It was 12:40 in the morning on the Wednesday after we lost in Austin, and there was one sheet of paper on my desk that hadn't been there when I went down the hall for coffee.
In This Chapter
- The Hook: The Receipt
- 34.1 What a Defensive Number Is: Count, Denominator, Decision
- 34.2 Expected Points and EPA: Putting Every Snap in the Same Currency
- 34.3 Success Rate: How Often the Offense Stayed on Schedule
- 34.4 Pressure: Blitz Rate, Pressure Rate, and the Race Against the Clock
- 34.5 Coverage Efficiency: What Each Coverage Gives Up — and Why the Season Table Lies
- 34.6 Explosive Plays: Where Defenses Actually Lose Games
- 34.7 Turnovers and Turnover-Worthy Plays: Habit, Luck, and the Quarterback Who Gives You Chances
- 34.8 The Autopsy: My Cover 0 Call Against Owen's Numbers
- 34.9 The Weekly Dashboard: Eight Numbers and the Decisions They Drive
- Spaced Review
- Project Checkpoint: Defensive Metrics Dashboard
- Chapter Summary
- What's Next
Chapter 34 — Analytics and Defensive Decision-Making: Pressure Rate, Coverage Efficiency, Expected Points, and What the Numbers Say
The Hook: The Receipt
It was 12:40 in the morning on the Wednesday after we lost in Austin, and there was one sheet of paper on my desk that hadn't been there when I went down the hall for coffee.
Owen Kessler doesn't leave binders. He leaves single pages, squared to the edge of the desk, with a sticky note on top. This note said: You asked. — O.
I had asked. Monday morning I'd shown the 52-yard touchdown to the whole defense four times and told them the call was mine. Tuesday, in the staff room, Owen had said the line that got repeated around the building all week: "Bad decision, good outcome — that's the most dangerous thing in football." On my way out Tuesday night I'd stopped at his door and said, "Write me a receipt. What did that call actually cost us? Not the scoreboard. The call."
Here's what the page said, near enough.
| Line on Owen's receipt | What it said |
|---|---|
| The snap | Week 6, 4th quarter, 4:12, tied 27–27. Austin 3rd-and-8 at its own 48. |
| The call | Nickel, Over, Mug, Mac Dog, Zero, Lurk. Six rushers, five in man, nobody deep. |
| What the play cost (the result) | About six expected points on one snap. A likely punt became a touchdown. |
| What the call cost (the decision) | About a quarter of a point per snap, on average — before adjusting for Vance. |
| After adjusting for Vance | More. His row against five-plus rushers (28 of 36) makes zero's bad outcomes more likely, not less. |
| The game-state cost | Tied, 4:12 left. A stop was enough. We bought variance we didn't need. |
| The sample | 36 attempts is a plan-sized sample, not a certainty. The direction is clear. The size is fuzzy. |
| Bottom line | Grade the decision, not the result. The decision was still wrong. |
I read it three times. In thirty-one years of coaching I had been fired once, praised in newspapers a few times, and yelled at by more head coaches than I can count, but I had never been handed an invoice for a single call. The thing that bothered me wasn't the six points. I knew about the six points; I'd watched them on a loop for two days. The thing that bothered me was the second row. A quarter of a point per snap. That's not a disaster. That's a leak. A coordinator who makes that kind of call twice a game, every game, loses four or five points a week and never sees a 52-yard touchdown to remind him — until he does.
At 7:15 the next morning Owen was in my office before the coffee was done.
Russ: The second row. Walk me through it.
Owen: It's the whiteboard picture from Chapter 16's meeting. Per hundred third-and-8s near midfield, zero and a four-man rush get stopped about the same amount. Zero turns some stops into sacks and some first downs into touchdowns. I put a value on each outcome and added them up. Zero came out about a quarter of a point worse every time you call it. That's with an ordinary quarterback. Vance isn't ordinary.
Russ: And if Coleman gets there half a second sooner, the receipt says what?
Owen: Same second row. Different first row. That's the point of the receipt. The first row is what happened. The second row is what you chose.
That conversation is this chapter. Analytics isn't a separate department that sends you numbers. It's a way of separating what happened from what you chose, so that you learn from the right one. It tells you which pass rush is actually working, which coverage is actually leaking, which quarterback is actually dangerous with the ball, and how big your sample really is. It also has limits, and a coordinator who doesn't know them will take a decimal point more seriously than a film clip. I've done both.
You'll learn the numbers the way our staff learned them: from Owen, with his sample sizes and caveats attached, on our own Sentinels data (all of it fictional) and with honest pointers to the public sources real analysts use.
🏃 Fast Track: If you already speak EPA and success rate, skim 34.2 and 34.3. Read 34.4 on why pressure rate and blitz rate are different questions, and spend your time on 34.5 (why a coverage's season EPA can lie), 34.8 (the Cover 0 autopsy), and 34.9 (the eight-number dashboard that is this chapter's project).
🔬 Deep Dive: Appendix H holds every metric definition in this chapter with the counting rules Owen uses. For the game theory of mixing pressure and coverage, reread 25.7 in Chapter 25. For the public data, start with the nflverse play-by-play described in 34.1.
📺 Film Watcher: This Sunday, pick one defense and count two things on every dropback: how many rushers came, and whether the quarterback had to move or throw early. Then watch where the ball went when he got rid of it fast. You'll have charted blitz rate, pressure rate, and the hot throw — the three numbers this chapter keeps coming back to.
34.1 What a Defensive Number Is: Count, Denominator, Decision
Owen has a rule he gave our staff his first week in the building, back in my first season in Sacramento: a defensive number with no definition and no sample size is an opinion with a decimal point. Every number he brings into my office has to survive three questions.
- What exactly did you count? "Pressure" means one thing at one charting service and something else at another. "Explosive play" means 12+ yard runs and 20+ yard passes in this book, and different thresholds elsewhere. If you don't know the definition, you don't know the number.
- Out of what? A count needs a denominator — the total it's divided by. "We gave up eleven explosive passes in Cover 1" means nothing until you know Cover 1 was the coverage on 108 dropbacks, not 20 and not 400.
- What decision does it change? If the answer is "none," it doesn't go on the sheet. A coordinator's attention on Saturday night is the scarcest resource in the building.
Owen adds a fourth question, almost under his breath, every time: how many? That's the sample size, and it decides how hard you're allowed to lean on the number.
The confidence ladder
You met Owen's ladder in Chapter 24. It's worth repeating because it governs everything here.
| Snaps (or attempts) behind the number | Owen calls it | What you may do with it |
|---|---|---|
| Under about 10 | A story | Show it in the film room. Don't build a call around it. |
| About 10 to 25 | A lean | Tilt a decision when other evidence agrees. |
| About 25 or more | A plan | Build a call-sheet box around it, and keep checking. |
And Owen's line from Chapter 26 belongs next to the ladder: "A tendency is a count plus a reason." A number with no mechanism behind it is a coincidence until proven otherwise. A number with a mechanism — Ruffin stays in to block, so the sixth rusher is never free — is a plan.
How fuzzy is a percentage? The square-root rule of thumb
Here's the one piece of arithmetic Owen makes every new intern memorize. For a percentage built from a count, a rough "worst-case" margin of uncertainty is one divided by the square root of the count.
- Count of 36: the square root is 6, and 1 ÷ 6 is about 0.17. Treat the percentage as fuzzy by roughly 17 points either way.
- Count of 100: the square root is 10. Fuzzy by about 10 points.
- Count of 400: the square root is 20. Fuzzy by about 5 points.
It's a deliberately cautious rule — real margins are often a bit smaller, especially when the percentage is far from 50 — but it's the right instinct for a coach. Now apply it to the most important row in this book.
Vance against five or more rushers: 28 of 36, about 78 percent. Fuzzy by up to about 17 points: somewhere between about 61 and 95 percent is consistent with that sample.
Vance against a four-man rush with post-snap rotation: 31 of 52, about 60 percent. Fuzzy by up to about 14 points (the square root of 52 is a bit over 7): somewhere between about 46 and 74 percent.
Those two ranges overlap. If completion percentage were the only evidence, Owen would tell you the gap is real-looking but not proven. It isn't the only evidence. The yards per attempt (9.4 against 6.1), the touchdowns and interceptions (4 and 0 against 1 and 3), and the mechanism — Austin kept Ruffin in to block on 19 of 27 long third downs, and Vance's hot throw comes out in about a second — all point the same direction. Several independent signals pointing one way is how a fuzzy sample becomes a plan. This is the sentence Owen said in Chapter 6, and it's the right way to hold any number: "I'd bet on the direction. I wouldn't bet the house on the size."
💡 Coach's point: Rare events are the fuzziest numbers you'll ever see. Three interceptions in 52 throws could easily have been one or five with the same quarterback making the same decisions. Never build a plan on a count in single digits unless the film shows you why.
Where real analysts get their numbers
Everything about the Sentinels in this chapter is fictional. The methods are real, and so are the public sources that analysts, media, and many front offices use. Know what each one is built from, because that decides what it can and can't tell you.
| Source | What it's built from | What it gives a defensive coach | What to watch out for |
|---|---|---|---|
| nflverse / nflfastR | The league's official play-by-play, cleaned and published free, going back to 1999, with open-source expected-points and win-probability models added | EPA, success rate, win probability, explosive plays, situational splits — for every team, free | Play-by-play doesn't say what coverage was called or how many rushed. In recent seasons nflverse has also distributed some third-party charting fields (for example from FTN) for certain seasons; check what's currently available |
| NFL Next Gen Stats | Player-tracking chips in the shoulder pads, recording position and speed many times a second | Time to throw, receiver separation, player speed, and stats derived from tracking | The public site shows a subset; definitions are the league's; tracking tells you where, not what the assignment was |
| Pro Football Focus (PFF) | Human charters watching every play, plus player grades | Pressures by player, coverage snaps, targets allowed, turnover-worthy plays | Charting involves judgment; definitions are PFF's own; much is subscription |
| Sports Info Solutions (SIS) | Human charting of every play | Coverage scheme charting, pressure, blitz, and their own value metrics | Different definitions from PFF; numbers won't match across services |
| FTN / DVOA | Football Outsiders' DVOA (a success-based, opponent-adjusted efficiency system), continued at FTN after Football Outsiders wound down, plus FTN's own charting | Opponent-adjusted team defense ratings over many seasons | DVOA is a model with its own baselines; compare it with itself, not with raw EPA |
📈 By the numbers: If you take one habit from this table, take this one: compare within a source, never across sources. Most charting services put league-average pressure rate somewhere between the mid-20s and mid-30s percent, and the spread is mostly definition, not football. A defense that is "30 percent" at one service and "38 percent" at another hasn't changed; the ruler has.
