How to Actually Read NFL Player Performance Metrics Beyond the Box Score
Most people look at a box score and call it analysis. Points, yards, tackles. That's the surface layer. The actual metrics that explain why a team won or lost are buried in tracking data, EPA, CPOE, run-block win rates, and pressure rates. When I first got into this stuff, I spent hours pulling raw data from various sites, cross-referencing PFF grades with NFL Next Gen Stats, and realizing most of it didn't tell the whole story. The Commanders-Patriots matchup is a good case study because both teams have been in transition phases, which makes their performance metrics even more telling. Let me walk through how I actually broke this down when I was putting together a video essay on it. First, I pulled snap counts and defensive snap counts from Pro Football Reference. That tells you who's actually on the field and in what packages. From there, I cross-referenced with NFL's official play-by-play data to get EPA per play. Expected Points Added is the single most useful metric if you actually want to understand what happened on each down. It measures how much a play changed the probability of scoring based on down, distance, field position, and game state.
I also grabbed CPOE (Completion Percentage Over Expectation) for both quarterbacks. This shows whether a QB is throwing to open receivers or forcing balls into tight windows. In that Commanders-Patriots game, the difference in CPOE between the two was more revealing than the raw completion percentage. One guy might throw 60% completions with high CPOE, while the other throws 65% with near-zero CPOE. The second guy is completing easy reads but not moving the ball effectively when it matters.
The Data Sources I Actually Use
NFL Next Gen Stats gives you player tracking data. You can see separation on routes, time to throw, pressure rate against, and yards after catch. It's not perfect, but it's the best publicly available source. PFF has their own grades which are controversial but useful when taken in aggregate. I don't trust individual player grades blindly. I've seen PFF grade guys wildly differently than what the tracking data shows, especially on defense where coverage assignments aren't always clear from film alone. I also use Football Outsiders' DVOA when it's available for the season context. It's not game-by-game useful, but it tells you whether a team's performance was sustainable or luck-based. After the Commanders-Patriots matchup, I checked DVOA to see if the winner's performance aligned with their season trend or if it was a one-off.
Get the Full Details
What Most People Miss About These Metrics
The biggest mistake I see is treating EPA as the final word. EPA doesn't account for scheme. A team that runs a lot of draw plays and short routes will naturally have lower EPA per attempt than a team that bombs it downfield. You have to contextualize. When I analyzed that Patriots offense, their EPA looked weak until I broke it down by play type. Their designed pass plays actually had above-average EPA. It was the quarterback pressures and negative plays on rollout that dragged the number down. Another thing: pressure rate against the pocket is useless without knowing what kind of pressures. Sack rate matters. But hurries and hits don't always lead to negative plays. Sometimes they force a quick throw that still gains yards. I've found that combining pressure rate with drop rate gives you a clearer picture. If a quarterback is getting hit and still releasing in under 2.5 seconds with decent accuracy, that's not always a broken offensive line situation. Sometimes it's a feature, not a bug.
A Problem I Hit and How I Worked Around It
When I was digging into the Commanders' defensive metrics for that breakdown, I noticed their EPA allowed ranked poorly against the run but well against the pass. Superficially that means they can't stop the run. But when I looked at the situational data, the issue was specific to early downs. On third and long, their EPA allowed was actually elite. The problem was they were in those situations constantly because the offense couldn't sustain drives, which meant the defense was on the field too much and getting gashed on running plays when they were tired or out of position. The workaround was looking at first-and-ten success rate allowed and opponent yards per carry by down and distance. That separated the schematic failure from the situational fatigue problem. It turned out the Commanders' defensive front had decent individual matchups on early downs but got worn down by sustained drives they couldn't prevent. That's a play-calling issue, not a talent issue. Most casual analysis would just say "can't stop the run" and move on.
Putting It Together For Your Own Analysis
If you want to do this yourself, start with the basic box score to identify what seemed unusual. Then pull EPA by play type, CPOE, pressure rate, and run-block win rate. Don't take any single metric as gospel. Cross-reference everything. If one data point conflicts with film, the film usually wins unless you have a compelling reason not to. I keep a spreadsheet with columns for each key metric across the last five games. Trend lines matter more than single-game numbers. A quarterback who looks bad in one game might be a blip. A quarterback whose CPOE has dropped consistently over five games while his pressure rate has climbed is trending in the wrong direction regardless of what one box score says. The Commanders-Patriots matchup had enough narrative around it that casual fans would come with opinions. The data helped separate signal from noise. Both teams had moments that looked bad on the stat sheet but made sense when you looked at the context. And vice versa. That's the whole point of doing this work properly.
