Big

numberFire has developed a metric -- known as Net Expected Points (NEP) -- that helps us understand the value each player and each play bring to the game of football.

NEP describes the contribution a play (or player) makes to their team’s chances of scoring. By adding down-and-distance value to the box score, we can see just how much each play and each team as a whole influence the outcome of games. For more info on NEP, check out our glossary.

By using NEP, we can break down the game of football into piece-by-piece looks at the worth of each individual play on the gridiron, or we can zoom all the way out and look at the big trends in the league and how they influence the sport as a whole.

We’re going to look at the latter here and see how NEP relates to that most basic of NFL goals: the win.

NEP is broken down into a few component parts: Offensive NEP, Defensive NEP, and Special Teams NEP. All clearly relate to the phase of football that they accrue value to or subtract it from. Obviously, bad teams can eke out a win here or there, and the “human element” and luck play some role in determining victory or defeat; still, the large majority of the time, the "better" team will win.

But better in what regard?

The first thing we need to help us figure out the value of a win is to determine what component of NEP relates the closest to winning.

To do that, I ran a correlation on a few different variables comparing NEP value and win percentage. What this tells us is how closely related two variables are, and the closer they are to 1 or -1, the stronger that relationship is (think a straight line going diagonally). As the correlation gets closer to 0, the relationship between the two is less strong (think a scattered mess of points all over).

Let's start with the grand poobah of value: Total NEP.

Correlation With Win Percentage Correlation
Total NEP 0.89

Clearly, the value of Total NEP (the combined value earned on offense, defense, and special teams) relates closely to teams’ win percentages.

This type of correlation is ridiculously strong, and while not exactly descriptive or a one-to-one causality, it’s a really good bet that a team who produces on offense and limits opponents on defense will win. In fact, each point of Total NEP since 2000 has been worth about 2.68 percent of a win.

So, our model is really good at predicting a win in the NFL by Total NEP, but what can we do with that?

In the interest of illustrating how the components of NEP also relate to winning, I split apart Offensive and Defensive NEP, as well as marking how closely Offensive Passing NEP and Offensive Rushing NEP correlate to winning.

How do they stack up (keeping in mind that a good Defensive NEP score is negative rather than positive)?

vs. Win Percentage Correlation
Offensive NEP 0.66
Defensive NEP -0.52
Off. Passing NEP 0.65
Off. Rushing NEP 0.39

We hear all the time that the NFL is an offensive league, and this helps to show just how true that is.

A 0.66 correlation for Offensive NEP isn't nearly as ironclad as Total NEP's correlation with wins, but that makes sense: we see plenty of games in which a great offense is undone by an abhorrently bad defensive showing (sorry, New Orleans Saints fans).

It's not as if there's no relationship to a good defense and winning, but offense is simply a stronger indicator of this. For all of you fans of the "other side of the ball," this means that we should be rooting for the diva receivers and hotshot quarterbacks when our NFL teams are drafting.

I specifically mention the skill players involved in the passing game because we can see that Offensive Passing NEP is significantly better as an indicator than Offensive Rushing NEP.

Is the league pass-happy or just happy to be passing? All the time, we talk about why the air game is just more efficient than the ground one, and this data helps to show that clearly.

If you wonder why our editor-in-chief JJ Zachariason could ever think that Rookie of the Year candidate Ezekiel Elliott was a huge reach in 2016's draft, it's this: running the ball relates so much less strongly to winning than passing that when you remove it from Offensive NEP, Offensive Passing NEP's strength is nearly unchanged.

So, what's the value of a win? That, and how to get to a win, will always be tough to nail down exactly, but we can see that they often come due to offense and prowess in the passing game.

As long as the opposing lineman doesn't toss you too far.