🔄 Check your understanding: Tess tells you, "Our nickel has allowed 7 of 9 on slants." Ask Owen's four questions and say what you'd do with the number this week.
Answer
What was counted: completions on throws charted as slants with Vo as the nearest defender (you need to confirm Tess's definition — targeted on him, or in his zone?). Out of what: 9 targets. Decision it changes: possibly Vo's leverage or the call on slant-heavy downs. How many: 9 — under 10, so it's a story. The square-root rule says 78 percent from 9 attempts is fuzzy by more than 30 points. Show the nine clips in the film room and look for a reason (a technique flaw, a leverage rule, a specific route stem). If the clips show the same mistake, fix the technique — the film, not the percentage, is the evidence. Don't change the call sheet on nine snaps.
34.2 Expected Points and EPA: Putting Every Snap in the Same Currency
Yards lie. Five yards on 3rd-and-4 is a first down; five yards on 3rd-and-12 is a punt. Twenty yards from your own 10 is a nice gain; twenty yards from the opponent's 20 is a touchdown. A defense that "allowed 340 yards" might have played brilliantly or terribly — the yards alone can't tell you which. Analysts fixed this by converting every situation into one currency: points.
Expected points (EP) is the average number of points the team with the ball goes on to score next, from a given down, distance, and field position (most models also account for time remaining), built from many seasons of real play-by-play. The idea has a long history. It shows up in operations-research work in the early 1970s co-written by Virgil Carter, who was an NFL quarterback at the time; it was popularized for fans in The Hidden Game of Football by Bob Carroll, Pete Palmer, and John Thorn; and in the internet era it was spread by analysts such as Brian Burke and built into the free nflfastR models. "Next score" includes the other team scoring next, which is why expected points can be negative deep in your own territory.
Expected points added (EPA) is the change in expected points from before a play to after it. If the offense was worth +1.0 before the snap and +2.8 after, the play's EPA is +1.8. From the defense's chair, negative EPA is good. A defense's "EPA per play allowed" is the average EPA of every offensive play it faced.
The ladder
Here's the shape of expected points for a first-and-10 at different spots on the field. These numbers are illustrative and rounded — they show the general shape public models produce, not any model's exact output. Exact values differ by model and season.
EXPECTED POINTS, 1st-AND-10, BY FIELD POSITION (ILLUSTRATIVE ONLY)
Each # = 0.2 expected points for the offense.
Ball on EP 0 1 2 3 4 5 6
own 5 -0.5 <|
own 15 +0.2 |#
own 25 +0.8 |####
own 40 +1.6 |########
midfield +2.2 |###########
opp 35 +3.0 |###############
opp 20 +4.0 |####################
opp 5 +5.5 |############################
📊 Diagram (described): A horizontal bar chart showing the illustrative expected points an offense has with first-and-10 at eight spots on the field. At its own 5-yard line the value is slightly negative, about minus half a point, drawn as an arrow pointing left of zero. At its own 15 it is about plus 0.2; its own 25, about 0.8; its own 40, about 1.6; midfield, about 2.2; the opponent's 35, about 3.0; the opponent's 20, about 4.0; and the opponent's 5, about 5.5. The bars grow steadily longer as the offense moves toward the opponent's goal line, and the growth speeds up inside the opponent's 20.
Two things to notice. First, the value of field position rises smoothly — there's no magic line at midfield. Second, down and distance matter as much as field position. A third-and-8 at midfield is worth far less than a first-and-10 at midfield, because on third-and-8 the most likely outcome is a punt.
Three worked snaps
Owen teaches EPA with three snaps from our own season. Same rule each time: state the situation before, state it after, subtract.
Snap 1 — Ruffin's 16-yard run, Week 6, fourth quarter. On the drive that ended in the 52-yard touchdown, Austin started at its own 30 and Ruffin ran for 16 on first down.
- Before: 1st-and-10 at the Austin 30. Illustratively about +1.0 for Austin.
- After: 1st-and-10 at the Austin 46. Illustratively about +1.9.
- EPA: 1.9 − 1.0 = +0.9. A good run. Not a catastrophe; nobody scored, and they still had 54 yards to go.
Snap 2 — the 52-yard touchdown, three plays later. 3rd-and-8 at the Austin 48.
- Before: 3rd-and-8 near midfield. The most likely outcome is a punt, so it's worth only about +1.0 for Austin, illustratively.
- After: a touchdown — 7 points with the extra point, minus a little for the kickoff Austin must now give us. Call it about +6.5 net in round numbers.
- EPA: 6.5 − 1.0 = about +5.5. Owen rounded to "about six" on the receipt, and he told me his rounding was generous in my favor, not his. One snap. More than any five ordinary snaps combined.
Now look at what a stop would have been worth on that same snap.
- A stop → Austin punts → we get the ball around our own 15. That's worth about +0.3 for us, which is about −0.3 for Austin.
- EPA of a stop: −0.3 − 1.0 = about −1.3.
So the swing between "stop" and "touchdown" on that snap was almost seven points. Between "stop" and "conversion short of a big play" it was about three — which is exactly what Chapter 26 told you: a third-down stop near midfield is worth roughly a field goal, and nobody has to kick anything.
Snap 3 — the Mug Cat Fire sack, Game 2. Anchor B: 3rd-and-7 at the Austin 38, tied 13–13, third quarter. Bell's sack lost 9 yards, to the Austin 29: 4th-and-16, punt.
- Before: 3rd-and-7 at the Austin 38. Illustratively about +0.6 for Austin.
- After the sack: 4th-and-16 at the Austin 29. They'll punt from deeper; we'll likely start around our own 25. About −0.9 for Austin.
- EPA: −0.9 − 0.6 = about −1.5.
- Compare an incompletion on the same snap: 4th-and-7 at the 38, punt, we start around our own 15 to 20. About −0.4 for Austin, so EPA about −1.0.
The sack was worth about half a point more than an incompletion. That's real, and over a season those halves add up. But notice how small it is next to the gap between any stop and a touchdown. On third down, a sack and an incompletion are cousins. A first down and a touchdown are strangers. That one sentence is most of the analytics case against calling zero on third-and-long, and it falls straight out of the arithmetic.
🔍 Why this works: EPA works because it measures every play against the same question — how much closer did the offense get to scoring than it was before? — instead of against an arbitrary yardage line. That lets you add a 4-yard run on 1st-and-10, a 9-yard sack on third down, and a 52-yard touchdown into one honest average. Yards treat a meaningless 15-yard completion on 3rd-and-25 as a good play for the offense. EPA correctly shows it as a loss, because they're punting anyway.
EPA per play, pass and run
Add every snap up and divide by the number of snaps, and you have EPA per play allowed — the best single summary of how much a defense let an offense move toward points. Owen splits it two ways every week: EPA per dropback (every pass play, including sacks and scrambles, since those began as passes) and EPA per designed run.
Last season, by Owen's charting of our games (fictional), the Sentinels allowed:
| Plays | EPA per play allowed | |
|---|---|---|
| Dropbacks (passes, sacks, scrambles) | 590 | −0.02 |
| Designed runs | 460 | −0.08 |
| All plays | 1,050 | about −0.05 |
What do those mean? A negative number means the average snap against us moved the offense away from scoring. Across a 62-snap game, −0.05 per play is about three points of expected value taken away from an opponent, compared with an average-ish defense.
📈 By the numbers: In most recent seasons, by most public expected-points models, the league's average dropback has been worth slightly more than zero to the offense, and the average designed run slightly less than zero. That's one of the most-discussed findings in public football analytics, and it's why analysts argue that pass defense matters more than run defense per snap. It doesn't mean run defense doesn't matter — a defense that can't stop the run sees more runs in better situations, and Chapter 30's two-high trade-off is exactly that fight. Treat the exact league averages as moving targets; they shift from season to season and model to model.
What EPA can't see
Owen is the first to list EPA's blind spots, and he makes every new staff member say them back.
- It doesn't know your opponent. −0.05 against three top-10 offenses is better than −0.05 against three bottom-five offenses. Opponent-adjusted systems (DVOA is the best-known) try to correct this.
- It doesn't know garbage time. A team down 24 in the fourth quarter is playing a different game. Owen trims those snaps or flags them, the same sample-hygiene rule he used on the Austin report in Chapter 24.
- It treats one play as one play. A single snap's EPA is mostly noise about the decision and all signal about the result. Averages over many snaps are where it earns its keep.
- It can't see assignments. EPA knows a 41-yard post happened. It doesn't know the Lock/Quarters split bust caused it. That's what the film and the charting are for.
Win probability (WP) is EPA's cousin: the chance a team wins from a given score, clock, down, distance, and field position, again built from many seasons of games. EPA answers "how much did this play help the offense score?" WP answers "how much did this play change who wins?" Late in close games, WP is the better lens, because a field goal and a touchdown can mean the same thing (or totally different things) depending on the score. Chapter 16's "volatility is a tool when you're losing, a tax when you're tied or ahead" is win-probability thinking in plain English.
🔄 Check your understanding: Two plays, both allowed by our defense. (a) 3rd-and-15 at the opponent's 30: a 12-yard completion, they punt. (b) 2nd-and-3 at our 45: a 4-yard run, first down. Which one has positive EPA for the offense, and which one did your film grader probably mark as the "bigger play"?
Answer
(a) is almost certainly negative EPA for the offense: they were probably punting before the snap, and they're punting after it, from only slightly better field position — and they burned a down. (b) is positive: a new set of downs in our territory moves them meaningfully toward points. A grader counting yards would call (a) the bigger play (12 yards versus 4). EPA calls (b) the one that hurt you. That's the whole reason the currency exists.
