CPOE (completion percentage over expected) tells you how often an NFL quarterback finishes throws above or below what an average QB would do on the same attempts. Raw completion percentage lies a lot: screens and checkdowns puff it up, and tough deep shots drag it down even when the passer is on point under pressure.
What is CPOE in football?
CPOE is a quarterback accuracy stat that stacks actual completions against the completion rate you’d expect on each throw once difficulty is in the mix. Plain version: given target depth, receiver separation, pressure on the passer, and similar factors, what share of those throws should an average NFL QB complete, and did this guy clear that bar?
The idea comes from tracking data, most famously NFL Next Gen Stats. Sensors and models estimate how completable each pass was before the ball left the hand. A short dump-off to a wide-open back might sit near a 90% expected completion rate. A tight-window deep out on third-and-long might land closer to 40%. Hit 10 of 12 on the easy stuff and you are not “elite.” You did what the play called for. Hit 6 of 10 on low-probability throws and that is the skill CPOE is built to catch.
You will usually see CPOE in percentage points. A +3.0 CPOE means the QB completed about three more percentage points of his attempts than expected. A -2.5 means he under-completed relative to that throw mix. Season samples beat one Sunday the same way you would not crown a hitter after two series. For the broader football stats frame that pairs with this, see our breakdown of EPA and why play value beats raw yards.
Why does completion percentage mislead on NFL QBs?
Completion percentage misleads because it treats every attempt as equal when the throws are nothing alike. A 68% completer who lives on screens is not automatically sharper than a 62% completer who pushes the ball downfield into tighter windows.
Scheme and supporting cast drive a huge share of the raw rate. West Coast and quick-game offenses mint high-percentage targets on purpose. Vertical offenses eat more incompletions in exchange for chunk plays. A clean pocket with elite route runners lifts completion rates without the QB becoming a different person overnight. A collapsing pocket with contested catches drops them for the same reason.
Old box scores make the trap obvious. Plenty of high-volume game managers have posted pretty completion rates while rarely stressing a defense past the sticks. Plenty of aggressive starters have looked “inaccurate” on paper while creating the throws that actually move expected points. If you only rank QBs by completion percentage from Pro-Football-Reference season passing tables, you will overrate safe distribution and underrate difficult accuracy.
| Throw profile | What raw Comp% says | What CPOE asks |
|---|---|---|
| Screens and checkdowns | Looks elite | Did he only hit layups? |
| Intermediate timing routes | Looks solid | Was separation already won? |
| Contested deep shots | Looks sloppy | Did he beat a hard expected rate? |
| Pressure / off-platform | Looks shaky | Was the throw mix unusually hard? |
| Play-action and RPO game | Mixed bag | Did scheme create freebie completions? |
Takeaway: completion percentage grades the result. CPOE grades the result against how hard the test was.
Before you trust a QB's completion percentage for a prediction:
- Check average depth of target and how often he throws behind the line versus past the sticks.
- Separate clean-pocket work from pressure snaps so scheme and protection do not hide the skill gap.
- Prefer multi-game CPOE over one hot or cold afternoon, especially after a role or coordinator change.
How do you read a good or bad CPOE number?
A good CPOE stays positive across a real sample, not a one-week spike off a soft throw chart. Roughly, season-long marks a couple points above expected put a QB in strong company. Large negatives over many attempts are a warning that the misses are not just “tough luck on deep balls.”
Context still matters. A backup forced into pure dropback work behind a bad line can post ugly CPOE without being permanently broken. A starter in a screen-heavy plan can sit near average CPOE while looking great in the box score. Pair the number with attempt volume, pressure rate allowed, and who is catching the ball. CPOE is an accuracy residual, not a full quarterback grade.
Watch regression patterns too. Extreme early-season CPOE, up or down, often cools as the throw sample grows and defenses adjust. That does not make the stat useless. It means you treat week 2 the way you should treat any small football sample: useful direction, weak conviction. The same caution shows up when people lean on yards alone; our note on success rate and why yards lie hits a related point from the play level.
How should CPOE change what you predict next?
CPOE should change your next-game QB read by moving weight off “he completes a high share” and onto “he completes more than this throw mix deserves.” If you are forecasting completions, completion percentage, or whether a passer will keep torching a soft zone look, CPOE helps you separate repeatable accuracy from schedule and scheme juice.
Use it as a correction factor. When a QB’s recent completion rate jumped because the offense lived on short outs and backs in the flat, dial back the idea that true accuracy leaped overnight. When completion rate dipped but CPOE held steady on a tougher chart, do not panic-sell the skill. The prediction question is always about the next environment: opponent coverage strength, likely game script, protection, and whether the play caller will keep asking for the same throw difficulty.
A practical sequence looks like this. Start with role and volume: is this QB still the full-time passer, and will the game stay on schedule enough for normal dropbacks? Then check throw difficulty trends: depth, pressure, contested share. Then bring in CPOE as the accuracy residual on top of that mix. Only after that should you decide whether last week’s completion percentage is a fair baseline for this week. If you want a place to score those reads against real player lines with friends, Download GAGE and put the process to work on game day.
What CPOE does not measure
CPOE does not measure decision quality, sack avoidance, run threat, or whether the throw was the right idea against the coverage. A QB can post strong CPOE and still kill drives with late reads, tip-drill interceptions, or awful pocket movement. Another can sit near average CPOE while winning with mobility, play-action leverage, and smart checkdowns that keep the offense on schedule.
It also does not fully clean up receiver and officiating noise. Drops, unique catch radius, and contested-catch talent still move outcomes after the ball is thrown. Tracking models try to price separation and difficulty, but football stays messy. Treat CPOE as one sharp lens on accuracy, then stack it with EPA-style value, turnover-worthy play rates when you have them, and plain film notes on who is open.
And CPOE is not a personality award. “Gunslinger” and “game manager” labels are fan shorthand. The useful split is simpler: how hard were the throws, and did the QB beat expectation on them? That framing keeps you out of narrative traps when a completion percentage looks pretty or ugly for the wrong reason.
How does CPOE compare with plain completion percentage for scouting form?
CPOE beats plain completion percentage for scouting form because it adjusts for throw difficulty instead of rewarding the easiest chart. Completion percentage is still a fine descriptive stat for what showed up in the box score. It is a weak causal stat for why it happened and whether it should continue.
If two QBs both sit at 65% over a month, and one got there on a diet of short, high-probability targets while the other posted the same rate on deeper and tighter throws, they are not the same projection going forward. CPOE is the cleaner way to say that without pretending one number explains the whole position. For public season-level counting stats you can still verify the raw side on Pro-Football-Reference and then ask the CPOE question on top: was that completion rate earned the hard way?
What does CPOE stand for in the NFL?
CPOE stands for completion percentage over expected, a tracking-based measure of how a quarterback’s actual completion rate compares with the rate expected on throws of similar difficulty.
Is a higher CPOE always better for predicting the next game?
Higher CPOE is generally a better accuracy signal, but only when the sample is large enough and the upcoming throw mix looks similar. A tiny hot streak or a sudden scheme change can still fool you.
Can completion percentage and CPOE disagree for a long time?
Yes. A QB in a short-pass offense can keep a high completion percentage with average CPOE for months, while a downfield passer can look worse on raw rate and still clear expected completion on tougher shots.