Appendix H — Offensive Analytics Reference
How to use this appendix
This is the page you keep open when Marisol Vega's Sunday-night dashboard lands in your inbox and you want to know what each line means, how it was counted, and how much to trust it. Every metric here is defined in plain English, computed by hand on a worked example, and paired with the thing it can't see. There's no code and no formula you can't do on a napkin: counting, percentages, and a little subtraction. Sections H.1 through H.8 define the metrics. H.9 is Marisol's honesty kit: her five rules, her sample-size rule of thumb, and her four-line slide. H.10 and H.11 are fourth-down and two-point logic. H.12 is our season dashboard, every number in one place. H.13 names the public data sources, and H.14 is a one-table reference card.
Two warnings before you start. First, every expected-points number in this appendix is illustrative — round numbers on a simple ladder so the arithmetic is easy. Public models differ from one another and from these by a few tenths here and there; learn the shape, not the decimals. Second, every Austin Outlaws, Sacramento Sentinels, or other team number here is fictional, from the season this book follows. League-wide figures are given as hedged magnitudes, never as precise citations. The chapter that owns this material is Chapter 28; the forms the numbers live on are in Appendix F.
Marisol's sentence to every new coach is the right frame for all of it: "I'm not here to make the decision. I'm here to make sure you know what you're deciding."
H.1 The Metrics at a Glance
| Metric | The question it answers | How it's counted | What it can't see |
|---|---|---|---|
| Expected points (EP) | What is this down, distance, and spot worth? | The average next score from many seasons of similar situations | Assumes league-average teams and plays |
| Expected points added (EPA) | What was this one snap worth? | EP after the snap minus EP before it | Noisy; one big play swamps a sample; credits the call and the player together |
| Success rate | Did the play keep us on schedule? | Share of plays that gain 40% of the distance on 1st down, 60% on 2nd, 100% on 3rd and 4th (or, by the other definition, EPA above zero) | Ignores explosives |
| Explosive rate | Can we skip the schedule? | Runs of 12+ and passes of 20+, per play or per game | The first thing defenses adjust to; volatile |
| CPOE | Does the quarterback complete more than the throws' difficulty predicts? | Actual completion rate minus a model's expected rate | Depends on the model; doesn't value the decision |
| Early-down pass rate | How often do we throw on 1st and 2nd down? | Passes ÷ early-down snaps, usually leaving out garbage time and two-minute | Doesn't say whether the situation called for it |
| PROE | Do we throw more or less than the situation predicts? | Actual pass rate minus a model's expected pass rate | Selection: you choose when to run |
| Third-down conversion by distance | Where is our cliff? | Conversions ÷ attempts in each distance band | Small samples per band |
| Tendency rate | What does a look tell the defense? | "X of Y," with a floor of about 70% on 10+ snaps for a "tell" | Opponents change on purpose; perfect streaks may be traps |
H.2 Expected Points (EP)
Expected points put a number on how valuable a situation is. Take every play in many seasons of NFL football that started from, say, first-and-10 at the opponent's 20. Follow each one to the next score — yours (positive) or theirs (negative), a touchdown, a field goal, a safety. Average those next scores. That average is the expected points of first-and-10 at the opponent's 20. Public models built from large play-by-play datasets refine this with down, distance, time, and more, but that's the core.
EXPECTED POINTS FOR 1st-AND-10, BY FIELD POSITION
(illustrative round numbers; every public model differs a little)
each # = 0.2 points
1st-and-10 at EP
own 5 -0.3 <|
own 20 +0.4 |##
own 30 +1.0 |#####
midfield +2.0 |##########
opp 35 +2.8 |##############
opp 20 +3.8 |###################
opp 10 +4.6 |#######################
opp 2 +5.6 |############################
📊 Diagram (described): A horizontal bar chart of illustrative expected points for first-and-10 at eight spots. Pinned at your own 5 the value is slightly negative, about −0.3, because the opponent is a little more likely to score next. It climbs to about +0.4 at your own 20, +1.0 at your 30, +2.0 at midfield, +2.8 at the opponent's 35, +3.8 at their 20, +4.6 at their 10, and about +5.6 at their 2.
How to read the ladder
| Reading | What it means for the call |
|---|---|
| Field position is worth points before anybody scores | Your 30 to midfield is worth about a point — the same as a made extra point |
| It isn't linear at the ends | The last ten yards are expensive: the field shrinks and the defense gets the end line as help |
| Down and distance matter as much as the spot | Third-and-9 at your 38 is worth far less than first-and-10 at your 38, because most of those snaps end in a punt |
| It's model-dependent | Know the shape; don't argue about the second decimal |
Worked example — reading a spot. You take a punt at your own 20 (about +0.4). Your returner breaks one to your 35. On the ladder that's halfway between your 30 (+1.0) and midfield (+2.0) — call it about +1.3. The return was worth about +0.9 points before your offense took a snap. That's why special teams coaches count hidden yards.
H.3 Expected Points Added (EPA)
EPA is the change in expected points from before a play to after it.
How to compute it, in three steps
- Before: look up the EP of the situation before the snap (down, distance, spot).
- After: look up the EP of the situation after the snap. If you scored, it's the points you got (minus what the opponent's next possession is worth to them). If you turned it over, it's the opponent's EP from their new spot, with the sign flipped — their +1.8 is your −1.8.
- Subtract: after minus before.
Worked examples from our season
All on the illustrative scale. Marisol's model gives slightly different decimals; the shape is the point.
| Snap (fictional Outlaws season) | Before | After | EPA (approx.) | The lesson |
|---|---|---|---|---|
| Game 1, Q4: "Inside PA, Gaines Post," 1st-and-10 at the SAC 39, Gaines +35 | ~+2.6 | 1st-and-goal at the SAC 4, ~+5.3 | ~+2.7 | One explosive is worth more than most whole drives |
| Game 1, Q4: inside zone, Ruffin +5 on 1st-and-10 at the AUS 49 | ~+2.0 | 2nd-and-5 at the SAC 46, ~+2.1 | ~+0.1 | A "boring" five on first down barely moves the needle — but it keeps you on schedule |
| Game 2, Q3: 3rd-and-7 at the AUS 38, the 9-yard sack | ~+0.7 | 4th-and-16 at the AUS 29 — a punt, ~−0.9 | ~−1.6 | A third-down sack kills a drive twice: lost yards and a lost down |
| Game 3, Q3: 3rd-and-7 at the AUS 41, Gaines hitch +10 vs. the mug | ~+0.8 | 1st-and-10 at the SAC 49, ~+2.0 | ~+1.2 | A ten-yard hitch on third down is worth more than a twenty-yard gain on first |
| Game 3, Q4: 3rd-and-9 at the AUS 38, the interception | ~+0.6 | Sentinels' 1st-and-10 at the SAC 46, ~−1.8 for us | ~−2.4 | A turnover costs the possession you lose and the one you hand over |
Worked in words — the Game 3 hitch. Before the snap: third-and-7 at our 41, worth about +0.8 (a lot of those end in punts). After: first-and-10 at the Sacramento 49, about +2.0. Subtract: about +1.2. Ten yards, more than a point.