34.3 Success Rate: How Often the Offense Stayed on Schedule
EPA is an average, and averages are dominated by big plays. One 52-yard touchdown outweighs a dozen stops. That's correct for scoring, but it hides something a coordinator needs to know: how often the offense did what it wanted. That's what success rate measures — the percentage of plays the offense would call a "win."
There are two common definitions, and you must know which one you're reading.
| Definition | A play is a success for the offense if… | Where you'll see it |
|---|---|---|
| Yardage version (the classic) | It gains at least 40% of the yards needed on 1st down, 60% on 2nd down, and 100% on 3rd or 4th down. Some analysts use slightly different cutoffs (50% on first down is common). | Coaching staffs, older analytics writing, many charting sheets |
| EPA version | Its EPA is greater than zero | nflfastR and most modern public analytics |
The two mostly agree. On first-and-10 a 4-yard gain is a success in the yardage version and roughly break-even in EPA; on 2nd-and-10, 6 yards is a success in the yardage version and probably slightly positive in EPA. They disagree at the edges — a 3-yard run on 1st-and-10 near the goal line can be positive EPA without reaching 40 percent — and that's fine as long as you never compare a number built one way with a number built the other.
Worked: grading a drive both ways
Owen used the first four snaps of a Week 11 Stallions drive (this season, at St. Louis) to teach this to Tess's interns. The yardage rule is easy to apply by hand; the EPA version he pulled from his model.
| Snap | Situation | Result | Yardage rule | EPA version |
|---|---|---|---|---|
| 1 | 1st-and-10, STL 25 | Day runs for 3 | Needed 4 (40% of 10) → fail | About −0.3 → fail |
| 2 | 2nd-and-7, STL 28 | Haugen completes for 6 | Needed 4.2 (60% of 7) → success | About +0.3 → success |
| 3 | 3rd-and-1, STL 34 | Day runs behind the guard for 2 | Needed 1 → success | About +0.6 → success |
| 4 | 1st-and-10, STL 36 | Haugen boot, incomplete | Needed 4 → fail | About −0.6 → fail |
Two of four: a 50 percent success rate on those snaps, and both definitions agree here. The EPA values are rounded illustrations, not Owen's actual model output; the point is the method.
Why a defense needs both numbers
EPA tells you how much you gave up. Success rate tells you how often. Put them on two axes and you get the most useful picture in defensive analytics — the four kinds of defense.
THE TWO-NUMBER MAP (success rate allowed vs. EPA per play allowed)
EPA per play allowed
LOWER (good) HIGHER (bad)
+-------------------------+-------------------------+
SUCCESS | "Bend, then hit" | "Death by a thousand" |
RATE | They move the chains | They win often AND |
HIGHER | but rarely hit big; | hit big. Broken. |
(bad) | holds in the red zone | |
+-------------------------+-------------------------+
SUCCESS | "Smothering" | "Boom or bust" |
RATE | Win most snaps, give | Win most snaps, but |
LOWER | up little. | explosives and busts |
(good) | Sentinels, last season | erase it. Sentinels, |
| (40%, -0.05) | Russ's first season |
+-------------------------+-------------------------+
📊 Diagram (described): A two-by-two grid. The horizontal axis is EPA per play allowed, with lower (good) on the left and higher (bad) on the right. The vertical axis is success rate allowed, with higher (bad) on top and lower (good) on the bottom. Top left is "Bend, then hit": offenses move the chains but rarely hit big plays. Top right is "Death by a thousand": offenses win often and hit big; the defense is broken. Bottom left is "Smothering": the defense wins most snaps and gives up little; the Sentinels last season sit here, at a 40 percent success rate and minus 0.05 EPA per play. Bottom right is "Boom or bust": the defense wins most snaps but explosives and busts erase the gains; the Sentinels in Russ's first season sit here.
My first Sentinels season was the bottom-right box, and I didn't know it until Owen drew this. Our success rate allowed was respectable — about 43 percent by his charting — but our EPA per play allowed was about +0.03, worse than average, because we gave up big plays: 29 communication busts that year, 11 of them explosives, and a 14 percent missed-tackle rate. We won lots of snaps and then lost the ones that mattered. The fix wasn't more pressure or a new coverage. It was communication (Chapter 8's check system and "No echo, no check"), tackling (Chapter 23), and run fits. By last season the busts were down to 9 and the missed tackles to 9 percent, and we'd moved left on the map.
Last season, by the EPA version, we allowed a 40 percent success rate overall — about 41 percent on dropbacks and 38 percent on designed runs. Most public definitions put league-average offensive success rate somewhere in the 40s, so 40 is good, not freakish. The EPA number (−0.05) is what moved us into the good box, and it moved because the explosives came down.
⚠️ Common mistake — "Bend, don't break" as a slogan instead of a measurement. A coordinator with a high success rate allowed and a decent EPA tells his head coach, "We bend, we don't break." Then one season the red-zone luck turns, the field goals become touchdowns, and the same defense is ranked 28th. The fix: if your success rate allowed is high, your EPA is being held up by a few situational stops (red zone, third-and-long) that are noisier than every-down efficiency. Know which box you're in and why. A defense that lives in "Bend, then hit" needs to be excellent in the red zone every year, and that's a hard thing to be every year.
🔄 Check your understanding: Austin's offense in Owen's 4-game sample would be in which box from its side — high success, high EPA? What does that tell you about where to aim the game plan?
Answer
Austin converted 47 percent of third downs and produced 23 explosives in 247 snaps, so from the offense's side it was both efficient and explosive — the defense's nightmare box. When an offense is good at both, you usually can't take away both. Our plan chose to take away the explosive side first (two-high shells, rotate, no zero, make Vance read) and accept some efficient underneath completions — then tackle them. "Let him check it down for five yards all day" is a choice to live in the "bend" row while forcing Austin's EPA down.
34.4 Pressure: Blitz Rate, Pressure Rate, and the Race Against the Clock
🧩 Productive struggle: Before you read this section, look at two numbers from Owen's charting of our defense last season. On dropbacks where we rushed four, we pressured the quarterback 31 percent of the time. On dropbacks where we rushed five or more, we pressured him 42 percent of the time. Write down your answer to this: should we have blitzed more? Then write down what other number you'd want to see before you'd be sure. Keep your answer. You'll grade it at the end of the section.
Three words get mashed together in every broadcast booth in America, and a coordinator has to keep them apart.
- Blitz rate — how often you send five or more rushers. It describes your call. You control it completely.
- Pressure rate — how often the quarterback is pressured: sacked, hit as or after he throws, or forced off his spot or into an early throw by a rusher, before the ball comes out. It describes a result. You influence it; you don't control it.
- Time to pressure — how long after the snap the pressure arrives. It decides whether the pressure mattered, because a rusher who arrives after the ball is gone has produced a highlight for his position coach and nothing for the defense.
Every service counts differently
Owen's definitions are his own, and he prints them at the top of every report.
| Question | Owen's rule | How others differ |
|---|---|---|
| What's a blitz? | 5+ rushers committed by about one second after the snap | Some services count any linebacker or defensive back rushing as a blitz even in a four-man rush (a sim pressure becomes a "blitz" in their data). Some count only rushers beyond four. |
| What about a Hug (green dog)? | Counted as a rusher only if he crosses the line before the throw | Some charters count the man by his pre-snap assignment, some by what he did |
| What's a pressure? | Sack, QB hit, or a hurry: a rusher forces the QB off his launch point or into an early throw. Must happen before or at the release. | Some include pressures that arrive after the throw. Some require a hurry to change the throw visibly. |
| Who gets credit? | The first rusher to cause it ("first to the quarterback") | Some credit every rusher who contributes |
That's why this book always says it the same way: most charting services put league-average pressure rate somewhere between the mid-20s and mid-30s percent, depending heavily on definition. Our 34 percent is measured on Owen's ruler and compared only with our own history on the same ruler.
Owen's pressure table, last season
Last season we faced about 590 dropbacks. We rushed five or more on 159 of them — our 27 percent blitz rate — and four or fewer on the other 431. Here is what Owen charted (fictional):
| Rush | Dropbacks | Pressures | Pressure rate | EPA per dropback allowed |
|---|---|---|---|---|
| Four-man rush (incl. sims) | 431 | 133 | 31% | −0.02 |
| Five or more | 159 | 67 | 42% | −0.04 |
| All dropbacks | 590 | 200 | 34% | −0.02 |
And here's the five-plus row broken out by type, using the same 159 blitzes and categories you saw in Chapter 18:
| Blitz type | Dropbacks | Unblocked rusher within 2.5 s (Ch 18) | Pressures | Pressure rate |
|---|---|---|---|---|
| Five-man fire zone with a lineman dropping | 46 | 19 | 22 | 48% |
| Five-man fire zone, no lineman drop | 42 | 9 | 17 | 40% |
| Five-man man-free (Cover 1 pressure) | 41 | 7 | 14 | 34% |
| Six or more (Thin, six-man Cover 1, Zero) | 30 | 11 | 14 | 47% |
| Total | 159 | 46 | 67 | 42% |
Now go back to your productive-struggle answer. Blitzing added about 11 points of pressure rate. What did it cost? Every extra rusher is one fewer coverage player, and the EPA column tells you the net: −0.04 per blitz versus −0.02 per four-man rush. A small edge in our favor overall — but look inside it. The fire zones were the good blitzes. The five-man man-free pressures, which needed a corner or nickel to win in man with no help underneath, were positive EPA for the offense (about +0.08 per dropback). The extra pressure on those 41 snaps didn't pay for the extra holes.
So the honest answer to "should we have blitzed more?" is: not more — differently. More fire zones in the right situations, fewer man-free pressures against quarterbacks who can find the hole. And against one particular quarterback, much less of everything — which is the next part.
🔍 Why this works: Pressure rate is a means, not an end. A pressure helps the defense only because pressured dropbacks are much worse for offenses than clean ones — by Owen's charting last season, about −0.40 EPA per pressured dropback against about +0.17 per clean one. Public charting consistently finds the same direction, at a size that varies by service. A blitz "works" only if the pressure it adds is worth more than the coverage it removes. That trade, not the pressure number alone, is what you're grading.
