clock menu more-arrow no yes

Filed under:

Cowboys analytics roundup: Exploring the meaning of the commonly used metrics

We talk about these stats for the Cowboys all the time, but what exactly do they mean?

Dallas Cowboys v New England Patriots Photo by Maddie Meyer/Getty Images

Starting this season, we’ve been putting out weekly pieces rounding up all of the most important analytics out there as it relates to the Cowboys to help give a better idea of where the team stands relative to the rest of the league. In case you’ve missed one of those weeks, they’re available down below:

Week 1
Week 2
Week 3
Week 4
Week 5
Week 6

Since the Cowboys didn’t play this past week, there isn’t much of an update to give on their standings in these areas. While some rankings have shifted, it’s entirely because other teams either played really well or really not-well this past weekend.

Instead, let’s take this as an opportunity to get a better understanding of what these analytics actually mean. It can be easy to point to a chart and say “See, Dak Prescott is playing great!” But the actual meaning can often get lost in all of that, so providing context can help all of us make better use of these metrics.

DVOA and EPA

Two of the more commonly used metrics when talking about team strength are DVOA and EPA. DVOA stands for Defense-adjusted Value Over Adjusted. The metric, created by Football Outsiders, is defined in simplest terms as “measur[ing] a team’s efficiency by comparing success on every single play to a league average based on situation and opponent.”

A good example to summarize this concept, one that Football Outsiders uses to explain it, is looking at two different teams who each run for three yards. It’s the same end result of rushing for three yards. But if one team runs for three yards on second and 15 while another runs for three yards and fourth and one, the efficiency between those two plays will be drastically different.

That’s the basic idea behind DVOA, which seeks to create overarching rankings of teams on how efficient they are with respect to league averages while also adjusting for factors like quality of opponent. DVOA can be broken down into offense, defense, and special teams to delineate between teams that are good because of offense and teams that are good because of defense, for example. DVOA also exists on an individual player basis, which we’ll get to in a bit.

EPA seeks similar insights. Standing for Expected Points Added, EPA measures the value of each individual play by potential points gained by the team. To quote this thorough breakdown of EPA from Inside the Pylon:

This is done by calculating the Expected Points (EP) of the down, distance, and field position situation at the start of a play and contrasting it with the situation at the end of the play.

EPA has recently become very popular in uses for things like fourth-down models and when to go for two points instead of kicking the extra point. It can also be used to measure individual players, although a criticism of the metric of late is that as NFL teams become more efficient the EPA numbers become larger in total. In other words, it would be meaningless to compare, say, Mac Jones’ EPA this year to Tom Brady’s EPA from his first Super Bowl year. To get the most insight out of EPA, it’s best to only make comparisons over the same timeframe.

Quarterback Efficiency Metrics

There is no position on the football field more important than the quarterback, and that explains why there are so many different stats out there by which one can judge a quarterback’s play.

Given how many things a quarterback is responsible for on a given play - and even the things he’s not responsible for, like running correct routes or not dropping passes or even questionable calls from officials - it’s important to look at a group of quarterback efficiency metrics as opposed to just one individual number. For the purposes of our weekly analytics roundups, we rely on the following: QBR, EPA, DVOA, DYAR, and CPOE.

QBR stands for Quarterback Rating and was created by ESPN back in 2011. While ESPN has never released the actual formula of how they create the statistic, QBR has quickly become one of the better methods to evaluate quarterback play. To use ESPN’s own words on the stat:

ESPN’s Total Quarterback Rating (Total QBR), which was released in 2011, has never claimed to be perfect, but unlike other measures of quarterback performance, it incorporates all of a quarterback’s contributions to winning, including how he impacts the game on passes, rushes, turnovers and penalties. Also, since QBR is built from the play level, it accounts for a team’s level of success or failure on every play to provide the proper context and then allocates credit to the quarterback and his teammate to produce a clearer measure of quarterback efficiency.

In short, QBR provides context to a quarterback’s play. It does this using EPA and adjusting for other factors like quality of opponent. Think of it as EPA revealing a quarterback’s average value added on a per play basis, while QBR is how well the quarterback played on an overall level.

These two metrics have a similar relationship to the duo of DVOA and DYAR. Standing for Defense-adjusted Yards Above Replacement, DYAR is looking at an individual player’s performance relative to the level an average replacement player would perform at. When looking at quarterback play, DYAR tells us about a quarterback’s total value, while individual DVOA tells us about a quarterback’s value per play.

