The Throws You’re Not Supposed to Complete

The Lie Inside the Box Score

Turn on the tape. A quarterback rips a throw into a window that barely exists. A receiver contorts his body, survives contact, and somehow comes down with it. The broadcast calls it “perfect.” The stat sheet calls it “complete.”

And inside every football building in America, the same argument starts: who gets the credit, and do you think he could do it again?

For decades, we’ve answered that question with opinions dressed up as conviction.

• “Elite ball placement!”
• “Strong hands!”
• “Trust between QB and WR!”

All true, yet all incomplete, because the one thing we’ve never been able to quantify until now is this:

“How difficult was that play, really?”

Completion percentage, one of the game’s most cited metrics, sits at the center of that problem. A 3-yard checkdown counts the same as an 18-yard hole shot versus Cover 2. Both go in the stat book as 1-for-1, but they’re not the same. Not even close. 

Enter CPOE: Football’s Reality Check

At SkillCorner, we built a model that answers the question coaches often ask when the clicker stops:

“How often should that ball be completed?”

Completion Probability Over Expected (CPOE) has gained attention in football discussions as a cutting-edge metric for evaluating quarterback and receiver performance. At a high level, CPOE compares what did happen to what should have happened. For example:

• How do you quantify an impressive throw into tight coverage?
• How do you reward an incredible catch from a receiver?
• How do you penalize QBs for poor ball placement? 
• How do you punish WRs for dropped passes?
  ‍

At SkillCorner, we’ve developed an advanced prediction model to determine the probability that each pass will be completed. Using player tracking data, each pass is evaluated based on the environment surrounding it. Measurements such as separation between intended target and primary defender(s), the amount of pressure being applied to the quarterback while attempting the pass, etc. The goal is to transcend traditional box score accounting by bringing nuance to the quantitative evaluation

Simply defined, CPOE = Actual Completion % - Expected Completion %. Where the actual completion percentage is 100% for a completed pass, or 0% for an incomplete pass and the expected completion percentage comes from SkillCorner’s advanced predictive model.

When a quarterback completes a difficult pass, they are rewarded with a high CPOE value. When an easy pass falls incomplete…well we’ll look at some different examples next.

CPOE In Action

Let’s take a look at some real in-game examples.

It’s 4th and 5 with 11:28 left in the 4th Quarter of the 2026 CFB National Championship Game. Fernando Mendoza rips a laser from his own left hash to find Charlie Becker on a back shoulder throw, 20 yards downfield, and on the opposite tick marks along the sideline. Based on SkillCorner’s tracking data, that pass had a 29.8% chance of being completed. Not a high expected pass completion probability given that Becker was covered closely by a Miami defender and the ball had to travel a long distance. 

On this play, CPOE = 100% - 29.8% = +70.2%. We can then reward Mendoza and Becker for this positive CPOE value of +70.2%.

Now of course with high risk comes high reward, so what happens when a low risk pass falls incomplete?

‍

In this example Mendoza throws a short hitch route with an expected completion percentage of 87.5% and that pass falls incomplete. The CPOE value on this play can be calculated as CPOE = 0% - 87.5% = -87.5%. 

As one can see, not all completions and incompletions are created equal. A quarterback can put the ball right in the perfect spot, and the intended target somehow fails to come up with the reception. Or, the quarterback can miss a wide open receiver and also have an incompletion. The punishment and reward go both ways. 

Over Time, The Truth Emerges

Where CPOE becomes especially powerful is when you aggregate it over time. Anyone can make one great throw.

What separates prospects, starters, and championship quarterbacks is what happens over 300+ attempts when nothing is perfect.

Fernando Mendoza, the projected 1st overall pick in the 2026 NFL Draft, finished his 2025-2026 Heisman campaign with a CPOE value of +14.2% on all passing attempts. This means that he is completing 14.2% more passes than expected. To put that into context, Mendoza attempted 379 passing attempts during his 2025-2026 season. When compared against the expectation for all of his specific attempts, he completed about 54 more passes than expected. That’s not just the difference in a game, that’s creating a difference over an entire season. 

When you zoom in further, you can visualize CPOE in particular situations and for specific throws. 

• +20.5% CPOE on tight-window throws
• +9.6% CPOE when under high pressure
  ‍

Those are the situations that translate. Those are the throws teams care about most. One could even filter for 3rd/4th downs, Red Zone, Backed Up, 2 Minute, and more!

Where Current Evaluations Fall Short 

Here’s the uncomfortable truth; a lot of quarterbacks can look accurate when looking at completion percentage and this is where traditional evaluations often fall short. 

Some quarterbacks produce efficiently within structure - clean pockets, defined reads, and schemed separation - but see that efficiency decline as conditions become more demanding. CPOE does not diminish those players; it contextualizes them. It identifies the environments in which they succeed and, just as importantly, where they struggle. 

At the highest levels of football, the difference is rarely about making routine plays. It is about making the ones that are not supposed to be made.

Here is a list of how FBS & FCS QBs from the 2025-26 season compare: 

It Was Never JUST About The Quarterback

CPOE isn’t just a quarterback stat. Mendoza has said it many times that he was fortunate enough to have an excellent surrounding cast of receivers, running backs, tight ends, and of course an elite offensive line unit, in addition to the innovative offensive minds of HC Curt Cignetti and OC Mike Shanahan.

CPOE is a variable stat that does not completely isolate an individual player from their teammates. In football, it’s very difficult to evaluate one player's impact on a play without acknowledging the other 21 players on the field. However, we’ve proven that CPOE offers valuable insight towards evaluating quarterbacks and other offensive skill players. 

Here is a list of how FBS & FCS WRs from the 2025-26 season compare: 

When Margins are Thin, Context Matters Most

At the highest levels of football the margins are thin. The separating factor is performance under constraint—when coverage is tight, pressure is high, and the design of the play begins to break down.

For evaluators, the value is not in replacing the eye test, but reinforcing it. Every experienced scout recognizes the instinctive reaction to a play that stands out, even if articulating why can be challenging. CPOE provides structure to that instinct. It quantifies what the eye already senses, aligning subjective evaluation with objective context.

By isolating those moments, one can shift the conversation from, “Did he complete the pass?” to “How impressive was that completion given the context?” and “Based on the data, I bet he could do it again.”

And that’s a much more valuable question.

The next time you’re watching a game and see a highlight throw or catch, take a second and think, “How likely was that play to succeed?”, because the answer to that question probably tells you more about a player than the result ever could.

‍