Where the pressure comes from
Owen also charts which rusher caused each pressure first. Of last season's roughly 200 pressured dropbacks, about 110 came off the edge first, about 60 from the interior first, and about 30 from a second-level blitzer first. The interior pressures were the most damaging — opponents completed about 40 percent at about 4.6 yards per attempt when the first pressure came up the middle, against about 53 percent and 6.2 yards when it came off the edge. That's the arithmetic behind Big Lou's rule from Chapter 19: a quarterback can step up away from an edge rusher, but he can't step away from a man in his face. It's also why Owen will never let me judge Rashad Coleman by his sack total. Coleman's pressures arrive in the quarterback's lap, and they change throws even when nobody gets a sack.
📈 By the numbers: Sacks are a small slice of pressures — by most charting, only a minority of pressured dropbacks end in a sack, with the exact share depending on the service — and sack totals jump around from season to season far more than pressure totals do. That's why analysts treat pressure rate (or tracking-based measures such as ESPN's pass rush win rate, which is built from Next Gen Stats tracking data and asks whether a rusher beat his block within about two and a half seconds) as a steadier read on a pass rush than sacks. A rusher with many pressures and few sacks is often a candidate for more sacks next year. A rusher with few pressures and many sacks is often a candidate for fewer.
Time to pressure versus time to throw
Here is the most important chart in this section. It's the race every dropback runs. Owen built the top rows from his timing study of our film (you saw the numbers in Chapter 18); the bottom rows add our pressure timing and two opposing quarterbacks' release times.
THE RACE: WHEN THE BALL COMES OUT vs. WHEN PRESSURE ARRIVES
Seconds after the snap. Owen's charting (fictional).
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
| | | | | | | |
Hot throw [==] 1.0-1.3
Free A-gap mugger [==] 1.6-1.9
Free edge rusher [===] 2.1-2.5
3-step last read [==] 2.2-2.5
Free DB from 6-8 [===] 2.4-2.8
Deep over / dagger [====] 2.8-3.3
- - - - - - - - - - - - - - - - - - - -
Our blitz pressure ^ median 2.1
Our 4-man pressure ^ median 2.6
Oduya release X avg 2.4
Vance vs. 4-man X avg 2.9
📊 Diagram (described): A timeline from zero to three and a half seconds after the snap, marked every half second. The top rows show time windows from Owen's film study: a hot throw comes out between 1.0 and 1.3 seconds; a free A-gap mug rusher arrives between 1.6 and 1.9; a free edge rusher between 2.1 and 2.5; a three-step drop's last read between 2.2 and 2.5; a free defensive back blitzing from six to eight yards between 2.4 and 2.8; and a deep over or dagger route between 2.8 and 3.3. Below a dashed line, a caret marks the Sentinels' median time to pressure on blitzes at 2.1 seconds and on four-man rushes at 2.6 seconds. An X marks Ospreys quarterback Marcus Oduya's average release at 2.4 seconds and another X marks Darius Vance's average release against a four-man rush at 2.9 seconds.
Read it from two quarterbacks' chairs.
Marcus Oduya gets the ball out in about 2.4 seconds. Our four-man rush, which typically gets home around 2.6, arrives after the ball on most snaps. Last season our pressure rate against Oklahoma City was only about 19 percent — and the pass rush wasn't bad that day. Oduya simply didn't hold the ball long enough to be pressured. Against a quick-game quarterback, pressure rate is the wrong number to grade. The right numbers are completion percentage on the short throws, yards after the catch, and missed tackles — which is why Chapter 17's press-versus-off decision against the Ospreys was built on completion and explosive rates, not on the rush.
Darius Vance is the opposite problem. Against a four-man rush he holds the ball (about 2.9 seconds on average in Owen's sample), hunting the deep over and the post. Our four-man rush has time to get there — that's when you get pressure from four. But blitz him, and look where his hot throw lives: 1.0 to 1.3 seconds, before even a free A-gap rusher can arrive. The blitz doesn't buy pressure against Vance. It buys him a shorter decision. That's the mechanism behind 28 of 36, and it's why Owen's number was a plan and not a story.
🗣️ The offense's view: Lyle Petrakis doesn't fear your pressure rate; he fears the pressure that arrives before his answer. His whole protection plan is built on making the answer arrive first: a hot throw at 1.1 seconds when you send more than he can block, max protection with the back in when he wants 3.3 seconds for the deep over, and a quarterback trained to know which one he's got before the snap. When a defensive coordinator brags about blitzing, Petrakis checks one thing on the film: when the blitz came, where did the ball go, and how fast?
🎬 Film room: On the end-zone film, start a stopwatch at the snap and stop it at the release on 20 straight dropbacks. Then time it again to the moment the first rusher reaches the quarterback's original spot. You've just built your own time-to-throw and time-to-pressure numbers for that opponent. If most of the releases come before most of the arrivals, rush lanes and tackling will win you more snaps than blitzes will.
🔄 Check your understanding: Big Lou wants to reward his linemen for "pressures" in the weekly grade. Name one way that reward could make the defense worse, and the rule you'd add to prevent it.
Answer
Rushers chasing pressure credit take the widest, fastest arc to the quarterback and abandon rush-lane integrity — the defensive end runs past the launch point and the quarterback steps up and escapes, which is how Rollins broke contain last season against Oklahoma City on a third-and-9 Star Fire and Oduya ran for 14. When a measure becomes a target, people bend their behavior to hit the measure (economists call this Goodhart's law). The rule: a pressure only counts in the grade if the rusher kept his lane — "never deeper than the quarterback" — and a lost contain on a scramble is a minus that cancels the pressure.
34.5 Coverage Efficiency: What Each Coverage Gives Up — and Why the Season Table Lies
Every coverage has a weakness. The analytics question is how much each weakness is actually costing you, against whom, and in what situations. Owen's coverage table is the most argued-over page in our building. Here it is, for last season (fictional; every dropback assigned to the coverage actually played after the snap).
| Coverage family (post-snap) | Dropbacks | EPA per dropback | Success rate | 20+ yard passes | Pressure rate |
|---|---|---|---|---|---|
| Cover 3 family (Sky, Spin 3, Buzz, Cloud) | 148 | +0.03 | 44% | 12 | 31% |
| Quarters and Six | 143 | −0.05 | 40% | 6 | 31% |
| Cover 1 family (incl. man-free pressures) | 108 | +0.02 | 43% | 11 | 34% |
| Fire zones (five- and six-man) | 96 | −0.09 | 37% | 3 | 44% |
| Two (plain Cover 2) | 43 | +0.06 | 46% | 2 | 28% |
| Two Man | 40 | −0.12 | 33% | 0 | 33% |
| Zero | 10 | −0.15 | 30% | 0 | 60% |
| Other (Two Pole, prevent, goal line) | 2 | — | — | 0 | — |
| All | 590 | −0.02 | 41% | 34 | 34% |
Before you rank anything, notice the rows that already match what earlier chapters told you. The Cover 1 family gave up 11 explosive passes, the most per dropback of anything we play — the crossers, rubs, and scrambles of Chapter 15. The Cover 3 family gave up 12, ten of them on flood/sail, dig, and four verticals, the concepts Chapter 9 charted. Plain Two gave up 2, both hole shots. Quarters gave up the fewest explosives per dropback of any base coverage, which is what a four-deep coverage is for.
Now the trap. If you sorted that table by EPA, you'd conclude that Two Man is our best coverage and that we should play it more. Our staff almost did. Owen stopped us.
Selection: coverages aren't called at random
A coverage's season numbers are the product of the coverage and the situations you called it in and the offenses you called it against. Two Man is a third-and-long call. On third-and-long, almost every defense posts good numbers, because the offense's most common outcomes — an incompletion, or a completion short of the sticks — all end in a punt. Zero is a goal-line and short-yardage call with ten snaps behind it; it's a story, not a number. Plain Two lives on first-and-10 and in the red zone. Comparing those rows with each other is like comparing a closer's earned-run average with a starter's.
The fix is to compare coverages inside the same situation. Owen pulled every 3rd-and-7-to-10 snap from last season — the same bin Chapter 26 uses — and the picture flipped.
| 3rd-and-7 to 10, last season | Snaps | Conversions | Conversion rate |
|---|---|---|---|
| Two Man | 18 | 6 | 33% |
| Cover 1 family | 9 | 2 | 22% |
| Quarters and Six | 10 | 3 | 30% |
| Fire zones | 18 | 3 | 17% |
| Cover 3 family | 3 | 1 | — (a story) |
| Runs faced (draws) | 4 | 0 | — |
| All 3rd-and-7 to 10 | 62 | 15 | 24% |
The coverage that looked best on the season table was the worst coverage we played on third-and-7 to 10. Its season EPA was carried by the 3rd-and-11+ snaps, where it really is good. Eighteen snaps is only a lean, so Owen didn't stop there — he went to the film for the reason. The reason was on every clip.
The mechanism: trail technique concedes the sticks
In Two Man, five underneath defenders play man with trail technique — they line up inside and under their man and chase from a step behind, because two deep safeties are protecting everything over the top. Trail technique is designed to concede the short throw and take away everything deep. On 3rd-and-12, the short throw is a punt. On 3rd-and-8, the short throw is the first down. Here's the snap Owen showed the staff — last season against Oklahoma City, fourth quarter, 3rd-and-8 at the OKC 34.
PRE-SNAP: "DIME OVER TWO MAN" vs. OSPREYS 11 PERSONNEL, 2x2
LAST SEASON, 4TH QUARTER, 3RD-AND-8 AT THE OKC 34
F $
(13 yds) (13 yds)
C * D C
M
E N T E
X H O O O O O Y Z
Q R
📊 Diagram (described): The Ospreys have 11 personnel in a two-by-two set: the split end (X) wide left with the slot receiver (H) inside him, five linemen with the center in the middle, the tight end (Y) attached on the right, and the flanker (Z) wide right. Oduya is in the shotgun with the back beside him. The Sentinels are in dime: two safeties at 13 yards, Royce (F) on the left and Holloway ($) on the right toward the tight end; corners Simmons (left, on X) and Brennan (right, on Z) about 7 yards off; Vo (*) over the slot; dime back Lucas (D) over the tight end; Bell (M) alone at linebacker depth in the middle; and four rushers — two ends outside, the nose, and the 3-technique toward the tight end. Eleven defenders: four linemen, one linebacker, six defensive backs.
