NFL QB Advanced Stats: EPA, CPOE & Pressure Rates
Every qualifying NFL quarterback, ranked by what actually predicts winning: EPA per dropback, completion percentage over expected (CPOE), and the pressure and accuracy profile underneath the box score. The quadrant chart plots efficiency against accuracy — the same view sharp bettors have used for years to separate real QB play from stat-padding — and the table underneath adds ANY/A, depth of target, time to throw, and how often each passer is under duress.
What you won’t find anywhere else is the regression flag. When a quarterback’s production is running well ahead of his accuracy profile, the board marks him with a red triangle — production that tends to come back to earth. A green triangle is the opposite: the underlying throwing is better than the results, the classic buy-low signal for futures and season-long markets. The board refreshes every morning from the previous night’s data during the season.
| QB | DB | EPA/db | CPOE | ANY/A | aDOT | TTT | AGG% | AYTS | INT% | Prs% | BadTh% | Reg |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Brock Purdy SF | 145 | 0.40 | +4.7 | 9.32 | 7.8 | 3.05 | 16.2 | -0.7 | 1.4 | — | — | ▼ |
| Lamar Jackson BAL | 103 | 0.36 | +8.1 | 9.50 | 10.1 | 3.11 | 12.5 | 1.2 | 1.0 | — | — | — |
| Jared Goff DET | 202 | 0.32 | +1.3 | 8.11 | 7.9 | 2.84 | 19.4 | -0.5 | 0.0 | — | — | ▼ |
| Josh Allen BUF | 148 | 0.29 | +5.0 | 7.78 | 9.5 | 2.94 | 10.9 | 0.7 | 2.9 | — | — | — |
| Trevor Lawrence JAX | 153 | 0.27 | +6.0 | 7.63 | 7.9 | 2.72 | 15.4 | -1.5 | 2.1 | — | — | — |
| Bryce Young CAR | 172 | 0.24 | +5.7 | 7.53 | 9.0 | 2.93 | 13.6 | 0.4 | 1.2 | — | — | — |
| Dak Prescott DAL | 198 | 0.23 | +1.9 | 7.04 | 7.5 | 2.70 | 21.4 | -0.6 | 1.6 | — | — | — |
| Patrick Mahomes KC | 134 | 0.19 | +0.8 | 8.13 | 7.0 | 2.70 | 8.6 | -1.4 | 1.6 | — | — | ▼ |
| Kirk Cousins LV | 180 | 0.19 | +1.4 | 7.11 | 8.2 | 2.93 | 15.7 | -0.4 | 2.3 | — | — | — |
| Joe Burrow CIN | 198 | 0.17 | +6.0 | 6.99 | 7.2 | 2.57 | 13.8 | -1.3 | 1.6 | — | — | ▲ |
| Matthew Stafford LA | 223 | 0.14 | +0.3 | 6.43 | 8.7 | 2.76 | 17.1 | 0.0 | 2.8 | — | — | ▼ |
| Drake Maye NE | 162 | 0.13 | +6.7 | 6.33 | 9.2 | 3.06 | 14.6 | 0.6 | 4.6 | — | — | ▲ |
| Jordan Love GB | 195 | 0.11 | -3.1 | 7.28 | 8.7 | 2.51 | 14.1 | 0.0 | 1.6 | — | — | ▼ |
| C.J. Stroud HOU | 199 | 0.09 | +2.5 | 7.18 | 9.6 | 2.96 | 20.0 | 0.8 | 0.0 | — | — | — |
| Geno Smith NYJ | 163 | 0.05 | +4.4 | 6.56 | 6.8 | 2.93 | 16.3 | -2.8 | 0.0 | — | — | — |
| Deshaun Watson CLE | 167 | 0.02 | -2.4 | 5.67 | 7.2 | 2.89 | 15.5 | -2.4 | 1.4 | — | — | ▼ |
| Jacoby Brissett ARI | 211 | -0.02 | +2.0 | 4.72 | 7.1 | 2.62 | 24.6 | -1.3 | 2.5 | — | — | — |
| Tyler Shough NO | 218 | -0.02 | +1.9 | 5.30 | 8.0 | 2.75 | 16.6 | -1.2 | 2.4 | — | — | — |
| Jameis Winston NYG | 143 | -0.06 | -5.9 | 5.12 | 9.6 | 2.71 | 21.2 | -0.1 | 2.3 | — | — | ▼ |
| Jalen Hurts PHI | 167 | -0.06 | -2.2 | 5.26 | 7.5 | 3.04 | 14.3 | -1.5 | 2.0 | — | — | — |
| Aaron Rodgers PIT | 198 | -0.07 | -2.5 | 5.74 | 8.6 | 2.82 | 17.4 | -1.0 | 2.2 | — | — | — |
| Bo Nix DEN | 197 | -0.08 | -2.4 | 5.04 | 4.9 | 2.95 | 9.5 | -3.2 | 1.6 | — | — | — |
| Kyler Murray MIN | 110 | -0.08 | +5.7 | 4.95 | 6.3 | 2.97 | 6.9 | -3.3 | 2.0 | — | — | ▲ |
| Daniel Jones IND | 185 | -0.13 | -2.4 | 4.57 | 7.3 | 2.77 | 14.8 | -1.3 | 2.3 | — | — | — |
| Sam Darnold SEA | 163 | -0.13 | -0.1 | 4.67 | 7.2 | 3.10 | 18.8 | -1.2 | 4.7 | — | — | — |
| Cam Ward TEN | 165 | -0.18 | +1.6 | 4.51 | 7.0 | 2.96 | 8.4 | -2.8 | 1.9 | — | — | ▲ |
| Justin Herbert LAC | 160 | -0.19 | -4.5 | 3.80 | 8.4 | 2.81 | 14.7 | -0.3 | 4.9 | — | — | — |
| Malik Willis MIA | 144 | -0.20 | -7.3 | 4.36 | 10.3 | 3.16 | 22.3 | 1.1 | 3.1 | — | — | — |
| Baker Mayfield TB | 109 | -0.27 | +0.3 | 4.63 | 8.9 | 2.99 | 12.5 | -0.9 | 2.1 | — | — | ▲ |
How to Read the Board
- The chart — each dot is a qualifying QB, placed by CPOE (accuracy vs. expectation, left to right) and EPA per dropback (value per play, bottom to top). The dashed lines are league averages, so the top-right quadrant is elite on both axes, and the bottom-right is production without accuracy — the “scheme-aided” zone. Hover any dot for the exact numbers.