Worked in words — the Game 1 run. Before: first-and-10 at our 49, about +2.0. After: second-and-5 at their 46, about +2.1. Subtract: about +0.1. Two snaps later, play-action off that same run, from the same picture, was worth about +2.7. Graded alone, the run was nearly worthless. Graded with the pass it set up, it was half a touchdown drive. (H.7 shows how we settled that argument.)
How to read EPA
| Use it for | Don't use it for |
|---|---|
| Valuing a call, a family of calls, or a situation over many snaps | Grading a player off one play |
| Comparing third-down and turnover outcomes with early-down gains | A small sample — one 60-yard play can make a bad day look good |
| Building fourth-down and two-point decisions (H.10–H.11) | Arguing about the second decimal |
⚠️ Common mistake: treating EPA as a grade for the player. The Game 3 interception was −2.4, and it went on Vance's line. But Vance read two-high before the snap, and the Sentinels rotated to single-high at the snap — the disguise they'd saved all season. The fix: attach EPA to the call and the situation first, then assign responsibility with film, not with the number.
📈 By the numbers: In public play-by-play data over many seasons, third-down results and turnovers swing expected points far more than individual early-down gains, and the biggest single-play swings belong to explosives and interceptions. That's why third down and turnovers get their own dashboard lines even though EPA already "contains" them.
H.4 Success Rate (Both Definitions)
Success rate asks a simpler question than EPA about every snap: did this play keep us on schedule? Two definitions are in common use. Always say which one you used.
Definition 1: the distance rule (the one we use)
| Down | A play is "successful" if it gains… | Examples |
|---|---|---|
| 1st | at least 40% of the yards to go | 4+ on 1st-and-10; 6+ on 1st-and-15 |
| 2nd | at least 60% of the yards to go | 3+ on 2nd-and-5; 6+ on 2nd-and-10 |
| 3rd and 4th | 100% — a first down or a touchdown | 7+ on 3rd-and-7 |
Definition 2: the EPA rule
A play is successful if its EPA is above zero — if it left you in a more valuable situation than it found you.
How to compute it
Grade every snap yes or no. Count the yeses. Write it as a count: "26 of 58," then the rate if the sample earns one. Leave out kneels and spikes; decide in advance whether penalties count, and keep the rule the same all season.
Worked example — three first downs (distance rule):
| Snap | Needed | Got | Successful? |
|---|---|---|---|
| Ruffin +3 on 1st-and-10 | 4 (40% of 10) | 3 | No |
| Powers +9 on 1st-and-15 after a penalty | 6 (40% of 15) | 9 | Yes |
| A 2-yard completion on 1st-and-goal from the 4 | 1.6 (40% of 4) | 2 | Yes |
Success rate scales with the distance, which is exactly why it beats raw yards per play as a measure of staying on schedule.
Worked example — a drive. Game 1's tying sequence: inside zone +5 on 1st-and-10 (needed 4: yes); inside zone +7 on 2nd-and-5 (needed 3: yes); play-action post +35 (yes); inside zone 4-yard touchdown (yes). Four snaps, four successes, and only one explosive. That's what a good drive looks like on a success-rate sheet: not a parade of big plays, but snaps that never let the defense choose the down and distance.
Where the two definitions part ways
They mostly agree. They disagree at the edges, and the edges are worth knowing:
| Situation | Distance rule | EPA rule |
|---|---|---|
| A gain of exactly 40% on first down | Success | Close to zero; can fall on either side depending on the model and the spot |
| A long-long third down that gains most of the distance and sets up a shorter punt or a makeable kick | Failure | May come out near zero or positive, because field position improved |
| A short gain deep in your own territory that gets you off the goal line | Judged only by distance | Often valued a bit more, because escaping the shadow of your own end zone is worth points |
🏫 Any level: You don't need a model for the distance rule. Have a manager chart 40/60/100 by hand on the sideline; after six games you'll know your own success rate by down, and your own fourth-and-short conversion rate, better than any league table can tell you.
Reading it. Success rate is steadier on small samples than EPA, because one long touchdown can't swamp it. That's also its blind spot: a 60-yard touchdown counts the same as a 10-yard gain on 3rd-and-9. Read it next to explosive rate.
H.5 Explosive Rate
Explosive play, as this book defines it: a run of 12 or more yards, or a pass of 20 or more yards. (Our companion book uses the same line. Other sources draw it elsewhere, so check before you compare.)
How to compute it
- Count every run of 12+ and every completion of 20+ (the yards on the play, including after the catch).
- Divide by the plays of that type, or by all plays, or by games — and say which.
- Keep your classification rules fixed: decide once whether a scramble counts as a run or a pass, and how to count a screen, and don't change either rule midseason.
Worked example — our season. 71 explosives in 17 games, a little over four a game. Examples from Game 1 alone: the Q2 45-yard run (a run of 12+), the 41-yard post to Gaines, the 35-yard play-action post, and the 52-yard zero-bin deep over (all passes of 20+). The +5 and +7 inside-zone runs that set up the 35-yarder are not explosives — they're successes.
How to read it
Success rate measures whether you can stay on schedule. Explosive rate measures whether you can skip the schedule. A good offense has both, and a good defensive coordinator will take one away and dare you to live on the other. Portland's Cover 2 made us live on success rate; Sacramento's rotating two-high shell was built to take away our explosives while keeping enough people in the box to lower our early-down success rate too.
Marisol's season split: of our drives that included at least one explosive, a little under half ended in a touchdown; of drives without one, about one in eight did. For most offenses, the explosive is the difference between a field goal and a touchdown.