POST-SNAP: TWO MAN -- TRAIL UNDERNEATH, HALVES OVER THE TOP
[DEEP HALF] [DEEP HALF]
F $
^ ^
| X go | Y seam
| |
| o<-------+ CATCH, +10 | v Z curl
C | D C
- - - - - - - - - | - - STICKS (8 YDS) - - - - - - - - - - - - -
|
* (trails a step) M (spies R)
H option
E N T E
Q R (stays in)
📊 Diagram (described): After the snap the two safeties retreat to deep halves: Royce on the left, Holloway on the right. Every underneath defender trails his man from inside and underneath. On the left, the split end runs a go route with Simmons trailing and Royce over the top. The slot receiver runs an option route: straight up the field to about 10 yards, then breaks out toward the left sideline; Vo trails a step behind and inside, and the ball is caught about 10 yards downfield, two yards past the first-down line drawn at 8 yards. On the right, the tight end runs a seam with Lucas trailing and Holloway over the top, and the flanker runs a curl at 12 yards with Brennan underneath. The back stays in to block, so Bell spies him and the quarterback. The four linemen rush; the ball comes out before they arrive.
Every defender did his job. Vo trailed exactly as taught. Royce and Holloway had nothing to do but watch. Oduya threw a 10-yard option route into the space trail technique gives away, and the Ospreys moved the chains. That's what 6 of 18 looks like on film: not busts, not bad technique — a coverage doing what it was built to do, in a situation where what it was built to do is the wrong thing.
The change was simple and you've already seen it. On our Week 15 third-down page in Chapter 25, Two Man is starred only on 3rd-and-11+, and the 3rd-and-7-to-10 box leads with Six. The number didn't make that decision. The number made us look at the film, and the film made the decision.
⚠️ Common mistake — "The coverage leaderboard." A staff sorts the season coverage table by EPA and installs more of the "best" coverage. The fix: never compare coverages across situations. Compare within a situation bin (down, distance, field zone, and personnel), and then demand the mechanism on film. If a coverage looks great only because of where you call it, it isn't great; it's well-placed. Keep it well-placed.
Man versus zone, by quarterback
Every game plan asks whether this quarterback is better against man or zone. The honest answer is that it depends on the quarterback, and the reasons are mechanical.
Owen split Vance's 124 attempts in the 4-game sample by coverage type.
| Vance, 4-game sample | Attempts | Completions | Yards per attempt | Scrambles on called passes |
|---|---|---|---|---|
| Against man (Cover 1, Two Man, Zero, man pressures) | 41 | 30 (73%) | 9.0 | 9 — for 81 yards and 6 first downs |
| Against zone (incl. fire zones) | 83 | 53 (64%) | 6.6 | 5 — for 22 yards |
| Total | 124 | 83 (67%) | 7.4 | 14 |
Why does man hurt us against Vance? Two reasons you can see on film. In man, every defender's back is turned to the quarterback, so when Vance escapes the pocket there's nobody facing him to rally — the 9 scrambles for 81 yards. And man invites Petrakis's rub, stack, and mesh answers (11 of 17 times when a defense showed man on third down, per Chapter 15). Zone keeps eyes on the quarterback and lets us rotate after the snap, which is where Vance's interceptions came from.
Now the opposite quarterback. Chapter 17's numbers showed Oduya completing 41 of 52 (79 percent) at 6.3 yards per attempt with no completions of 20+ against corners playing off, and 17 of 29 (59 percent) at 5.9 yards with three completions of 20+ against press. Off coverage gave up completions and denied explosives; press denied completions and gave up explosives. Which is "better" depends on what you fear more on that down, and that's a coaching decision, not a statistical one. On first-and-10, we'll take the 6-yard completion and tackle it. On 3rd-and-6, the 6-yard completion is the whole game.
📈 By the numbers: In recent seasons, most charting services have shown NFL defenses playing some form of zone on a clear majority of dropbacks, with man coverage on a minority that varies a lot from team to team. Exact splits differ by service and by how each one classifies match coverages, which look like zone before the snap and play like man after it (Chapter 31). Be skeptical of any single "man versus zone" number for a quarterback from a public source until you know how that source classified pattern-match quarters.
📋 Try it — Make the call (with numbers): Week 15, Austin, 3rd-and-8 at the Austin 40, 11 personnel 2x2, second quarter, tied. (A practice scenario, not a snap from the game.) You have three calls in the box. Owen's numbers for this bin and this opponent:
Call Our 3rd-and-7 to 10 rate, last season Vance, relevant row Nickel Over Six Quarters and Six: 3 of 10 Vs. 4 rushers with rotation: 31 of 52, 6.1 YPA, 3 INT Dime Over Two Man 6 of 18 Vs. man: 30 of 41, 9.0 YPA; 9 scrambles for 81 yards Nickel Over Mug Cat Fire Fire zones: 3 of 18 Vs. 5+ rushers: 28 of 36, 9.4 YPA, 0 INT Which do you call, and what would make you call the fire zone instead?
Answer
Nickel Over Six. It keeps a four-man rush (Vance holds the ball against four, so the rush has time), keeps eyes on him for scrambles, and gives him a post-snap picture to read — the row where he threw three interceptions. Two Man is out twice over: it's our worst 3rd-and-7-to-10 coverage, and it's man against a quarterback who scrambles against man. Mug Cat Fire has our best bin number (3 of 18), but that bin number is against everyone; against Vance the five-plus row is 28 of 36. You'd call the fire zone only as the one-in-five mix call from Chapter 25 — once per half at most, from the one-high look, in a game state where its sack-or-hot-window outcome is worth the risk. That's exactly how Anchor B was called in the third quarter of Game 2. The sample sizes (10, 18, 18) are all leans. Combine them with the mechanism, not with each other.
34.6 Explosive Plays: Where Defenses Actually Lose Games
The conventional wisdom says: if your defense allows explosives, the coverage is broken somewhere. That's half right, and the wrong half will send you fixing the wrong thing.
The book's definition: an explosive play is a run of 12+ yards or a pass of 20+ yards. Owen counts scrambles as dropbacks, so a 15-yard scramble is not explosive by this rule (it was a called pass and needed 20). Other analysts use other cutoffs — 10 and 20, 15 and 25, or a single threshold for everything — so, again, compare within one definition.
Last season we allowed 52 explosive plays in 17 games: 18 runs and 34 passes. That's about 5 percent of the roughly 1,050 snaps we played, about 3 a game. Here are the 34 passes by coverage.
EXPLOSIVE PASSES ALLOWED (20+ YDS) BY COVERAGE FAMILY, LAST SEASON
Owen's charting (fictional). Each # = one play. Dropbacks in ( ).
Cover 3 family (148) ############ 12 8.1%
Cover 1 family (108) ########### 11 10.2%
Quarters / Six (143) ###### 6 4.2%
Fire zones (96) ### 3 3.1%
Two (Cover 2) (43) ## 2 4.7%
Two Man (40) 0 0.0%
Zero (10) 0 0.0%
---------------------------
34 on 590 dropbacks = 5.8%
📊 Diagram (described): A horizontal bar chart of the 34 explosive passes the Sentinels allowed last season, one mark per play, grouped by the coverage played. Cover 3 family: 12 explosives on 148 dropbacks, 8.1 percent. Cover 1 family: 11 on 108, 10.2 percent. Quarters and Six: 6 on 143, 4.2 percent. Fire zones: 3 on 96, 3.1 percent. Plain Cover 2: 2 on 43, 4.7 percent. Two Man: none on 40. Zero: none on 10. Total: 34 explosive passes on 590 dropbacks, 5.8 percent.
That chart tells you where explosives happened. It doesn't tell you why. For that, Owen charts the first cause of every explosive: the earliest thing that went wrong, the one that, if fixed, would most likely have prevented the play. It's the same rule Chapter 5 used for the 18 explosive runs.
| First cause | Runs (12+) | Passes (20+) | Total | Fixed by |
|---|---|---|---|---|
| Run-fit error (wrong gap, squeeze, wrong key) | 11 | — | 11 | Fit rules, eyes, keys (Ch 5, 20, 21) |
| Sound fit, then a missed tackle | 4 | — | 4 | Tackling (Ch 23) |
| Missed tackle after a catch short of 20 | — | 6 | 6 | Tackling, rally (Ch 23, 26) |
| Good blocking, sound defense | 3 | — | 3 | Nothing — credit the offense |
| Communication bust | — | 3 | 3 | Checks and echo (Ch 8, 14) |
| Safety false trigger on a run fake | — | 5 | 5 | Eye discipline (Ch 22) |
| Lost in man (beaten, no technique error by the call) | — | 9 | 9 | Technique, matchups, leverage (Ch 15, 17) |
| Concept beat the call (the coverage's known weakness) | — | 8 | 8 | Call selection, disguise (Ch 9–12, 25) |
| Scramble / rush-lane break | — | 3 | 3 | Rush lanes, scramble rules (Ch 33) |
| Total | 18 | 34 | 52 |
Add up the rows that are "the coverage is broken" in the narrow sense — lost in man (9) and concept beat the call (8). Seventeen of 52. A third. The other two-thirds were fits, tackles, busts, eyes, and rush lanes. If I'd responded to 52 explosives by buying a new coverage, I'd have attacked a third of the problem with the most expensive tool in the building. Most of our improvement from my first season came from the cheap tools: communication, tackling, and fits.
💡 Coach's point: Explosives are the most valuable number on your dashboard because they're where the points are. A 45-yard run is worth more expected points than most offenses get from a whole series of 4-yard gains. But "explosives allowed" is an alarm, not a diagnosis. Every explosive gets a first-cause code before Monday's meeting, and the fix follows the code.
📋 Try it — Whose fault? (the analytics version): Code the first cause of Anchor C. Game 1, second quarter, 2nd-and-6 at the Sacramento 45. Inside zone to the tight-end side. Tavon Greer, a 9-technique (aligned on the outside shade of the tight end) with a "C-gap/contain" card, squeezed inside chasing the tackle's down block. Jordan Ames, the C-gap filler, filled B. Ruffin bounced into the crease outside both of them. Tre Holloway took a flat angle in the alley and missed. 45-yard touchdown. What's the first cause, what's the secondary code, and which numbers on the dashboard does this one play move?
