Chapter 34 — Instructor Notes

What this chapter is for

Chapter 34 teaches analytics as a way of separating what happened from what you chose. Lead with that sentence, because it is what keeps the chapter from becoming a glossary of metrics. The hook is Owen's receipt, left on Russ's desk at 12:40 a.m. on the Wednesday after the Week 6 loss. It is one page with two rows that matter. The result row says the Cover 0 call ("Nickel, Over, Mug, Mac Dog, Zero, Lurk" on 3rd-and-8 at the Austin 48, tied 27–27 with 4:12 left) cost about six expected points on one snap. The decision row says the call costs about a quarter of a point per snap on average, more after adjusting for Vance's 28 of 36 against five or more rushers. The game-state line says a stop was enough and Russ bought variance he didn't need. Bottom line: grade the decision, not the result. The decision was still wrong.

The arc: what a number is, with Owen's three questions, "how many?", the confidence ladder, and the square-root rule (§34.1); expected points and EPA (§34.2); success rate (§34.3); pressure, blitz rate, and time to pressure (§34.4); coverage efficiency and why the season table lies (§34.5); explosives and first-cause coding (§34.6); turnovers and dangerous balls (§34.7); the full autopsy of Russ's Cover 0 call (§34.8); and the eight-number weekly dashboard (§34.9). The math stays at the book's level: counting, rates, and expected values worked in plain English. Every Sentinels number is fictional. The public sources (nflverse, Next Gen Stats, PFF, SIS, FTN) are named honestly and never quoted for invented figures.

Teaching objectives

By the end of the session a student should be able to:

  1. Interrogate any defensive number with Owen's three questions (what did you count, out of what, what decision does it change) plus "how many?", and label it story, lean, or plan.
  2. Estimate the fuzziness of a percentage with the 1 ÷ √n rule (36 snaps is about ±17 points).
  3. Explain EPA and success rate in plain English, and place a defense on the two-number map (Smothering; Bend, then hit; Boom or bust; Death by a thousand).
  4. Distinguish blitz rate, pressure rate, and time to pressure, and explain why the blitz doesn't buy pressure against Vance (hot throw at 1.0–1.3 s versus a free A-gap rusher at 1.6–1.9 s).
  5. Detect selection bias in a season coverage table by re-binning by situation, and find the mechanism on film.
  6. Code explosives by first cause and match the fix to the code.
  7. Grade a big call twice, once on the film and once on the card, and build an eight-number dashboard with definitions, sample notes, targets, and if-then triggers.

Timing

In a 75-minute session: 10 minutes on the hook and the receipt table. 12 minutes on §34.1: three questions, the ladder, and the square-root rule, practiced on 25, 144, and 400 snaps from exercise A5. 10 minutes on §34.2–§34.3: EPA and success rate, using the illustrative EP ladder and one worked snap. 13 minutes on §34.4. Open with the 🧩 Productive struggle (31% four-man pressure versus 42% blitz pressure: should we blitz more?) and have students write and keep their answers. 15 minutes on §34.5, the Two Man trap (see the activity below). 15 minutes on §34.8, the expected-value table, ending on resulting.

In a 150-minute block: add 20 minutes on §34.6, coding Anchor C in the 📋 Whose fault? (analytics version) and then all 52 explosives by code. Add 15 minutes on §34.7, dangerous balls and recoveries near half. Add 20 minutes on §34.9 with the Week 6 dashboard card. Add 10 minutes for the 🪞 Learning check-in: which number did you want to disagree with?

What to cut when you are behind: §34.1's public-source table and §34.3's worked-drive table can go to reading. §34.7's dangerous-ball paragraph can shrink to the definition. Do not cut §34.5 or §34.8. The first is the most transferable analytics lesson in the book. The second is the chapter's moral argument, and it closes Russ's Cover 0 thread from Chapter 16.

Where students get stuck

Blitz more, because the blitz pressures more. Diagnostic: the §34.4 🧩, which students answer before the section. Intervention: the pressure table and the race chart. Four-man rushes got 133 pressures on 431 dropbacks (31%, EPA −0.02). Five-or-more rushes got 67 on 159 (42%, EPA −0.04). The difference in outcome is small, and it hides a split: fire zones paid and five-man man-free didn't (34% pressure, about +0.08 EPA). Then show the timing. Against a quarterback who throws hot in about a second, pressure that arrives after the release is a highlight, not a stop. Have students grade their own 🧩 answers.

Trusting the coverage leaderboard. Diagnostic: quiz item 6 and exercise C3. Intervention: the activity below. Two Man has the best season EPA (−0.12) and the worst 3rd-and-7-to-10 conversion rate (6 of 18, 33%). Teach the ⚠️ "The coverage leaderboard": a coverage that looks great only because of where you call it isn't great. It's well-placed.

Resulting. Diagnostic: quiz item 18 (would Mug Cat Fire still be a good call if Vance had completed it?). Intervention: the two-row receipt. If Coleman gets home half a second sooner, the Cover 0 call becomes a sack and a newspaper story about Russ's guts, and the second row doesn't change. Case Study 34.1 runs the same receipt on Mug Cat Fire: a Tuesday pre-receipt with a staff-estimate outcome table of about −0.44 EPA per call. That is a good decision that would have stayed good even if it failed.