⚠️ Common mistake: the explosive addiction. Explosives are the most valuable plays in football and the ones defenses adjust to fastest. Curtis Bale led his conference in explosives in his first year as a coordinator; the next year defenses played deep, and instead of living on schedule he kept calling for the big one. The fix: build the offense to survive on success rate, and let the explosives come from the plays the success rate sets up.
H.6 Completion Percentage Over Expected (CPOE)
CPOE asks whether a quarterback completes more passes than the difficulty of his throws would predict.
How it's built, in plain English
A model looks at a very large number of past throws and learns how often throws like this one are completed. "Like this one" means similar on the things the model can measure — commonly how far downfield the ball travels, where on the field it's thrown, and, in models built on player-tracking data, things like how open the receiver is and how much pressure the quarterback is under. The inputs differ by provider. Each throw gets an expected completion chance. Add those chances across a game or a season and you have expected completions.
CPOE = actual completion percentage − expected completion percentage.
Worked example (illustrative, not a real player)
| Step | Number |
|---|---|
| Attempts | 30 |
| Expected completions (sum of each throw's chance) | 19.5 → 65% |
| Actual completions | 21 → 70% |
| CPOE | 70% − 65% = about +5 points |
Read it as: on throws a typical quarterback completes about 65% of the time, this one completed 70%.
How to read it
| It's good at | It can't see |
|---|---|
| Separating accuracy from scheme: an offense full of easy throws has a high expected rate, so it doesn't get credit just for being easy | Whether the throw was the right one — a safe checkdown and a tight-window completion both count as one completion |
| Comparing quarterbacks across offenses | Drops, throwaways, and spikes, which most models treat in their own ways |
| Season-long trends | One game's worth of throws, which is a small and noisy sample |
The ten-line dashboard in Chapter 28 doesn't carry CPOE, because it measures the quarterback more than the call. If you add it, put it on the quarterback's page and read it beside sacks (by cause) and interceptions (by cause), which tell you about his decisions.
H.7 Early-Down Pass Rate, PROE, and the Selection Trap
Early-down pass rate
Early-down pass rate is the share of first- and second-down snaps that are passes. Most analysts leave out garbage time and two-minute situations, and most count a dropback that ends in a sack or a scramble as a pass, because a pass was called. Say what you left out.
Pass rate over expected (PROE)
PROE compares how often you passed with how often a model predicts an average team would pass in the same down, distance, field position, score, and time.
PROE = actual pass rate − expected pass rate. Positive means you throw more than the situation predicts; negative means you run more.
Worked example (illustrative):
| Step | Number |
|---|---|
| Early-down snaps (garbage time and two-minute removed) | 200 |
| Model's expected passes (sum of each snap's predicted pass chance) | 104 → 52% |
| Actual passes | 112 → 56% |
| PROE | 56% − 52% = +4 points |
Our season: early-down PROE "a little above zero" by Marisol's estimate — a few points more pass-heavy than an average team in our situations. In recent seasons the league as a whole has drifted toward throwing more on early downs, and toward running more efficiently when it does run.
The RPO problem
Is an RPO a run or a pass? Classify it by what happened and every correct throw becomes an efficient "pass" and every correct handoff into a box that just lost a defender becomes an efficient "run" — and the credit goes to the wrong family. Marisol codes RPOs as their own family. In the four-game sample the Sentinels charted, our 187 early-down snaps broke down: called runs 37%, RPOs 20%, play-action 21%, dropbacks and quick game 17%, screens 5%.
The selection trap
Coaches don't call plays at random. We run when we're ahead late and on third-and-1; we throw when we're behind and on third-and-9. Compare "all runs" with "all passes" and you're comparing two different sets of situations.
Worked example — play-action. Our play-action passes averaged about 8.6 yards per dropback; straight dropbacks about 6.2. But about four of every five play-action calls came on first or second down, while barely half of our straight dropbacks did — the rest were third downs, where yards per attempt drops for everyone. First and second down only: about 8.6 against about 7.0. Still better. Not two and a half yards better. Most of the gap had been the downs we chose.
The fix: compare like with like — same down, same distance band, same box count, same score state — and accept that the sample gets small when you do.
"Establish the run": grade the family
Some public analysts have argued from league-wide data that play-action's effectiveness doesn't depend much on how often or how well a team has been running. That challenge is real. The answer that survived in our building is narrower: the run doesn't earn play-action by volume; it earns it by picture and by the specific defender who is taught to stop it. We tested it by grading the whole family from one look.
THE INSIDE FAMILY FROM GUN DOUBLES -- 1st and 2nd down, Weeks 1-8
(Marisol's bye-week ledger; EPA per play on her model; each # = 0.05)
Member Calls Successful EPA/play
Inside (plain handoff) 58 26 of 58 -0.02 |
Inside Slant / Glance 31 18 of 31 +0.19 |####
Inside PA (same look) 22 12 of 22 +0.41 |########
-----------------------------------------------------------
The family 111 56 of 111 +0.12 |##
📊 Diagram (described): A table-and-bar chart of every first- and second-down call from the same Gun Doubles picture over our first eight games. Plain inside-zone handoffs: 58 calls, 26 successful, about zero expected points per play. The RPO members, Slant and Glance: 31 calls, 18 successful, about +0.19 per play. Play-action from the identical look: 22 calls, 12 successful, about +0.41 per play. The family together: 111 calls, 56 successful, about +0.12 per play.
Marisol's verdict: "You win the call sheet. You don't win the argument." The ledger shows the family is good and the runs don't drag it down much; it doesn't prove the runs made the passes better. The evidence that came closer was defender-level: Teo's card on Holloway's first step — downhill on 15 of 21 snaps on film, and four for four on our fourth-quarter sequence in Game 1. "That's better evidence than my table," Marisol said. "It's also twenty-five snaps of one safety."
The comfort call: the argument I lost
A comfort call is the play a caller reaches for after the last one went badly. Mine was the second-and-long "reset run."
| Weeks 1–8, 2nd-and-8 or longer | Snaps | Successful | Average 3rd-down distance after |
|---|---|---|---|
| Designed runs (not RPOs) | 25 | 7 of 25 | about 3rd-and-7 |
| Everything else (dropbacks, PA, RPO throws, screens) | 38 | 17 of 38 | about 3rd-and-5 |
| All 2nd-and-8+ | 63 | 24 of 63 |
After the bye the run became half of a Kill call (light box keeps it; anything else throws). Weeks 10–18: 68 second-and-long snaps, 14 runs, 30 of 68 successful. Marisol's caveat on the slide: different defenses, a better November quarterback, 68 snaps — "it moved the right way" is all she'd sign.