Answer
First cause: run-fit error, by a lineman (Greer). He was the earliest defender to leave his responsibility; with his squeeze, the C-gap lost its outside wall, and Ames's wrong fit then opened the crease completely. Owen codes the first error by position, and this one is a defensive lineman — the same bucket as 5 of last season's 11 fit-error explosive runs. Secondary code: missed tackle, category "angle" (Holloway) — Owen charts it as one missed tackle, exactly as Chapter 23 recorded it. Coding the first cause matters because it points to the first fix: Greer's eyes and the "C-gap/contain" rule. Fixing Holloway's angle alone would likely have turned a 45-yard touchdown into a 10-to-14-yard run — still an explosive. On the dashboard, the play moves explosives allowed (+1 run), EPA per play (a big positive spike for the offense), and missed-tackle rate (+1 miss). It moves success rate no more than a 7-yard run would have; success rate is one yes-or-no per play. That's why you need both.
🔄 Check your understanding: Your explosives went from 61 one season to 52 the next. Give two reasons you shouldn't call the coverage change you made "the cause" yet.
Answer
First, you need the first-cause codes: if the drop came from fewer busts and fewer missed tackles, the coverage change didn't do it. Second, explosives are fairly rare events — 61 versus 52 is a difference of 9 plays across 17 games, and the square-root rule says counts in that range wobble by several plays from season to season on luck and schedule alone. You'd want the per-snap rate by coverage, the opponent quality, and the film before crediting the change.
34.7 Turnovers and Turnover-Worthy Plays: Habit, Luck, and the Quarterback Who Gives You Chances
We had 27 takeaways last season — 17 interceptions and 10 fumble recoveries by Owen's count. My players will tell you turnovers are a habit, not luck, because I've told them that every day since training camp of my first season. Owen will tell you that's half true, and he's right.
Here's the honest split. Some parts of a turnover are skill and habit. Some are a bounce.
| Part of a turnover | Mostly habit (coachable) | Mostly luck (noisy) |
|---|---|---|
| Creating a dangerous throw (disguise, pressure in the face, tight coverage) | ✔ | |
| Getting hands on the ball (break on the throw, track it, high-point it) | ✔ | |
| Catching the ball you got hands on | Partly | Partly — dropped interceptions are common |
| Attacking the ball on every tackle (strip, punch, "ball in every hand") | ✔ | |
| Whether the ball comes out on a punch | Partly | Partly |
| Who recovers a loose ball | ✔ — close to a coin flip in most analyses |
That last row is the one coaches hate. Public analysts have looked at fumble recoveries for years, and the common finding is that recovering a loose ball is close to a coin flip once it's on the ground, with little carryover from one season to the next. Last season opponents fumbled 21 times by Owen's count (14 of them forced by a strip or punch, the habit), and we recovered 10 — a little under half, right where the coin usually lands. Owen's projection for this season wasn't "10 again." It was "about half of whatever we force." So we worked on forcing: our ball attacks per game — strips, punches, rakes at the ball after the wrap — rose from about 7 to about 11 (Chapter 35 tells that story).
Interceptions are less of a coin flip, but they're noisy too, because they're rare. That's why Owen charts the thing upstream of the interception: the dangerous ball.
Turnover-worthy plays and Owen's "dangerous ball"
PFF charts a stat it calls turnover-worthy plays (TWP): throws its charters judge to have a high chance of being intercepted, and ball-security mistakes likely to cause a fumble, whether or not a turnover actually happened. It's a judgment stat — two charters won't always agree — but the idea is sound, and Owen's version for our own film is simpler: a dangerous ball is any throw a defender had a realistic play on — thrown into a defender's zone or lane, behind a receiver into a trailing defender, or late over the middle — whether it was caught, dropped, or picked.
Last season opponents threw 34 dangerous balls against us. We intercepted 15 of them. Our other 2 interceptions came on tipped balls off good throws — the kind of luck that comes from having eleven hats near the ball. Fifteen of 34 is about 44 percent. That rate wobbles. The 34 is the habit.
Now apply it to a quarterback. Owen charted Vance's dangerous balls in the 4-game sample:
| Vance, 4-game sample | Attempts | Dangerous balls | Interceptions |
|---|---|---|---|
| Vs. 5+ rushers | 36 | 1 | 0 |
| Vs. 4 rushers, post-snap rotation | 52 | 6 | 3 |
| Vs. 4 rushers, shell held | 36 | 2 | 1 |
| Total | 124 | 9 | 4 |
Against pressure, Vance threw one dangerous ball in 36 attempts. He wasn't careless against the blitz; he was precise, because the blitz handed him a pre-snap answer and a short throw. Against rotation, he threw six dangerous balls in 52, because what he saw before the snap wasn't what he got after it. That's the mechanism, and the mechanism matters more than the three interceptions, because — as Chapter 24 flagged in its sample hygiene — two of those three came in a single game against an opponent that rotated a lot like we do. Three interceptions is a story. Six dangerous balls against rotation, with a reason, is a lean worth building a disguise around. It's the lean we cashed in on 3rd-and-9 in the playoff game, Royce sitting in the deep middle where Vance thought nobody would be.
The decision rule Owen writes at the top of the opponent page:
- High dangerous-ball rate (a quarterback who throws into tight windows or late over the middle): aggressive coverage is rewarded. Robbers, traps, jumped routes, and disguised rotation — the rewards are interceptions.
- Low dangerous-ball rate (a veteran who gets the ball out on time to the right place, like Oduya): aggression costs you more than it pays. The quarterback won't give you the ball, so a jumped route is just a missed tackle with extra steps. Play sound, rally, tackle, and win third down.
🩺 Player safety: "Attack the ball" is a tackling skill, not a license. The strip and the punch come after the ballcarrier is wrapped and secured by the first defender — "first man stops, second man attacks the ball" (Chapter 23). A defender who leads with a swing at the ball instead of a wrap misses the tackle, and a defender who leads with his helmet to "dislodge" the ball is committing a foul that endangers both players. Count strip attempts in practice only when the tackle was secure.
🔄 Check your understanding: Your defense had 34 takeaways last season, with 21 fumble recoveries on 28 opponent fumbles and 13 interceptions on 30 dangerous balls. What do you predict for next season, and which number do you coach?
Answer
Expect fewer. Recovering 21 of 28 fumbles (75 percent) is well above the near-coin-flip most analysts find, so the recovery count will likely fall toward half of whatever you force. The 30 dangerous balls and the number of forced fumbles are the habit — coach those (disguise, pressure up the middle, break on the ball, strip after the wrap). A staff that celebrates 34 takeaways without looking upstream will be confused by 24 next year, when the habit didn't change at all.
34.8 The Autopsy: My Cover 0 Call Against Owen's Numbers
You know the play. Chapter 16 is its home, and it graded every player. This section grades the decision with the tools of this chapter, the way Owen did on the receipt. I'm going to walk it the way Owen walks new staff through it, because it's the cleanest example I have of a coordinator overriding his data and paying for it.
What I knew before the snap
| Input | Value | What kind of evidence |
|---|---|---|
| Vance vs. 5+ rushers | 28 of 36, 9.4 YPA, 4 TD, 0 INT | A plan-sized count (36), with a mechanism |
| Vance vs. 4 rushers with rotation | 31 of 52, 6.1 YPA, 1 TD, 3 INT | A plan-sized count (52); the INTs are a story |
| Austin kept the back in on 3rd-and-7+ | 19 of 27 dropbacks (about 70%) | A plan-sized count with an obvious reason |
| Protection ledger for six rushers vs. back-in | 6 rushers, 6 blockers, 0 wasted, 0 free | Arithmetic (Chapter 18) — not a sample at all |
| Game plan line, in my handwriting | "No zero pressure unless they're in 12 personnel inside the 10" | Tuesday's decision, made with a clear head |
| Game state | Tied 27–27, 4:12 left, Austin 3rd-and-8 at its own 48 | A stop was enough to give our offense the ball with a chance to win |
| Gabe on the headset | "Not zero, Russ. Not with Cole on Eleven. Give me a post." | Expert judgment, matching the data |
Look at the evidence column. It's not one fuzzy number. It's three plan-sized counts, an arithmetic certainty, a pregame decision, and an expert all pointing the same direction. When independent kinds of evidence agree, the fuzziness of any one of them stops mattering.
What I told myself
I've confessed the reasoning in Chapter 16: the four-man rush hadn't touched Vance since the third quarter; he'd converted three third downs on his last two drives against two-high; we hadn't called zero on third-and-long all season, so he wouldn't expect it; I kept Vo on Powers to take away the hot and tagged Lurk to take away the shallow. And the honest one: "I was tired of watching him do it."
Owen took each one apart with a number.
"The four-man rush hadn't touched him." Our pressure rate in Game 1 was about 22 percent by Owen's charting — 7 of 32 on four-man rushes, and 2 of 9 on blitzes. The blitzes weren't getting there either. Star Fire went three-for-three for Vance on third down that day. The rush being cold was a reason to change the rush — a game, a twist, a different launch-point plan — not a reason to send six against a protection built to block six.
"He'd converted three third downs against two-high." Three snaps is a story. And the alternative wasn't "more of the same"; it was a rotation he'd thrown three interceptions against in the scouting sample.
"He wouldn't expect zero." That's the best argument I had, and it's a real one: a call's value includes surprise (Chapter 25's mixing logic). But Austin's protection rule didn't depend on expecting zero. Petrakis's answer to a double-A-gap mug on third-and-long was "six-man, back on the Mike" whether or not he expected it, and Owen had that in the 19-of-27 number. Surprise doesn't help you against a rule.
"I was tired of watching him do it." There's no number for that one. There's only a rule: when the reason is a feeling, the call goes back to the card.
