Chapter 28 — Instructor Notes

What this chapter is for

Chapter 28 teaches students to use numbers without being used by them. Marisol Vega's sentence to new coaches is the thesis: "I'm not here to make the decision. I'm here to make sure you know what you're deciding." The chapter gives students the working vocabulary (expected points, EPA, success rate, explosive rate, pass rate over expected, the break-even conversion rate) and then spends most of its length on the harder skill: knowing when the number should win and when it shouldn't.

The hook carries the whole lesson. Week 12 against St. Louis, third quarter, Austin up 13–10, Moose Odom short by "a long foot" on 3rd-and-2, and now 4th-and-1 at the Austin 43. Teo's booth read is "Bear. Eight. Mabry at three." Marisol says go, not close. Wade Harmon, a former defensive coordinator, asks the question students should carry out of the room: "What's the call if we go?" Lyle answers "Heavy, Check," Vance sneaks for a yard and a half, and the next snap is the Load Right boot to Brandt for 14. The model assumes a league-average fourth-and-1 play. Making that assumption true is the coordinator's job.

The chapter also shows Lyle on both sides of the evidence. He wins the "grade the family" argument in §28.4 (plain Inside, Slant/Glance and Inside PA charted as one ledger, Weeks 1–8) and loses the reset-run argument in §28.5 (7 of 25 successful on second-and-8-plus; Wade: "a punt with extra steps"). Seeing a coordinator lose to the data, and change his call sheet, teaches more than seeing him vindicated.

Teaching objectives

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

  1. Define EP and EPA in plain English and estimate the EPA of a snap from the illustrative ladder in §28.2.
  2. Grade snaps by success rate (40/60/100) and explosive rate (runs 12+, passes 20+), and explain why a dashboard needs both.
  3. Explain the selection trap using the Outlaws' play-action numbers (about 8.6 vs. 6.2 yards per dropback overall, 8.6 vs. 7.0 on early downs only) and compare like with like.
  4. Compute a fourth-down break-even for a described situation and say what conversion rate makes "go" correct.
  5. Apply "Name it, then price it" to an override: say whether an objection names something the model can't see and whether it changes the decision or only the call.
  6. Read a two-point situation by the score and use the chart in §28.8.
  7. Report small samples honestly: counts before percentages, a "This doesn't know…" line, and a question mark on perfect streaks.

Timing

In a 75-minute session: 10 minutes on the hook, stopping at Wade's question. 10 minutes on §28.2 (the EP ladder and two EPA examples). 10 minutes on §28.3, success rate and the third-down cliff. 20 minutes on §28.6: the 🧩 Productive struggle worked on paper before anyone sees the decision tree, then the break-even line (about 53%), then "The call that made the number true." 15 minutes on §28.7, "Name it, then price it," with the 📋 Try it — Make the call (the hypothetical 4th-and-2 at the opponent's 38 with Kalu at left tackle). Use the last 10 minutes on the Project Checkpoint.

In a 150-minute block: add 20 minutes on §28.4 and §28.5 together, the argument Lyle won and the one he lost, read as a pair. Add 15 minutes on §28.8, two-point decisions by score. Add 20 minutes on §28.9 and its 📋 Try it — Whose fault? (the max-protect plan built on "15 of 26"). Finish with 20 minutes on Case Study 1, the Week 10 fourth-and-1 at the Portland 18 that failed and was still right.

What to cut when you are behind: "What the public data says" in §28.4 and "Talking to players" in §28.9 go to reading. Shorten §28.8 to one rule: two-point decisions are driven by the score. Do not cut §28.6 or §28.7. The break-even arithmetic and the override test are what the Project Checkpoint asks students to produce.

Where students get stuck

"If it failed, the decision was wrong." This is the most common error and the one the chapter fights hardest. Diagnostic: quiz item 12 (true/false) and exercise D5 (the "spreadsheet coach" columnist). Intervention: Case Study 1. The Portland fourth-and-1 was graded on Monday in a fixed order: decision right, call reasonable, execution lost the rep, and only then the result. Pratt slanted into the right A gap across Lorusso's face; Talbert finished. Then tie it back to Chapter 25's threshold idea: a 70% play fails three times in ten.

"Analytics says always go." Students who have just learned the break-even sometimes stop thinking. Diagnostic: quiz item 6 and exercise D1. Intervention: the model's assumptions (exercise A8 asks students to name all five). The first is a league-average play. Wade's question tests it. If you can't name your call, the chart's GO is about some other team.

"My gut says otherwise" counts as a reason. Intervention: "Two kinds of gut" in §28.7. Pattern recognition can be named and priced: a backup tackle losing every rep is a probability change. Fear of blame can't. Step 4 of the decision framework on the key-takeaways card asks it directly: "Is the objection about the play, or about the blame?" Point out that most named objections change the call, not the decision. Kalu at left tackle sends you to Sprint Right, Snag, not to the punt team.

"Bigger percentage, more trust." Diagnostic: quiz item 13 (is a 29-for-29 streak more trustworthy than 23 for 29?). Intervention: the streak that ended on the Game 3 interception. A perfect streak may be real, or it may be the opponent letting you get good at the wrong test. Marisol's rule is "write 9 of 14, never 64 percent," and perfect streaks get a question mark.