How to find your own comfort call: sort your self-scout by "the play I call most often right after a negative play," then check its success rate against the same down and distance when it didn't follow a negative play.
H.8 Third-Down Conversion by Distance
How to compute it
Sort every third down into a distance band and count conversions (a first down or a touchdown) over attempts in each band. Use our four words so the dashboard, the scouting report, and the call sheet line up:
| Band | Yards to go |
|---|---|
| Short | 1–3 |
| Medium | 4–6 |
| Long | 7–10 |
| Long-long | 11+ |
Our season (regular season, fictional)
| Band | Converted | Attempts | Rate |
|---|---|---|---|
| Short (1–3) | 38 | 57 | about two in three |
| Medium (4–6) | 30 | 62 | a little under half |
| Long (7–10) | 19 | 58 | about one in three |
| Long-long (11+) | 5 | 27 | fewer than one in five |
| All third downs | 92 | 204 | about 45% |
THIRD-DOWN CONVERSION BY DISTANCE -- Outlaws, regular season
(each # = about 5 percentage points)
Short 1-3 38 of 57 #############
Medium 4-6 30 of 62 ##########
Long 7-10 19 of 58 #######
Long-long 11+ 5 of 27 ####
📊 Diagram (described): A bar chart of third-down conversions by distance for the season. Third-and-1 to 3: 38 of 57, about two-thirds. Third-and-4 to 6: 30 of 62, a little under half. Third-and-7 to 10: 19 of 58, about a third. Third-and-11 or longer: 5 of 27, under one in five. Each bar is shorter than the one above it; the drop past six yards is the cliff.
The four-game sample the Sentinels charted before Week 6 (55 third downs, split slightly differently) showed the same cliff: 1–3, 10 of 14; 4–6, 13 of 26; 7–9, 2 of 9; 10+, 1 of 6 — 26 of 55 in all.
How to read it
Every yard of distance is a yard of freedom handed to the defensive coordinator. On third-and-2 the defense must defend the run, the quick game, and the sneak. On third-and-9 it can drop eight, sit at the sticks, and dare you — or bring a disguised pressure, because you have to hold the ball for a 9-yard route. That's why early-down success rate matters: it measures how much freedom you gave the other coordinator on third down.
Three reading rules:
- Judge early-down calls by the third down they leave. Marisol's second-and-long table (H.7) is the example: the run left third-and-7; everything else left third-and-5.
- Medium is the offense's down. In the four-game sample we were in three-by-one on 22 of 26 third-and-4-to-6 snaps, called dig or flood on 14, and converted 8 of those 14.
- On long-long, judge the call by what it costs, not only by the conversion rate. On 3rd-and-10 or longer, our 7 screens produced 2 conversions, no sacks, and no turnovers; our other 34 snaps at that distance produced 8 conversions, 4 sacks, and 2 interceptions. Marisol's note: "Seven snaps. Do not put the conversion rate on a slide." Two of seven and eight of 34 aren't different numbers at that sample size. What the table does say is that the screens never cost us the ball and never cost us a kick.
Taught in: Chapter 26 (third down by distance, "sticks plus one"); screens on third-and-long in Chapter 20.
H.9 Sample-Size Honesty
Football's samples are tiny. A team runs roughly sixty to seventy offensive snaps a game, split across dozens of situations; slice to "third-and-7 from two-by-two against a two-high shell" and you might have nine snaps in a season. Marisol wins arguments in our building because she loses some on purpose: she'll tell you when her number is too thin to use, and that's why you believe her when she says it isn't.
H.9.1 Marisol's five honesty rules
- Counts before percentages. Write "9 of 14," never "64 percent." The count tells you how much to trust the rate.
- Every number carries its sample and its caveat. A number without a sample size is a rumor.
- Say what the number can't see. Every slide has a bottom line that begins "This doesn't know…"
- Separate the decision from the result. A good decision can lose; a bad decision can win. Grade the process, then look at the result to learn something about the inputs.
- When you don't know, say "I don't know." If more than about one snap in ten in a column is tagged "?", she won't give it a percentage at all.
H.9.2 How far luck alone can move a rate
Marisol's rule of thumb for a success rate (or any yes/no rate near the middle):
| Snaps in the sample | The rate can wander, by luck alone, roughly… |
|---|---|
| 20 | about 20 points either way |
| 100 | about 10 points either way |
| 400 | about 5 points either way |
So a call that's "60% successful" on 20 snaps could honestly be a 40% call that got lucky or an 80% call that didn't. Under about 20 snaps, call it a lean, not a fact.
H.9.3 The tell threshold
A look is a tell when it predicts one play family about 70% of the time or more, on 10 or more snaps. Both parts matter. The floor isn't proof: 9 of 14 can happen by luck about one time in five.
Worked example. Man coverage shown on third down → we went stack, bunch, or mesh on 11 of 17. Marisol: "It's sixty-five percent. My line for a tell is seventy percent on ten or more snaps. It's not a tell yet. It's a lean." The Week 6 tells on the Sentinels — 14 of 18, 9 of 12, 17 of 22 — all cleared it.
H.9.4 Four traps
| Trap | What it looks like | Our example | The fix |
|---|---|---|---|
| Aggregation | A combined number that describes no actual opponent | "Pressure on 15 of 26 long third downs" — but 8 of 9 against rookie QBs and 7 of 17 against veterans | Split every tendency by the condition that matters most to your team; report counts so "7 of 17" can't hide inside "15 of 26" |
| Selection | Comparing plays called in different situations | Play-action 8.6 vs. 6.2 yards per dropback shrank to 8.6 vs. 7.0 on early downs | Compare like with like (H.7) |
| Regression to the mean | A small-sample hot or cold streak treated as permanent | Inside Slant: a little over six yards a call through Week 14, then about 13 yards on 4 calls in Game 2 | Four calls can't tell a broken play from a bad afternoon — and check whether the defense changed (it had: Ames sat) |
| The opponent reads the data too | A perfect streak, trusted in bold | Two-high on 3rd-and-7+ vs. 2x2, 29 straight times since their bye; the thirtieth rotated, and Royce intercepted | Perfect streaks get a question mark; every streak gets a post-snap answer; "Never let a tendency outlive its evidence" |
The fourth is what makes football analytics different from a weather forecast: the thing you're measuring is measuring you back. When a streak gets long, the question isn't "how sure are we?" It's "what would it cost us if it broke on the snap that matters most?"