The expected-value picture
Here is Owen's whiteboard from the Chapter 16 meeting — the illustrative per-100 table of third-and-8s at midfield — with a value attached to each outcome, using the rounded EPA values from 34.2. Every number here is illustrative, a teaching sketch of the shape, not league data and not ours.
| Per 100 third-and-8s (illustrative) | EPA each (to the offense) | Four-man, two-high, rotation | Six-man Cover 0 |
|---|---|---|---|
| Ordinary stops (incompletion, short completion) | −1.3 | 51 → −66.3 | 40 → −52.0 |
| Sacks | −1.7 | 6 → −10.2 | 14 → −23.8 |
| Takeaways | −3.5 | 3 → −10.5 | 4 → −14.0 |
| Ordinary conversions | +1.8 | 35 → +63.0 | 29 → +52.2 |
| Gains of 20+ (not touchdowns) | +3.5 | 4 → +14.0 | 7 → +24.5 |
| Touchdowns | +6.0 | 1 → +6.0 | 6 → +36.0 |
| Total per 100 snaps | −4.0 | +22.9 | |
| Per snap | about −0.04 | about +0.23 |
Check the arithmetic the way Owen makes his interns do it: four-man stops add to 60 (51 + 6 + 3) and conversions to 40 (35 + 4 + 1); zero stops add to 58 and conversions to 42 — the same totals as Chapter 16's table. Multiply each count by its value, add the column, divide by 100.
The four-man rush comes out about −0.04 per snap — slightly good for the defense. Zero comes out about +0.23 — good for the offense. The difference, about a quarter of a point per call, is the second row of the receipt. Notice that zero's extra sacks were real: 14 against 6 is worth about 13.6 points per hundred snaps in our favor. Its extra touchdowns were worth 30 points per hundred against us. The sack column is what you remember on the bus home from the games it works. The touchdown column is what the season remembers.
And that's an ordinary quarterback. Owen's adjustment for Vance moved snaps from the "stop" rows into the "conversion" rows — his 78 percent against pressure versus about 60 against rotation is the whole story — and widened the gap. He didn't put a precise number on that adjustment, and he was right not to; 36 attempts can tell you the direction and not much more about the size.
The game-state layer
Expected points treat every snap as if the score didn't matter. Late in a tied game, it does. Win probability puts the clock and the score back in. Owen didn't quote me a model's number, because different models give different answers for a specific game. He drew it as a picture: tied with four minutes left, we were near a coin flip. A stop would have tilted it slightly toward us, with our offense getting the ball. A touchdown would have tilted it hard toward Austin — down seven with about four minutes left, we'd need a touchdown drive just to tie. Zero added a little to our chances on the good outcomes and took a lot away on the bad ones. Volatility is a tax when you're tied or ahead. I paid it.
🎙️ On the headset (the version I should have heard): Third-and-8, their 48, tied, 4:12. Tess: "Twenty-eight of thirty-six against five-plus. Back stays in nineteen of twenty-seven." Gabe: "Give me a post." The card in my hand says no zero outside 12 personnel inside the 10. The rush is cold — that's a rush problem, not a coverage problem. A stop wins the down; a stop is enough. Call: Nickel Under Tex Six — four-man rush, Coleman and Rollins on the twist — give the rush a new picture, make Vance hold it, make him read a rotation he's thrown into before. If he checks it down short of the sticks, we tackle it and punt. That's the call. It's the call on the card.
Resulting, and why Owen's receipt has two rows
Poker players have a word for judging a decision by its outcome: resulting. (Annie Duke's book Thinking in Bets popularized the term outside poker.) If Coleman had gotten to Vance half a second sooner, the same call would have been a sack, a punt, and a newspaper story about my guts. The decision would have been exactly as bad. That's what Owen meant on Tuesday: bad decision, good outcome is the most dangerous thing in football, because it teaches you to make the bad decision again.
The reverse matters just as much. Mug Cat Fire in Game 2 — Anchor B — was a good decision: one call in five, from a different look, with Coleman in the hot window and a sack-or-punt downside. If Vance had found Powers for 6 and Austin had converted on fourth down, it would still have been a good decision, and we'd have called it again in the right box. A staff that grades results instead of decisions will throw away good calls after bad bounces and keep bad calls after good ones. Both errors compound.
💡 Coach's point: Grade every big call twice: once on the film (what happened, who did what) and once on the card (what you knew, what you chose, and what the expected outcome was). Write both grades down. Over a season, the second grade is the one that predicts whether you keep your job.
What the numbers couldn't tell me
Owen insists I say this part, because he doesn't want anyone leaving this chapter thinking the numbers would have called the game. They couldn't tell me that Brennan would peek at the top of Gaines's stem. They couldn't tell me whether Coleman would beat the guard at 2.6 seconds or 3.1. They couldn't tell me what was on Petrakis's zero list that week; I learned the next spring that the max-protect deep over was third on it "because we didn't think you'd call it." They couldn't see that Big V was gassed on that drive, or that the sideline was loud, or what was in Vance's head. What they could tell me was the shape of the bet. I took a bet with a known shape and a bad expected value because the alternative felt passive. That's not an analytics failure. That's a coordinator failure, and the fix was a coordinator's fix: the zero card on every call sheet, and the one-word veto "Card" that Gabe, Tess, or Dre can say on the headset, after which I give a one-sentence reason or default to the written call.
🪞 Learning check-in: Stop and look at how you've been reading this chapter. Which number did you trust the moment you saw it, and which did you want to argue with? Most coaches trust the numbers that agree with what they already believe — the blitz-lover trusts the sack column and doubts the touchdown column. Pick one number in this chapter you wanted to disagree with, and write down the specific evidence that would change your mind: a sample size, a situation bin, a film clip. Then ask the opposite question about a number you liked: what would it take to convince you it was luck? If you can't name the evidence in either direction, you're not reading a number. You're reading your own opinion with a decimal point on it. Rate yourself 1 to 5 on each skill so far — reading EPA, separating blitz from pressure, spotting a selection trap, coding a first cause — and reread the section for your lowest score before you build the dashboard.
34.9 The Weekly Dashboard: Eight Numbers and the Decisions They Drive
You can drown in numbers. Every service sells hundreds. Owen's rule: if a number doesn't change a decision on Tuesday, it doesn't go on the front page. After four seasons together, our weekly defensive dashboard has eight numbers. Chapter 23 told you missed-tackle rate is one of them. Here are all eight, with last season's values, and the snapshot Owen handed me on the Monday after Game 1.
SENTINELS DEFENSE -- WEEKLY DASHBOARD (Owen Kessler; fictional)
WEEK 6 at AUSTIN: Outlaws 34, Sentinels 27 (1 game = a story)
# NUMBER GAME LAST SEASON TARGET
- ---------------------------- -------- ----------- -------
1 EPA/play allowed +0.16 -0.05 < 0.00
pass | run +.24|+.01 -.02|-.08
2 Success rate allowed 49% 40% < 42%
3 Explosives allowed (R12/P20) 5 (2R 3P) 52 (18R 34P) <= 3/gm
4 Pressure rate 22% 34% >= 32%
4-man | 5+ 22%|22% 31%|42%
5 3rd-down conversions allowed 8 of 14 36% < 38%
6 Red-zone TD rate allowed 2 of 3 48% < 50%
7 Takeaways | dangerous balls 0 | 1 27 | 34 --
8 Missed-tackle rate 12% 9% < 10%
-------------------------------------------------------------------
WATCH (not graded): comm. busts 1 (Lock/Quarters split, Q1)
📊 Diagram (described): A one-page dashboard for the Week 6 game at Austin, which the Outlaws won 34–27, labeled "one game equals a story." It lists eight numbers with three columns: the game, last season, and the target. One, EPA per play allowed: plus 0.16 in the game (plus 0.24 on passes, plus 0.01 on runs) against minus 0.05 last season (minus 0.02 passes, minus 0.08 runs); target below zero. Two, success rate allowed: 49 percent against 40; target below 42. Three, explosives allowed: 5 in the game (2 runs, 3 passes) against 52 last season (18 runs, 34 passes); target 3 or fewer per game. Four, pressure rate: 22 percent against 34, with both the four-man and five-plus rates at 22 percent in the game against 31 and 42 last season; target 32 or higher. Five, third-down conversions allowed: 8 of 14 against 36 percent; target below 38. Six, red-zone touchdown rate: 2 of 3 against 48 percent; target below 50. Seven, takeaways and dangerous balls: 0 and 1 against 27 and 34 last season; no target. Eight, missed-tackle rate: 12 percent against 9; target below 10. A watch line at the bottom notes one communication bust, the first-quarter Lock and Quarters split.
Read that page the way I read it on Monday morning. The five explosives were the Harlan dig for 22 on the opening drive, Gaines's 41-yard post off the Lock bust, Ruffin's 45-yard run, Ruffin's 16-yard run to start the last drive, and Gaines's 52-yard touchdown. Three of the five were coded to us — a bust, a fit error, and my call. The pressure rate was cold on four-man rushes and on blitzes, which is the line that should have told me in the fourth quarter that blitzing wasn't the cure. The missed-tackle rate was three points above our standard. One game is a story, and Owen printed that on the page so I couldn't forget it. But the stories pointed at the same three fixes we'd already proven: communication, fits, tackling.
The eight, and what each one decides
| # | Number | Exact definition (Owen's) | Decision it drives | Trap |
|---|---|---|---|---|
| 1 | EPA per play allowed, pass and run | Average EPA of every snap, dropbacks and designed runs split | Where to spend practice time: pass game or run game | Opponent quality and garbage time; one game is noise |
| 2 | Success rate allowed | Share of plays with EPA > 0 (we also keep the 40/60/100 version for by-hand checks) | Whether we're losing the "every snap" fight or only the big ones | Compare only within one definition |
| 3 | Explosives allowed | Runs 12+, passes 20+ (scrambles count as passes), each with a first-cause code | Which tool to fix: fit, tackle, bust, eyes, or call | Treating every explosive as a coverage problem |
| 4 | Pressure rate, split by rush count, plus median time to pressure | Pressures (sack, hit, hurry before the release) ÷ dropbacks | Rush plan and pressure package; whether pressure is arriving before the ball | Comparing across charting services; ignoring time to throw |
| 5 | Third-down conversion rate allowed, by bin | Conversions ÷ third downs, in bins 1–3, 4–6, 7–10, 11+ | Third-down menus (Chapter 26) | Average distance to go hides everything; always bin it |
| 6 | Red-zone TD rate allowed | Touchdowns ÷ possessions reaching our 20 | Red-zone package (Chapter 27) | Small samples — a handful of trips a game |
| 7 | Takeaways and dangerous balls | Turnovers gained; throws a defender had a realistic play on | Takeaway drills; how aggressive the coverage plan can be | Celebrating recoveries, a near coin flip |
| 8 | Missed-tackle rate | Missed tackles ÷ tackle attempts, coded by type | Tackling periods (Chapter 23) | Graders who don't agree on "attempt" |
What's not on the front page matters too. Yards allowed per game isn't there, because it depends on how many plays the opponent ran and where their drives started, which depend on our offense and special teams. Points allowed isn't there, because it's the outcome these eight are meant to explain, not a lever. Sacks aren't there, because pressure rate is the steadier read. Communication busts sit on a watch line — we count every one, but by last season there were only 9 all year, so they live on the page as a tripwire, not a weekly grade. If busts come back (seven of last season's nine came on the road), they move to the front.
