Dallas Cowboys: What EPA tells us about the offense’s strengths in ’18

Dak Prescott #4 of the Dallas Cowboys (Photo by Ronald Martinez/Getty Images)
Dak Prescott #4 of the Dallas Cowboys (Photo by Ronald Martinez/Getty Images) /

If scoring points is the offense’s primary objective, EPA is essential in determining what worked for the Dallas Cowboys in 2018 and what didn’t.

As the Dallas Cowboys remake their offense under new coordinator Kellen Moore, it’s important to understand what worked and what did not. Luckily for us, there’s an advanced statistic out there that’s perfect at determining such things and it’s called EPA.

What if I told you a 1-yard gain was a better play than a 40-yard gain? Or that a pass for 10-yards was more valuable than a touchdown run? Or that going for it on 4th and goal from inside the 2-yard line is typically beneficial to the Dallas Cowboys in both success AND failure?

You may be inclined to call me mad (or worse). That’s because all of these statements go against conventional wisdom and without any context to them, they seem like pretty absurd statements to make.

That’s why advanced stats like EPA are important. EPA looks at the context of the game: What was the down and distance? Where on the field did it occur? How much time was on the clock? These are all questions that must be considered when measuring the worthiness of a play.

In situations where the win probability is between 25% and 75%, the Dallas Cowboys actually ranked as the No. 8 passing attack rated by EPA/dropback.

EPA takes it a step further and runs the numbers on the statistical likelihood of points gained or lost by a play (all within context). Since the whole objective is to score points (unless you’re running out the clock late in the fourth) the success/failure of the play must be based on increasing/decreasing the offense’s chances of scoring plays.

So what is EPA?

EPA, or Expected Points Added, grades the success (or failure) of a play based on whether or not that play increased the offense’s net point expectancy. In other words, did the play increase the offense’s likelihood of scoring (and if so, by how much) or decrease it’s odds of scoring?

It’s an advanced statistic that uses historical data and actual context (field position, time on the clock, down and distance, etc…) to grade plays.

At the end of each play the statisticians add the necessary context of the situation and attribute the appropriate EP (expected points). After the next play they do the same. The difference between the two gives us our EPA. And from there we can determine whether the play was successful or not.

It’s important to keep in mind, just because positive yardage was gained, does not mean the offense increased its expected points. Three yards and cloud of dust on 1st-and-10 is deemed unsuccessful because the team was statistically better off at 1st-and-10 than they were at 2nd-and-7.

Context Matters

With counting stats, a one-yard gain is always less than a 50-yard gain in perceived value. But what if that one-yard gain converted a fourth down and moved the team to goal-to-go? And what if that 50-yard gain was a screen pass stopped short of the goal line as time expired?  Suddenly the one-yard gain is extremely valuable whereas the 50-yard gain is basically worthless. All it takes is a little context.

ESPN gives this example:

"“From your own 20-yard line, an 8-yard gain on third-and 10 is worth about minus-0.2 EPA because you don’t get a first down; the same 8-yards on third-and-seven is worth 1.4 EPA for converting a long third down and keeping the drive alive. EPA knows that not all yards are created equal.”"

Using cumulative stats, we could see a player gained 100-yards on 22 carries and a TD and call it a successful day. But what if 15 of those carries were deemed unsuccessful (negative EPA that actually decreased his team’s chances of scoring) and more often than not, he was the least helpful weapons in the quest to score points?

What worked for the Dallas Cowboys offense in 2018?

I think we can all agree scoring points is a great way to win a football game. That’s why EPA is great. EPA essentially tells us whether a play helped or hurt our team’s chances of scoring points and by how much. It’s making the points the goal rather than getting caught in the weeds of a perceived gain.

Dak Prescott only ran the ball 75 times last season so it may seem strange to see him as one of the best producers of EPA, but harken back to those runs and you’ll likely remember  those were pretty key times he tucked the ball and ran.

Cole Beasley produced the highest EPA per play last season. He may have only produced 3 touchdowns last season but EPA is about increasing scoring likelihood – not getting fantasy football points.

More from Dallas Cowboys

For example: If Zeke dives in for a 1-yard TD run he achieves less EPA than Beasley who just converted a third down and moved the team into field goal range. That’s because points are virtually expected (a high statistical likelihood) on the 1-yard line and the big EPA plays on that drive likely happened to get the team to that advantageous spot on the 1-yard line.

So the player that extended the drive by moving the team from the 40 yard line to the 31 yard line added to the likelihood of a field goal and kept the hope alive for more. Based on added points, it’s that catch at the 31 that “scored” more.

Passing worked well for the Dallas Cowboys

Forget the cumulative numbers that say Dallas was the 22nd ranked passing offense in the NFL. The Dallas Cowboys were actually a pretty darn effective in the ballistics department.

Now at first glance, it didn’t look great for the passing game. Even EPA looked pedestrian measured against the NFL’s elite. Dak’s passing attack was only 16th in the NFL in EPA per play for the season.

But if you take away the blowouts and only rate NFL teams in winnable games (games with a win probability between a 25% and 75%), you’ll see Dallas moves into the top-10 in the NFL

It’s clear one of the major flaws of the Scott Linehan-led Cowboys offense was it didn’t lean on the passing attack enough. That’s because the Dallas Cowboys actually ranked as the No. 8 passing attack rated by EPA/dropback. Just imagine how deadly they could have been if the sacks weren’t holding those numbers back (Dallas was second in the NFL in sacks against).

What does this say about Dak and Zeke?

In the coming days, weeks, and months we’ll dive more deeply into this but for now it’s important to understand what EPA is and why it might be the most important advanced statistic for offense there is. It’s not as empty and misleading as the cumulative/counting stats because it involves context. It’s especially valuable to study play-calling because it lets us know what’s successful and what is not successful.

What it doesn’t tell us is whether or not Ezekiel Elliott, Cole Beasley, Dak Prescott, Amari Cooper, etc… are good as individual players. There is a ton more that goes into player evaluation.

It’s important we all stay focused on the objective of the offense: scoring points. That’s why EPA is such a valuable tool because it keeps the points as the goal and primarily grades play-calls and philosophy rather than individual execution.

Dallas Cowboys Mock Draft: Combine Edition. dark. Next

In coming days, we’ll use EPA to help us draw more conclusions about the Dallas Cowboys offense and what it must do to improve. We’ll also roll it over into how we value and prioritize individual players because right now, questions are surrounding just about every corner of this roster.

  • Published on 03/03/2019 at 13:00 PM
  • Last updated at 03/03/2019 at 12:13 PM