Pass rate over expectation (PROE) has become a common stat for evaluating quarterbacks over the last few seasons. Check any data-centric NFL analyst; you’ll see the acronym referenced somewhere in their content or rankings. It’s littered throughout most of mine.
But, to be honest, I never quite understood the metric.
First off, I’ve seen the name change a few times. Nothing drastic. But there’s a passing rate over expected, a pass rate over expectation, a pass rate above expectation--we need some consistency.
Also, critically, the “over expected” component has been just a term or part of the calculation. I don’t know how anyone can expect anything in a game so chaotic. But, after doing some digging and building some charts, it’s become a bit clearer. So, let’s see if I can impart some knowledge on the stat and why it’s important for fantasy.
What is Pass Rate Over Expected?
Coincidentally, “Pass Rate Over Expected” means what it sounds like. It measures how much above or below expectations an offense calls a passing play. I’ll cover the expected part in a few sections, but its utility is why it’s become a larger part of our vocabulary.
Think about the other stats typically used to characterize offenses.
- Run/Pass Ratio
- Neutral Passing Rate
- Early-Down Passing Rate
- Red zone Pass Rate
While each can be useful (like YPC can be), they either lack context, have different interpretations, or require data filtering to work with a sample. And no two NFL teams are alike. Look at the number of early-down and neutral plays from this season alone.
Denver (66) called fewer than half of the plays the Chargers (133) ran in neutral situations using win probability. The Chiefs called 164 early-down plays to Tennessee’s 85.
Sampling rates alone will alter our conclusions using the stats mentioned above. But PROE incorporates every sample available.
PROE uses some of the same in-game lenses we use to parse on-field production (e.g., down, yard line, point differential, etc.) to measure passing rates. However, you don’t have to toss out any plays to calculate it. So, it’s the best of both worlds despite needing a bit of math to get us there.
How Do You Calculate Pass Rate Ove Expected? (PROE)
You can calculate Pass Rate Over Expected by subtracting a team’s “expected pass rate” from their “actual pass rate.”
Numerically, the “actual pass rate” answers a yes or no question. It’s a binary value of 1 or 0, meaning, yes, they called a pass (1) or, no, it was a run play (0).
The “expected pass rate” is the probability of a pass play based on multiple factors, from yard line and score differential to who’s the home team. But if probability and statistics aren’t your jam (they weren’t mine either), let’s think about expectations more intuitively.
Consider the extremes that would force even the Bears to call a forward pass play. Let’s say it’s 3rd and 7 with two minutes to go in the first half, and Chicago’s trailing by multiple scores. Given how far they have to go and their point deficit, we would at least expect Justin Fields to pass. Or, flip the scenario around.
Chicago’s leading by two touchdowns, and they have the ball with a minute left. Let Fields hand it off! In both scenarios, the game environment sets our expectations. And after using play-by-play data from nflfastR (and just this year’s data as a proxy), we can see how closely related our expectations are to the model inputs.
Less time, fewer attempts left to get a first down, and playing from behind all increase the expected pass rate for an offense.
Situations that you’d intuitively expect a team to pass more in any way. So, if we know the type of play that was actually run and how often to expect it, we can find PROE.
As I mentioned, PROE indicates how much over (or under) expectation each offense calls a passing play. Positive numbers mean a quarterback throws above expectation, and negative numbers point to a squad coached by a boomer. And with more samples, we can create splits within the data to inform our draft and in-season decision-making.
Types of Pass Rate Over Expected
Season PROE gives us the broadest view of a team’s intention. After averaging across hundreds of plays, much of the chaos of the regular season that holds our attention each week fades away. Short-term injuries or a bad matchup or two all get lost in the wash. And in doing so, we can compare each offense’s approach to play-calling.
But again, this is one season of data. And without added context (e.g., EPA, fantasy points per dropback, etc.), it isn’t easy to draw any other conclusions. At least ones we can readily verify.
Of course, seeing the Kansas City Chiefs (1st), Cincinnati Bengals (2nd), and Buffalo Bills (3rd) at the top makes sense. But the 49ers’ offense (23rd) produced two top-10 players at their respective positions.
Justin Fields (32nd) ended the year as the Fantasy QB6. So, unfortunately, we can’t necessarily use current PROE to predict future success.
Ten seasons of each team’s PROE not being correlated to their quarterback’s points per dropback tells us what we already know.
Passers create value in different ways. Long live the #KonamiCode. However, we can find something useful in PROE trends when factoring in personnel moves. The top-3 teams from last season help highlight my point.
More passing isn’t just a schematic shift but a philosophy shared by the head play-caller and the quarterback. And it’s not like we haven’t seen this elsewhere.
The Los Angeles Rams’ passing attack kicked into high gear after trading for Matthew Stafford despite having Sean McVay already pulling the strings on offense. Kevin O’Connell’s move to Minnesota vaulted Kirk Cousins from a fringe QB1 to his highest yardage season since 2016. So while the quarterbacks execute, understanding coaching tendencies can also help us evaluate an offense from a play-calling standpoint.
Last Four Games PROE
Looking at PROE from a seasonal view can help us draw overall conclusions but has a lesser impact once we get into the current season. And you can track each week as a single data point, but using the same three teams from earlier, see if you can spot a trend.
Everyone has a plan until they get punched in the mouth.
Week-to-week variability makes predicting what a team will do the following week impossible. However, with a set of games, we can start to form opinions on who’s above or below PROE and make comparisons.