Turnovers as skill. Diagnostic: exercise A7 and quiz item 9. Intervention: the Sentinels' 27 takeaways were 17 interceptions and 10 fumble recoveries. The 10 recoveries came from 21 opponent fumbles, 14 of them forced. Coach the upstream numbers: forced fumbles, hands on balls, and dangerous balls (15 of the 17 interceptions came on 34 dangerous balls). Expect recoveries near half. Case Study 34.2 shows Vic Maddox's 31 takeaways falling to 16 over three seasons as the bounces regressed.

Blaming the coverage for every explosive. Diagnostic: quiz item 8. Intervention: the first-cause codes. Of 52 explosives, only the "lost in man" (9) and "concept beat call" (8) codes are coverage problems in the narrow sense. The rest are fit errors, missed tackles, busts, false triggers, and rush lanes, and each has a different fix.

Whiteboard / film-room activity

The table that lies. Write two rows of the season coverage table on the board: Two Man, 40 dropbacks, −0.12 EPA, 33% success; Cover 3 family, 148, +0.03, 44%. Ask, "Which coverage should we play more?" Most rooms say Two Man. Then add the 3rd-and-7-to-10 bin: Two Man 6 of 18 converted, fire zones 3 of 18. Then draw the snap Owen showed the staff.

POST-SNAP: "DIME OVER TWO MAN" vs. OSPREYS 2x2, 3RD-AND-8
(last season, Q4, at the OKC 34)

      [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 both safeties retreat to deep halves, Royce on the left and Holloway on the right. Every underneath defender trails his man from inside and underneath. On the left the split end runs a go with Simmons trailing and Royce over the top. The slot receiver runs up to about 10 yards and breaks out. Vo trails a step behind, and the ball is caught about two yards past the line to gain. On the right the tight end runs a seam with Lucas, the dime back, trailing, and the flanker runs a curl in front of Brennan. Bell spies the back, who stays in to block. Four linemen rush. Eleven defenders.

What to ask. "Did anyone bust?" (No. Vo trailed exactly as taught.) "What is trail technique designed to give up?" (The short throw, in exchange for everything deep.) "On 3rd-and-12, what is a 10-yard completion?" (A punt.) "On 3rd-and-8?" (A first down.) "So why does Two Man look best on the season table?" (Selection. Its EPA is carried by 3rd-and-11+ snaps, where every defense posts good numbers.) "Eighteen snaps is a lean. What made it a plan?" (The mechanism on film, which was the same on every clip.)

What a strong answer sounds like: "The number didn't make the decision. It sent us to the film, and the film showed a coverage doing exactly what it was built to do in a situation where that's the wrong thing. So we keep Two Man and star it on 3rd-and-11+, and we lead the 7-to-10 box with Six."

Assessment notes

Assign: A2, A5, A6 (a 17-yard scramble is not an explosive under Owen's rules, because scrambles count as dropbacks), A10, B4 (the broadcast pressure-rate graphic), B5 (the Ospreys release clock), B6–B7 (first-cause coding), C3, C5 (the pressure trade), C7 (rare events), D1 (is a quarter of a point a lot?), D2 (sacks on the dashboard), D3 (Goodhart's law and scramble yards), and M5 (the 18 explosive runs, 11/4/3). E1, a public-data dashboard from nflverse, is the right capstone for an analyst section.

Quiz: items 1, 5, 6, 7, 8, 10, 12, 14, 16, and 18.

A strong Project Checkpoint (Defensive Metrics Dashboard) has eight numbers, each justified by the decision it drives. Each has an exact counting rule (what, out of what, who charts it, from which source, compared only within one source), a sample note placing it on the story/lean/plan ladder, and a target with a two-week if-then trigger. It includes a "not on the front page" box with three numbers excluded and the reasons, and one worked week. Apply the chapter's own tests: could someone else chart this number and get the same count, and could a coverage look good on it only because of where it was called? Common failures are listing sacks or yards per game without a decision attached, and writing a trigger that fires on one game, which is a story.

Any-level adaptation

High school staff: use the checkpoint's high school row (explosives coded by the press-box manager Friday and by the DC from Saturday film; two or fewer per game as the target). Skip EPA. Success rate and explosives with first-cause codes are chartable with no data service. Exercise D4 (no film exchange against one opponent, 140 snaps on another) is the right judgment item. Teach the §34.7 🩺 callout alongside any takeaway number. "Attack the ball" is a tackling skill, not a license. The first man wraps and secures the ballcarrier, and the second man attacks the ball. A swing at the ball instead of a wrap misses the tackle, and leading with the helmet to dislodge the ball is a foul that endangers both players. Count strip attempts only when the tackle was made with sound technique, head out and shoulder in.

College class: use exercise C1's illustrative EP ladder, and pair §34.8 with Chapter 16's point that volatility is a tool when losing and a tax when tied or ahead.

Analyst course: this is your chapter. Assign E1 through E3, and insist on Owen's habits: name the source, compare within one source, report sample sizes, and code first causes from film rather than box scores.

Connections

Backward: Chapter 16 (the Cover 0 call and the per-100 table); Chapter 18 (blitz types and the ledger); Chapter 19 (where pressure comes from); Chapter 24 (Vance's rows and sample hygiene); Chapter 25 (the mix line and self-scout).

Forward: Chapter 35 uses takeaways, dangerous balls, and missed-tackle rate as culture measures. Appendix H holds every definition. Appendix F holds the templates.