"EPA grades the player." Intervention: the common-mistakes table. EPA attaches first to the call and the situation. The Game 2 sack and the Game 3 hitch +10 are both in §28.2 as EPA examples, and in each case film, not the number, says who lost the rep.

Whiteboard / film-room activity

Make the number true. Draw the hook's fourth-and-1 and stop before anyone says what the defense is.

4th-and-1, AUS 43, up 13-10: "Heavy, Under Wing Right, Check"
vs. the Stallions' short-yardage bear (eight in the box)

                            F          <- Carrow, single high
   C                                                        C
              S             M                 $  <- walked down
            E          T    N    T         E
                U   O   O   O   O   O   Y
  X                         Q               B
                            R

 Check menu: bear = sneak | anything else = duo |
             safety down + both backers inside = PA to the Y

📊 Diagram (described): Austin is under center in Heavy personnel, with Brandt (U) attached on the left, Harlan (Y) attached on the right, Fale (B) a step off the line as a wing outside Harlan, Sutton (X) split wide left, and Ruffin directly behind Vance. St. Louis shows its bear: a nose head-up on the center, two tackles on the outside shoulders of both guards, ends outside both tight ends, Mabry stacked about three yards behind the nose, the Sam on the left, the strong safety walked down on the right, Carrow alone in the deep middle, and two corners. Eight defenders are near the line. Below the picture is Austin's three-play Check menu.

What to ask, in order. First, the decision: put the §28.6 numbers on the board (punt about −0.2; convert about +1.7; fail about −2.3) and have pairs find the break-even before you reveal "about 53%." Second, the call: "The center and both guards are covered. Where are the double-teams? Where is the softest yard?" Third, the safety question: Vance's designed-carry budget is about three a game. Does a sneak count, and who decides? The chapter's 🩺 note on the sneak says to count it.

A strong answer sounds like: "Going wins if we convert more than about half the time, and fourth-and-1 converts a lot more often than that, so it's a go. But that's only true if we have a play for this front. Against a bear there's nobody to double and both A gaps are covered, so duo is a bad bet. The sneak is the fastest yard, and the guards' job is to slam the A gaps so nobody knifes through." Students who add that the sneak set up the next snap's boot have connected this chapter to Chapter 29.

Assessment notes

Quiz: 22 items, 60 points. Items 1, 2, 5, 6, 7 and 12 cover the core. Short answer 17 ("What's the call if we go?" as a test of the model's assumption) is the single best item to grade closely. Applied item 22 is the "15 of 26" aggregation trap and pairs with the §28.9 Whose fault? drill.

Exercises: assign A6, A8, B1, B5 (draw and solve the fourth-and-1 at your own 35), B9 (Twelve personnel, 16 runs in 22 snaps: a tell and a zero), C1 ("Four-down" on 3rd-and-2), C4 (down 7, 4th-and-3 at the opponent's 33, against man), D2 (design a test of Gus's claim that the reset run "keeps the edge rushers honest") and M1 (the Portland fourth-and-1 as a three-line Monday review). D3 weighs Vance's carry budget against a fourth-quarter sneak and is the chapter's player-safety assignment; require a written rule, not a feeling.

Project Checkpoint: Offensive Metrics Dashboard + Fourth-Down / Two-Point Decision Chart. Part 1 is a one-page, ten-line dashboard with four columns (Metric, This week as a count, Sample, This doesn't know…). Part 2 is a laminated grid of field zones by distance marked GO / LEAN GO / TOSS-UP / KICK, adjusted for the student's kicker and short-yardage rate, plus the two-point table. A strong submission follows the worked line, Own 40 to opp 40, 4th-and-1: GO. Call: "Heavy, Under Wing Right, Check"… Break-even about 53%; our season rate 9 of 12, and names the call that makes every GO box true. Common weak submissions: percentages with no counts, an empty "This doesn't know…" column, and a chart copied from the sample without adjusting for the student's own kicker. Grade the last one hard at the high-school level, where a 25-yard kicker changes most of the right side of the grid.

Any-level adaptation

High school staff: replace EPA with points per drive and hand-chart success rate with a manager. Exercise C6 asks for a chart for a kicker reliable inside 25 yards and a 31-yard punter. Build it as a staff, and you'll see the GO region grow. Two-point tries come from the 3 (check your state's code). Treat the sneak with the same care as any designed quarterback run: count it and teach the guards' A-gap fits at walk-through speed before it's ever live.

College class: use Case Study 2 (Super Bowl LII: the fourth-and-goal at the 1 before halftime and the fourth-quarter fourth-and-1 near midfield). Students compare a widely documented real decision with the illustrative ladder. The case attributes no quotes to real people, and your discussion shouldn't either.

Analyst / film course: exercise E1 (chart one team's season of fourth downs against a public model and write a "name it" line for every disagreement) is the best data assignment in Part IV. Require counts and the share of disagreements with a nameable reason.

Connections

Backward: Chapter 14 (the short-yardage package that makes the chart's GO true), Chapter 23 (the "15 of 26" split and question-marked streaks), Chapter 25 (a call is a bet on a distribution), and Chapter 26 ("the fourth-down answer belongs in the third-down call").

Forward: Chapter 29 (Marisol's boot table, a family ledger for a different run) and Chapter 38 (grading a system, not a play). Reference: Appendix H for EPA, success rate and fourth-down logic.