H.9.5 The four-line slide
Marisol's slides look the same every week, and the sameness is the point: coaches stop reading a slide the moment they have to figure out how to read it. One number per slide, as a count; the sample; the blind spot, stated by the analyst before a coach can find it; and a recommendation in football words.
+--------------------------------------------------------------+
| THE NUMBER 2nd-and-8+ designed runs: 7 of 25 successful |
| THE SAMPLE Weeks 1-8, all opponents, 25 calls |
| THIS DOESN'T Box counts; what happened on the snap before; |
| KNOW... whether third-down rushes actually slowed |
| WHAT I'D DO Make the run half of a Kill call: light box |
| keeps it, anything else throws |
+--------------------------------------------------------------+
📊 Diagram (described): Marisol's slide template, filled in, as a box with four labeled rows. "The number": second-and-8-or-longer designed runs, 7 of 25 successful. "The sample": weeks 1 through 8, all opponents, 25 calls. "This doesn't know": box counts, what happened on the previous snap, and whether third-down pass rushes actually slowed down. "What I'd do": make the run half of a Kill call, kept only against a light box.
Blank slide:
+--------------------------------------------------------------+
| THE NUMBER .............. of ............. |
| THE SAMPLE weeks ........ opponents ........ snaps .... |
| THIS DOESN'T ................................................|
| KNOW... ................................................|
| WHAT I'D DO ................................................|
| ................................................|
+--------------------------------------------------------------+
📊 Diagram (described): The same four-row slide left blank: a count "of" a total for the number; weeks, opponents, and snaps for the sample; two lines for what the number can't see; and two lines for a recommendation in football words.
A second filled slide, built from our screen numbers:
+--------------------------------------------------------------+
| THE NUMBER 3rd-and-10+ screens: 2 conversions in 7; |
| 0 sacks, 0 turnovers |
| THE SAMPLE Regular season, 7 calls. Our other 34 snaps at |
| that distance: 8 conv, 4 sacks, 2 INT |
| THIS DOESN'T Whether 2 of 7 and 8 of 34 differ (at this |
| KNOW... size, they don't); the field position after |
| WHAT I'D DO Judge the long-long screen by what it costs: |
| no sack, no turnover, no lost kick |
+--------------------------------------------------------------+
📊 Diagram (described): A filled slide for third-and-10-or-longer screens. The number: two conversions in seven, with no sacks and no turnovers. The sample: seven regular-season calls, compared with the other 34 snaps at that distance, which produced eight conversions, four sacks, and two interceptions. What it doesn't know: whether the two conversion rates really differ — at this sample size they don't — and the field position that followed. The recommendation: judge the long-long screen by its cost, because it never produced a sack, a turnover, or a lost kick.
H.9.6 Talking to players
Players don't get EPA. They get film and one sentence, in their language. Marisol's count showed Ben Achterberg's oversets bunched after an edge rusher's second speed rush in a series; Ben never saw the table. Gus showed him three clips and said, "The third rush is the one. Set for the spin." Rome gives Darius "the ball goes where the pressure left." Rules with names travel; percentages don't. When Vance came off the field after the Game 3 interception he didn't say "I misread the rotation." He said, "Rule four, Coach." Somebody counted, a coach turned the count into a rule, and a player could say it back under the worst pressure of his season.
Taught in: Chapter 28 (sections 28.1 and 28.9); the "?" rule and the 70-on-10 floor in Chapter 23.
H.10 Fourth-Down Logic
H.10.1 The break-even number
Every fourth-down decision compares two or three futures. You need three numbers and one chance:
- The kick value: what punting (or trying the field goal) is worth, in expected points.
- The convert value: what you're worth if you go and make it.
- The fail value: what you're worth if you go and don't (the opponent's ball at the spot, sign flipped).
- Your conversion chance for this distance, against this front, with this call.
The break-even conversion rate is the chance at which going and kicking are worth exactly the same. In words: break-even = (kick value − fail value) ÷ (convert value − fail value). If your real chance is above the break-even, go.
For a field-goal try, the "kick value" is itself a blend: the make chance times what three points are worth (less what the opponent's next possession is worth to them), plus the miss chance times what the opponent's ball at the spot is worth. Most staffs let the chart do that arithmetic.
H.10.2 Worked example: Week 12, fourth-and-1 at our 43
Third quarter, up 13–10. Moose Odom's third-down run came up a long foot short.
FOURTH-AND-1, AUS 43, UP 13-10, THIRD QUARTER (illustrative EP)
PUNT --> Stallions 1st-and-10 near their 17 ............. -0.2
GO ----> convert (about 7 in 10): our 1st-and-10, ~AUS 45 +1.7
\--> fail (about 3 in 10): their ball at our 43 ... -2.3
GO overall = (0.7 x 1.7) + (0.3 x -2.3) = about +0.5
Difference: GO is worth about 0.7 points more than PUNT.
Break-even: GO wins whenever we convert more than ~53 in 100.
📊 Diagram (described): A decision tree with two branches. Punting leads to the Stallions starting near their own 17, worth about −0.2 points to us. Going splits: about seven times in ten we convert and have a first down near our 45, worth about +1.7; about three times in ten we fail and the Stallions take over at our 43, worth about −2.3 to us. The blended value of going is about +0.5, roughly 0.7 points better than punting, and the break-even is converting a little more than half the time.
The break-even, by the formula in words: kick value −0.2 minus fail value −2.3 is 2.1; convert value +1.7 minus fail value −2.3 is 4.0; 2.1 ÷ 4.0 is about 0.53. We needed to convert a bit more than half the time. The call was "Heavy, Under Wing Right, Check": the Stallions showed a bear, Vance took the sneak over the right A gap, +1½, first down. Our own fourth-and-1 number that season was 9 of 12.
The lesson: the model tells you the price of the decision. The call sheet decides whether you're paying league-average or better.
H.10.3 Worked example: a low break-even (hypothetical)
Next season, tied in the fourth quarter, fourth-and-2 at the opponent's 38, your starting left tackle just hurt, a field goal out of range. Illustrative values: punt about +0.1, convert about +2.8, fail about −1.2.
Break-even = (0.1 − (−1.2)) ÷ (2.8 − (−1.2)) = 1.3 ÷ 4.0 = about 0.33. You need to convert about one time in three. Even if the injury knocks your usual chance from the high fifties to the mid-forties, you're well above it: go. The injury changes the call — move the pocket away from the backup tackle ("Sprint Right, Snag") — not the decision.