CPOE is another valuable metric, and it stands for Completion Percentage Over Expectation. This one comes from the NFL’s Next Gen Stats hub. Essentially, they calculate the probability of a pass being completed based on factors such as receiver separation, where the receiver is on the field, how much pressure the quarterback was under when throwing, and more. This completion probability creates an expected completion percentage, and CPOE is the difference between the expected completion percentage and the actual completion percentage.

In short, CPOE is a good way to measure a quarterback’s actual accuracy, adjusting for things like deeper throws that bring a lower probability of completion or if a quarterback throws an incompletion because they were hit as they threw. CPOE also (generally) does a good job of isolating a quarterback’s accuracy from scheme issues, although this isn’t perfect. CPOE has yet to adjust for wide receiver drops, however, which is the biggest weakness of this metric. It’s still great for gauging how well someone is actually throwing the ball relative to the throws they’re attempting.

Pass Protection and Pass Rush Metrics

Offensive line play can be really hard to effectively measure, especially in run blocking. Football Outsiders has developed what they call Adjusted Line Yards, which attempts to separate line play from running back performance. It’s an impossible feat, but this metric comes about as close to doing it as possible.

Without getting too into the weeds of the math, adjusted line yards gives the offensive line all the blame for negative run plays and all of the credit for runs of four yards or shorter, splitting credit with the runner on gains between five and ten yards, and giving all of the credit to the running back on anything beyond that. It’s more complicated than that, and there are a lot of adjustments to it, but this yields a solid representation of how many rushing yards are actually a result of run blocking.

We also have Run Block Win Rate, a metric from ESPN that measures the rate at which an offensive lineman holds his block on a running play for 2.5 seconds or more. It isn’t always indicative of good rushing attacks, but this metric provides a glimpse of how consistent run blocking units are. Similarly, the Pass Block Win Rate measures how often linemen hold their blocks on pass plays for 2.5 seconds or more. The weakness here is that, when grading an offensive line as a whole, their overall win rate can be tanked by one bad lineman. Individual win rates are also tracked, but ESPN only releases the top ten individual win rates on a weekly basis.

Two other good measures of offensive line play is looking at pressures allowed and a team’s Adjusted Sack Rate. Pressures are most commonly recorded by Pro Football Focus and TruMedia, both of which require subscriptions, but the adjusted sack rate comes from Football Outsiders. This measures sacks and intentional grounding penalties on a per pass attempt basis, with adjustments for variables like down, distance, and opponent. This helps determine how often the quarterback actually gets taken down, although it doesn’t delineate between quality of pass protection and a quarterback’s ability to not take a sack.

On the other end of the spectrum, defenses can look at both Pass Rush Win Rate and Run Stop Win Rate. Very simply put, these happen when a defender beats their blocker in less than 2.5 seconds on a pass or run play. This helps to give a better understanding of how often defenders beat the blocker, although it becomes irrelevant when facing a quarterback with a quick trigger like, say, Tom Brady. Still, these win rate metrics do a good job of redirecting the focus from sacks, which can be an unreliable stat from week to week and especially season to season.

Why Passer Rating Matters for Defensive Backs but Not Quarterbacks

The disclaimer here is that we still don’t have a good way of quantifying coverage play, and probably never will. Without knowing all the intricacies of each defensive play call, it’s impossible to truly know which defender was responsible for which pass-catcher.

For the time being, though, Passer Rating Allowed is a sufficient way of getting an idea of how good a defender is in coverage. For a more specific breakdown of the passer rating stat itself, check out this summary. Passer rating was initially derived to judge a quarterback’s play, much like QBR. The issue is that passer rating doesn’t have all of the context QBR does.

For example, when Lamar Jackson has games where he only throws for 150 yards or so but also runs for 110 yards and two touchdowns, he finishes with a poor passer rating but a very good QBR. That’s because passer rating ignores rushing from a quarterback, thus making it a highly irrelevant stat in an era where dual threat quarterbacks are becoming more common.

However, judging a defender’s skills in pass coverage really only involves passing plays. Thus, passer rating redeems itself somewhat in this category. Passer rating allowed is simply just a measure of a quarterback’s passer rating when targeting a specific defender, thereby giving us an idea of how well the defender performs when being thrown at. Again, the caveat here is that we can’t ever truly know who the responsible defender is on a given play, but this is as close as we can get right now to meaningful analysis of pass coverage skills.