- Green and red chips — a value gets shaded only when it sits in the top or bottom 10% of qualifiers. If a cell is plain, it’s somewhere in the middle of the league. Direction is handled for you: a green INT% chip means a low interception rate.
- The Reg column — ▼ means production is outrunning the accuracy profile (regression risk); ▲ means the profile is ahead of the results (positive regression candidate). Hover the arrow for the plain-English read.
The QB Metrics Defined
EPA per dropback is expected points added, per play — how much each dropback moved the team’s scoring expectation, sacks included. It’s the single best public measure of quarterback value because it weighs down-and-distance and field position instead of treating every yard the same. League average among qualifiers sits near the dashed line on the chart; sustained marks above +0.20 are MVP-conversation territory.
CPOE (completion percentage over expected) compares each QB’s completion rate to what an average passer would complete on the same throws, given depth, direction, and coverage. It strips out the easy-throw diet: a QB living on screens can post a high raw completion percentage and a flat CPOE at the same time. CPOE is also stickier year to year than EPA, which is exactly why the gap between the two drives the regression flag.
ANY/A (adjusted net yards per attempt) folds touchdowns, interceptions, and sack yardage into a single per-dropback yardage number. It’s the old-school efficiency stat that still correlates strongly with winning, and it’s a useful cross-check on EPA.
aDOT (average depth of target) describes style, not quality — how far downfield the ball travels per throw. Read it alongside CPOE: a +3 CPOE at a 9-yard aDOT is a much harder trick than the same CPOE at 6 yards.
TTT and AGG% come from NFL Next Gen Stats player tracking. Time to throw (seconds from snap to release) separates quick-game processors from hold-and-hunt QBs, and it reframes the pressure column — a passer with a high pressure rate and a slow clock is generating his own heat. Aggressiveness is the share of throws into tight windows; paired with a low CPOE, it’s a turnover profile waiting to express itself.
AYTS (air yards to the sticks) shows whether a QB’s average throw travels past the first-down marker or short of it — the checkdown detector.
Pressure% and Bad Throw% come from Pro-Football-Reference’s charting. Pressure rate is context for everything else on the row; bad-throw rate is the closest free analog to a turnover-worthy-play rate, and a QB with a low INT% but a high bad-throw rate has been getting lucky.
How Bettors Use QB Advanced Stats
The board isn’t a picks page — it’s the layer under the picks. A few of the higher-leverage applications:
- Season win totals and futures. EPA/CPOE composites are among the most predictive public QB metrics from one season to the next, and the regression flags tell you which of this year’s numbers to trust. Check the flags before you touch a win total built on last season’s passing production.
- Game lines. When the market is still pricing a QB off his box score and the board says the profile underneath is better or worse, that gap is where early-week numbers get beat.
- Touchdown and passing props. aDOT, AYTS, and aggressiveness tell you whose production comes in the red zone versus between the twenties — context that matters for TD props more than raw yardage does.
Related boards: RB advanced stats · WR advanced stats · TE advanced stats
Where the Data Comes From
Play-by-play EPA and CPOE are built on the open-source nflverse (nflfastR) data, tracking metrics come from NFL Next Gen Stats, and the pressure and accuracy charting comes from Pro-Football-Reference via nflverse. The board refreshes every morning; a QB qualifies at 100 dropbacks (roughly three starts), so early-season boards fill in as the sample builds. EPA per dropback excludes scrambles.
Frequently Asked Questions
What is EPA per dropback?
Expected points added per dropback measures how much each pass play (including sacks) changed the team’s expected points, based on down, distance, and field position. It’s the most widely used single number for QB value because a 5-yard gain on 3rd-and-4 and a 5-yard gain on 3rd-and-15 are not the same play.
What is a good CPOE?
League average is zero by construction. Anything above +2 is genuinely accurate, +4 or better is elite, and negative CPOE at volume is a real warning sign regardless of what the raw completion percentage says.
What does the red triangle next to a quarterback mean?
His production percentile (EPA per dropback) is running at least 25 points ahead of his accuracy percentile (CPOE and on-target rate). Historically that gap closes toward the accuracy side — it flags QBs whose results are likely to regress, not QBs who are bad.
Why isn’t a quarterback I’m looking for on the board?
He hasn’t hit the 100-dropback qualifying minimum yet. Backups and injured starters appear once the sample is large enough for the rate stats to mean something.
How often do these stats update?
Daily, early each morning, from the previous day’s completed games. During the season the board reflects the full week’s action by Tuesday morning.
CLEATZ provides information and analysis, not betting advice. 21+. If you or someone you know has a gambling problem, call 1-800-GAMBLER.