H.10.4 What the league generally finds (hedged)
- For fourth-and-short across most of the field, the break-even is well below the rate at which teams actually convert. League-wide, fourth-and-1 has generally converted somewhere around two-thirds to three-quarters of the time, with quarterback sneaks at the high end.
- In recent seasons NFL teams have gone for it on fourth-and-short, especially between the 40s, noticeably more often than a decade or two earlier, and public fourth-down models are a big part of why.
- Those models use expected points early in games and win probability (the chance of winning from this score, time, and situation) late, when the scoreboard matters more than the average next score.
Treat every one of those as a magnitude, not a precise figure.
H.10.5 The five assumptions, and "name it, then price it"
Every fourth-down model assumes:
- A league-average play for this distance.
- A league-average kicker and punter (good charts adjust for yours; Marisol's adjusts for Evan Dorsey, whose range is about 52 yards).
- A league-average defense in this situation.
- Healthy, fresh players.
- Nothing has happened in this game that changes the odds.
Name it, then price it. To override the number, name a specific, observable thing the model can't see ("their goal-line front has stopped three different looks in a row"; "our left tackle just left the game"), then price it: turn it into a change in the conversion chance and re-run the arithmetic. Most named objections, priced honestly, don't flip the decision — they flip the call. Wade Harmon's question is the same rule in six words: "What's the call if we go?"
Two kinds of gut. Expert pattern recognition (Gus seeing the backside end chase flat; Teo seeing a safety's feet) can always say what it saw — it's data the model doesn't have. Fear of blame can only say how it feels. Nobody gets fired for a punt; that asymmetry is exactly the bias fourth-down models were built to correct. If you can't name it, the number wins.
H.10.6 "Four-down": decide before third down
The fourth-down decision isn't made on fourth down. When we cross into a zone where Wade's chart says we'll go on fourth-and-short, his decision reaches me before the third-down call — from Teo as his "one thing," or from Wade on my channel: "Four-down."
| Third down | What "four-down" changes |
|---|---|
| 3rd-and-6 | Call the play most likely to gain half of it safely; a four-yard gain sets up a fourth-and-2 we'll go for, and a sack is the one result that kills the drive |
| 3rd-and-1 or 2 | The play-action shot becomes a near-free down: its likely failure is an incompletion that still leaves a fourth-and-short we'll go for |
The defense reads it too. Russ Kimbrough told me in the spring: "When you're four-down, I'm not defending third down. I'm defending two snaps for a yard apiece, and I'm saving my best short-yardage call for fourth." The counter: make third down the shot down and fourth down the play you trust most. On the call sheet, the habit lives in the "4th: P / K / G" box on every third-down row (Appendix F, F.7.6).
H.10.7 The Thursday chart
On Thursday Marisol prints Wade a one-page chart for that opponent and that week's weather: for each distance (1, 2, 3, 4–5, 6+) and each field zone, GO, LEAN GO, TOSS-UP, or KICK, adjusted for Dorsey's range and our own conversion rates. Here is the teaching sample we give young coaches. It is illustrative, not Marisol's real chart and not a league table.
SAMPLE FOURTH-DOWN CHART -- early and middle of games, normal weather
kicker range line ~opp 35. GO / LG = lean go / TU = toss-up / K = kick
Field zone 4-1 4-2 4-3 4-4/5 4-6+
own 1-20 TU K K K K
own 21-39 GO TU K K K
own 40 to opp 40 GO GO LG TU K
opp 39-31 (no-man's) GO GO GO LG TU
opp 30-11 (FG range) GO LG TU K K
opp 10-1 (goal to go) GO GO LG TU K
Late in games: trailing moves every box toward GO; leading with a
good defense moves the own-territory boxes toward K.
📊 Diagram (described): An illustrative six-row, five-column grid. Rows are field zones from your own goal line to the opponent's; columns are fourth-down distances from 1 to 6 or more. Fourth-and-1 says GO everywhere except deep in your own territory, where it's a toss-up. The chart is most aggressive in the opponent's 31 to 39 — too far for a reliable field goal, too close for a useful punt. Inside field-goal range it leans toward going only on short distances, and goal-to-go fourth downs from the 1 or 2 say GO. A note says late-game score and time shift the boxes.
Blank chart:
| Field zone | 4-1 | 4-2 | 4-3 | 4-4/5 | 4-6+ |
|---|---|---|---|---|---|
| Own 1–20 | |||||
| Own 21–39 | |||||
| Own 40 to opp 40 | |||||
| Opp 39–31 | |||||
| Opp 30–11 | |||||
| Opp 10–1 |
Every GO box should name the call that makes it true. Worked line (pro level): Own 40 to opp 40, 4th-and-1: GO. Call: "Heavy, Under Wing Right, Check" — bear = sneak, otherwise duo, safety down and both backers inside = play-action to the Y. Break-even about 53%; our season rate 9 of 12.
H.10.8 Our fourth-down decisions, graded
| Game | Situation | Chart | Decision | Result | The grade |
|---|---|---|---|---|---|
| Game 1 (Wk 6) | 4th-and-goal at the SAC 3, Q1, up 7–0 | Go, close | Wade kicked (named: the front had stopped three looks; Jumbo Power Left being saved) | FG, 10–0 | A close call moved slightly toward kicking; passes the test. The saved card scored from the 2 nine weeks later |
| Week 2 at STL | 4th-and-2 at the STL 37, late Q4, down 17–20 | — | Went | Incomplete | — |
| Week 10 at POR | 4th-and-1 at the POR 18, late Q3, up 17–10 | Go | Went: duo | Stopped for no gain (Pratt slanted inside) | Decision right, call reasonable, execution lost the rep |
| Week 12 vs. STL | 4th-and-1 at the AUS 43, Q3, up 13–10 | Go, "not close" | Went: sneak vs. a bear | +1½, first down | Right, and paid |
| Game 2 (Wk 15) | 4th-and-1 at the AUS 36, early Q4, down 20–13 | Go, "not by a little" | Wade punted: I couldn't name a call I trusted against big nickel | SAC FG, 23–13 | The model's first assumption (a league-average play) wasn't true that day; the fix was a short-yardage package we trusted |
| Game 2 (Wk 15) | 4th-and-4 at the AUS 37, 0:58 Q2, tied 10–10 | Punt | Wade's timeout at my request, then punt | SAC FG, 13–10 at half | The decision belonged in the third-down call |
A failed fourth down you still believe in is the best teacher you'll get. If you grade fourth downs by result, you'll teach yourself to punt: the failures are memorable and the punts are invisible.