🏫 Any level: You won't have an expected-points model for a high school league; nobody has built one from enough games. You don't need it. A manager in the press box with a clipboard can chart five of the eight by hand every Friday night: success rate by the 40/60/100 rule, explosives (with a first-cause code from Saturday's film), third-down conversions by bin, red-zone touchdown rate, and takeaways. Missed tackles come from Saturday film; two coaches chart and must agree. Pressure rate needs film too — count "the quarterback had to move or throw early" — and in high school it's often the least useful of the eight, because many high school offenses throw quick and short by design. In college, hash marks 40 feet apart change where explosives happen; chart explosives by field and boundary, because the wide side gives up more of them.
Connecting the dashboard to the decisions
The dashboard isn't a report card. It's a set of triggers. Ours has a printed "if-then" line under each number. Two examples:
- If the four-man pressure rate is below 25 percent for two straight weeks against opponents who hold the ball, Big Lou and I rebuild the rush plan before we touch the pressure package (Chapter 19's rush-plan card first, Chapter 18's pressures second).
- If explosives exceed 4 in a game, every one gets a first-cause code before the Monday meeting, and the staff spends the first fifteen minutes on the most common code, not on the biggest play.
Analysts who want to build these numbers from public play-by-play — EPA, success rate, explosives, third-down bins, red-zone rate — can do all of it for every NFL team from the free nflverse data. DataField.Dev's companion title Professional Football Analytics and Visualization goes further into building those models; this book stays on the coordinator's side of the desk.
🔄 Check your understanding: Your dashboard shows pressure rate up from 28 to 36 percent over three weeks, but third-down conversions allowed are also up, from 33 to 41 percent. Name two explanations you'd check before deciding the pass rush is "working."
Answer
First, check time to pressure against time to throw: if the extra pressures are arriving after the release, they aren't changing throws (and the opponents may have been quick-game teams who throw short and convert anyway). Second, check how the pressure rose — if it came from more blitzing, look at what the blitzes cost in coverage, especially man-free pressures, and look at the third-down bins to see whether conversions came on hot throws against those blitzes. Also check rush lanes: a rising pressure rate with more scramble conversions is Goodhart's law in action. Three weeks is also a small sample; check the opponents.
Spaced Review
Answer each question before you read the reminder under it.
🔄 Check your understanding (Chapter 25): You're sorting a self-scout line: "On 3rd-and-7+, 4th quarter, within 8 points, we sent 5+ rushers on 7 of 8." Break, bait, or accept — and how does this chapter's data on Vance make the case stronger?
Answer
Break. It's a predictable tendency that hands the offense a free, reliable answer, and Chapter 25's mixing table shows that blitzing on almost every late long third down throws away the value of the mix. This chapter adds the mechanism: Vance's hot throw arrives at 1.0–1.3 seconds, before even a free A-gap rusher, so a known blitz against him buys neither pressure nor surprise. The fix is a pressure limit printed on the call sheet ("1 in 5 or fewer") and counted by Tess.
🔄 Check your understanding (Chapter 18): Write the protection ledger for the Week 6 Cover 0 call against Austin's back-in protection: rushers, blockers, wasted blockers, effective blockers, free rushers, and where the hot throw went.
Answer
Rushers 6; blockers 6 (five linemen plus Ruffin); wasted blockers 0; effective blockers 6; free rushers 0. There was no hot throw needed — every rusher was blocked, so Vance had about three seconds and threw the deep over to Gaines against man coverage with nobody deep. Blitzing is arithmetic, and the arithmetic said zero free rushers before the snap; Owen's 19-of-27 back-in number said that was the likely protection.
🔄 Check your understanding (Chapter 6, deep callback): Why does Vance throw more dangerous balls against a four-man rush with post-snap rotation than against a blitz?
Answer
Because the shell is not the coverage. Against a blitz, the picture before the snap is honest — he sees pressure, knows his hot answer, and throws it on time. Against rotation, the picture before the snap is a claim, and the coverage after the snap is different; his pre-snap decision is built on the wrong picture, which is when quarterbacks throw into defenders. The dangerous-ball counts (1 in 36 against pressure, 6 in 52 against rotation) measure exactly that.
Project Checkpoint: Defensive Metrics Dashboard
Where you are. In Chapter 33 you built your Mobile-QB / Screen / Play-Action Plan: spy and rush-lane rules, screen alerts, and the eye-discipline rules that keep your aggression from being turned against you. Several of those rules depend on numbers you haven't been tracking yet — scramble yards, explosives off play-action, missed tackles on screens. This checkpoint gives them a home.
This deliverable: your Defensive Metrics Dashboard — the eight numbers you'll track weekly, and why. One page for the dashboard, one page for definitions. Build it for your level.
- The eight numbers. Choose eight. You may use ours, but you must justify each one with the decision it drives. A number that drives no decision is cut, no matter how interesting.
- A definition for each. Write the exact counting rule: what counts, out of what, who charts it, from which film or data source. If you use a public source (nflverse play-by-play for EPA and success rate, a charting service for pressure or coverage), name it and note that you'll compare only within that source.
- A sample-size note for each. How many snaps per game feed this number? When does it become "a lean" and "a plan" (Owen's ladder)? Red-zone touchdown rate needs weeks; success rate is usable after a game or two.
- A target and an if-then trigger for each. The target is your standard. The trigger is the decision: "If X for two straight weeks, then Y."
- A "not on the front page" box. List three numbers you will not track weekly and why (yards per game, sacks, points allowed are common choices).
- One worked week. Fill in the dashboard for one real game you've charted (your team, or any game you can watch) — or use the Week 6 numbers from 34.9 if you're working from this book alone.
A filled-in row, at the high school level:
| # | Number | Definition | Source / charter | Sample note | Target | If-then |
|---|---|---|---|---|---|---|
| 3 | Explosives allowed | Runs 12+, passes 20+; scrambles count as passes. Each coded with a first cause: fit, tackle, bust, eyes, lost in man, concept beat call, rush lane | Press-box manager marks the play Friday; DC codes the cause from Saturday film | About 50–60 snaps a game; one game is a story, 4 games is a lean | 2 or fewer per game | If the same first-cause code shows up 3+ times in 3 games, the first 15 minutes of Monday's practice go to that fix |
Use your answer from 34.4's productive struggle and 34.5's selection trap as the test of your definitions: could someone else chart your number and get the same count? Could a coverage look good on your dashboard only because of where you called it? Fix the definition until the answer is no.
What's next. In Chapter 35 you'll write your Culture Document — standards, effort grading, and the takeaway circuit. Two of your eight numbers (takeaways with dangerous balls, and missed-tackle rate) are the measuring sticks for that document. Bring them.
Chapter Summary
The numbers a defensive coordinator needs, what each one measures, and the trap built into each:
| Metric | What it measures | Sentinels last season (fictional) | Use it to decide | The trap |
|---|---|---|---|---|
| EPA per play allowed | Average change in the offense's expected points per snap | −0.05 (dropbacks −0.02, runs −0.08) | Where to spend time: pass or run | Opponent quality; one play's EPA judges the result, not the call |
| Success rate allowed | How often the offense "won" the snap (EPA > 0, or 40/60/100) | 40% | Every-snap consistency | Mixing definitions |
| Blitz rate | How often you sent 5+ | 27% (159 of ~590) | Pressure package size and mix | Confusing it with pressure |
| Pressure rate | How often the QB was pressured before the release | 34% (four-man 31%, 5+ 42%) | Rush plan; whether blitzes pay for their holes | Cross-service comparisons; ignoring time to throw |
| Time to pressure vs. time to throw | Whether pressure arrives before the answer | Blitz median 2.1 s, four-man 2.6 s | Blitz vs. coverage against a specific QB | Averages hide the hot throw |
| Coverage efficiency | EPA, success, explosives by coverage | Fire zones best overall; Two Man best overall but worst on 3rd-and-7–10 | Which coverage in which box | Selection: compare inside a situation bin |
| Explosive plays | Runs 12+, passes 20+, first-cause coded | 52 (18 runs, 34 passes) | Which tool to fix | Blaming the coverage for fits, tackles, and busts |
| Dangerous balls / TWP | Throws a defender could have played | 34 dangerous, 15 picked; 17 INT, 10 fumble recoveries | How aggressive the coverage plan can be | Treating recoveries as skill |
Decision framework — before you let a number change a call:
- Definition. What exactly was counted, out of what? Is it the same ruler as the number you're comparing it with?
- Size. How many? Story, lean, or plan? How fuzzy (one over the square root of the count)?
- Situation. Was it measured inside the situation you're deciding about, or across situations that flatter it?
- Mechanism. What's the reason on film? A count plus a reason is a tendency; a count alone is a rumor.
- Decision versus result. Are you grading what you chose or what happened? Grade both, separately.
- Game state. Late and close, does this call add variance you need, or variance you're paying for?
What's Next
Numbers tell you where the leaks are; they don't make eleven people run to the ball. Chapter 35 is about the part of defense no dashboard can install — culture: the standards, effort grading, accountability language, and takeaway circuit that moved our missed-tackle rate from 14 percent to 9 and produced 27 takeaways. Owen's numbers will be there, but this time they'll be measuring people, not calls. Start with Chapter 35.