Taught in: Chapter 28 (sections 28.6–28.7); the short-yardage calls in Chapter 14; the end-of-half case in Chapter 26.
H.11 Two-Point Logic
H.11.1 The arithmetic
In the NFL a two-point try is snapped from the 2-yard line, and the extra-point kick has been snapped from the 15 since 2015. (College and high school snap the two-point try from the 3; check your code.) Over many seasons, NFL two-point tries have converted somewhere around half the time, and extra points since the move to the 15 have generally been made somewhere in the low to mid nineties percent.
| Choice | Points × chance | Worth, roughly |
|---|---|---|
| Two-point try | 2 × about one-half | about 1 point |
| Kick | 1 × a little over nine in ten | a little under 1 point |
On average, they're worth about the same. So the decision is driven almost entirely by the score — what the margin will be after the try, and what the opponent then needs.
H.11.2 The teaching chart
This is the chart we teach, written from the offense's side, using the margin after your touchdown and before the try. It's a late-game lean, not a law; early in a game, kick unless your two-point play converts well above the league's rough coin flip.
| Margin before the try | Kick makes it | Two makes it | Late-game lean | Why |
|---|---|---|---|---|
| Down 2 | Down 1 | Tied | Go | A kick still loses |
| Down 5 | Down 4 | Down 3 | Go | Down 3, a field goal ties |
| Down 8 (you were down 14) | Down 7 | Down 6 | Go — and debated | Convert and your next TD + PAT wins; fail and your next TD + two ties, the same as kicking twice would |
| Down 10 | Down 9 | Down 8 | Go | Down 8 is one possession; down 9 is two |
| Down 1 | Tied | Up 1 | Kick, usually | Go only if you'd rather finish it now |
| Up 1 | Up 2 | Up 3 | Go | Up 3, their field goal only ties |
| Up 5 | Up 6 | Up 7 | Go | Up 7, their TD + PAT only ties |
| Up 12 | Up 13 | Up 14 | Go | Up 14, two TDs + two PATs only tie |
The "down 8" row is the one analysts have pushed hardest in recent seasons; a growing number of NFL coaches now go there, and others still kick and play for overtime. Every other row is close to settled logic.
Worked example — up 5 with three minutes left. Convert and you're up 7: their touchdown and extra point only tie. Kick and you're up 6: their touchdown and extra point beat you. Fail the two and you're still up 5 — they still need a touchdown, exactly as they would if you'd kicked. Small downside, large upside: go.
H.11.3 Our two-point season
Three tries, two conversions, none against Sacramento. The base play is the sprint run-pass from Chapter 14: "Eleven, Gun Trips Right, Sprint Right, Snag," which threatens the whole width of the goal line and makes one flat defender choose. It lives on the Green panel next to the goal-line runs, and it's on the call sheet every week whether we expect to use it or not, because a two-point decision you haven't prepared a play for is just a hope. Wade owns the decision, as he owns fourth downs and timeouts.
🏫 Any level: Everything in H.10 and H.11 tilts toward going as you move down the levels. In high school, field goals and extra points are less reliable and punts are shorter and more often blocked, so the value of kicking falls and the break-even for going drops. The two-point try from the 3 nudges the conversion rate down. Build your chart from your own kicker and your own short-yardage rate.
H.12 The Outlaws Season Dashboard
Every number on our season, in one place, each with its sample and what it can't see. All of it is our fictional team's charting; none of it is a league benchmark. Regular season unless marked.
H.12.1 The one-page view
OUTLAWS SEASON DASHBOARD (regular season unless marked; fictional)
+--------------------------------------------------------------+
| RECORD 11-6, second in the division |
| Wild card: W 30-24 at OKC. Divisional: L 17-20 |
| 3RD DOWN 92 of 204 short 38/57 medium 30/62 |
| long 19/58 long-long 5/27 |
| RED ZONE 31 TDs in 51 trips |
| 4TH DOWN 26 tries, 15 converted (11 between the 40s) |
| 2-POINT 3 tries, 2 converted (none vs. Sacramento) |
| EXPLOSIVE 71 in 17 games (runs 12+, passes 20+) |
| SACKS 31 TURNOVERS 17 (11 INT, 6 fumbles lost) |
| SCREENS 52 calls for 301 yards (5.8 a call) |
| PLAY-ACT. ~28% of dropbacks; ~8.6 vs ~6.2 yds per dropback |
| SHOTS BOX 46 calls: 19 comp, 14 explosives, 3 sacks, 2 INT |
+--------------------------------------------------------------+
📊 Diagram (described): A one-page dashboard box for the Outlaws' season. Record 11–6, second in the division, a wild-card win at Oklahoma City and a divisional loss at Sacramento. Third downs 92 of 204, split 38 of 57 short, 30 of 62 medium, 19 of 58 long, and 5 of 27 long-long. Red zone: 31 touchdowns in 51 trips. Fourth downs: 26 tries, 15 conversions, 11 of them between the 40s. Two-point tries: 3, with 2 converted and none against Sacramento. 71 explosives in 17 games. 31 sacks and 17 turnovers. Screens: 52 for 301 yards. Play-action about 28 percent of dropbacks, about 8.6 yards per dropback against about 6.2. Shots box: 46 calls, 19 completions, 14 explosives, 3 sacks, 2 interceptions.
H.12.2 Identity
| Line | Count | This doesn't know… |
|---|---|---|
| 11 personnel | ~66% of snaps (68% in the four-game sample: 168 of 247) | What each grouping was for that week |
| 12 personnel | ~21% (20% in the sample: 49 of 247) | That against Sacramento, after Week 6, Twelve drew big nickel every time |
| Pre-snap motion | ~55% of snaps | Which motions were diagnostic and which were for a matchup |
| RPO-tagged plays | ~15% | How many were handoffs and how many throws |
| Play-action | ~28% of dropbacks | That about four in five came on early downs |
| Early-down PROE | A little above zero | Which situations drove it |
H.12.3 Third down, red zone, fourth down, two-point
| Line | Count | This doesn't know… |
|---|---|---|
| Third down | 92 of 204 (about 45%): short 38/57, medium 30/62, long 19/58, long-long 5/27 | What the early-down calls did to the distance |
| Red zone | 31 touchdowns in 51 trips (about 61%) | Which zone each trip stalled in (fringe, high red, low red, goal line) |
| Fourth down by distance | 4th-and-1: 12 tries, 9 conv (sneaks and duo) · 2–3: 7 / 4 · 4–6: 3 / 1 (all passes) · 7+: 4 / 1 (all late, trailing) · total 26 / 15 | Whether the call on each was one we'd repped against that front |
| Chart vs. head coach | Chart said GO or LEAN GO on 35; Wade went on 24 of them; went twice when it said kick (both Q4, both converted); kicked or punted 11 times when it said go | The reason Wade named for each override (he could name one for almost every one) |
| Two-point | 3 tries, 2 converted, none vs. Sacramento | Three tries is not a rate |
H.12.4 Explosives, sacks, turnovers
| Line | Count | This doesn't know… |
|---|---|---|
| Explosives | 71 in 17 games; drives with one: a little under half ended in TDs; drives without: about one in eight | Which plays set them up (Anchor D's +5 and +7 runs count as zero explosives) |
| Sacks by cause | 31: lost the rep 12 · held the ball 7 · count 6 · coverage 4 · unblockable 2 | The buckets are judgment calls; another analyst would move a few |
| Turnovers | 17: 11 interceptions, 6 fumbles lost | — |
| Interceptions by cause | 11: eyes beaten by a rotation after the snap 4 · forced into tight coverage late 3 · tipped or off a receiver's hands 2 · quick outs jumped by a corner 2 | The biggest bucket was the one a defense could schedule — and Sacramento did |
H.12.5 Screens
| Screens (season, RPO tags not counted) | Calls | Yards | Per call |
|---|---|---|---|
| All designed screens | 52 | 301 | 5.8 |
| Back screens (slip, swing, double) | 24 | 152 | 6.3 |
| Receiver screens (bubble, tunnel, standalone now) | 20 | 98 | 4.9 |
| Tight end and middle screens | 8 | 51 | 6.4 |
| vs. 5+ rushers | 13 | 109 | 8.4 |
| vs. 4 rushers | 39 | 192 | 4.9 |
| On 3rd-and-10+ | 7 | 55 | 7.9 (2 conversions, 0 sacks, 0 turnovers) |
This doesn't know: thirteen snaps against pressure is a small sample; and the third-and-long row is seven snaps (H.8).
H.12.6 Play-action splits
| Split | Yards per dropback |
|---|---|
| Play-action, all | about 8.6 |
| Straight dropbacks, all | about 6.2 |
| Play-action, 1st and 2nd down only | about 8.6 |
| Straight dropbacks, 1st and 2nd down only | about 7.0 |
| Play-action when one of the two previous snaps was a run from the same formation (38 calls) | about 10 |
| All other play-action | about 7.6 |
This doesn't know: the previous-snap split is a small sample with wide error bars, and it says the previous play matters, maybe — not the season rushing average. Two of our better play-action games came against St. Louis, the defense that made play-action pay. YAC note: on Wide PA and boot completions, a little over half the yards came after the catch; on dropback completions, about a third.
H.12.7 The boot table
| Evidence before the call (24 Boot and Naked calls) | Calls | Completions | Yards per call |
|---|---|---|---|
| End chased on two or more zone runs earlier on the drive | 13 | 10 | 10.2 |
| End chased once | 6 | 3 | 4.5 |
| End never chased, or no zone run yet on the drive | 5 | 2 | 1.8 |
Marisol: "I wouldn't bet the house on the middle row. But the top and bottom rows aren't close." That's the evidence behind the "chases twice" rule Teo brought into our booth the week after Week 2.
H.12.8 Y Cross
| Y Cross (season) | Count |
|---|---|
| Calls | 18 |
| Completions | 12 |
| Explosives | 5 |
| Calls vs. single-high | 13 of 18 |
| Calls vs. two-high | 5 — producing 1 first down (one of the five was a sack where the protection broke down before anybody crossed) |
Marisol: "Y Cross is a single-high call on our film." A lean, not a rule.
H.12.9 The SHOTS box
| Shots box (season; the zero bin counted separately) | Count |
|---|---|
| Calls | 46 |
| Throws / sacks / scrambles | 41 / 3 / 2 |
| Completions | 19 |
| Explosives | 14 |
| Interceptions | 2 |
| Defensive pass interference drawn | 4 |
| All Go calls (season) | 29 |
Marisol's summary: "The median shot is an incompletion. The average shot was still worth it, because the hits are huge and the interference flags count. But it's a high-variance play, so call it when you can afford a miss." The call-it-when rule is in Appendix F, F.7.4.
H.12.10 Inside Read
| Inside Read family (season, all versions) | Count |
|---|---|
| Calls | 88 |
| Vance kept | 27 (about 31%); the other 61 went to the back |
| Success rate on keeps | about 52% |
| Success rate on gives | about 46% — close to plain inside zone |
This doesn't know: with 27 keeps the difference is too small to call real; Marisol footnotes it on every slide. Vance's designed-carry budget is about three a game in the regular season, up to five in a playoff game with Wade's sign-off.
H.12.11 The run game by box count
| Called runs (not RPOs), by box | Carries | Successful |
|---|---|---|
| Six-man boxes | about 140 | a little over half |
| Seven-man boxes | about 160 | a bit under half |
| Eight or more | under 40 | about a third |
This doesn't know: against Sacramento's Quarters, the six-man-box runs looked like the seven-man number — the safeties were playing like box defenders. The rule that came from counting boxes: "Seven or fewer, run it; eight, take the throw."
H.12.12 Practice and communication
| Line | Count | This doesn't know… |
|---|---|---|
| Calls with 2+ full-speed reps that week | about 780 snaps, successful a little under half | The low-rep calls skew toward strange situations that are harder for everyone |
| Calls with 0–1 full-speed reps | about 190 snaps, successful about four in ten | 190 is a small pile; it doesn't prove reps cause success |
| Timeouts used for communication | 4 in 17 games (3 before the Week 9 bye) | Four timeouts prove nothing about a season; the direction was clear |
| Calls sent as a band number | from about 1 in 12 snaps to about 1 in 5 | — |
| Delay-of-game penalties | 2 | — |
H.12.13 A one-game dashboard: Game 1
For a single game the dashboard shrinks to a few lines. Game 1 against Sacramento (Week 6, won 34–27): success rate 49%; third downs 8 of 14; red-zone touchdowns 2 of 3. One game is a story, not a sample; read it for what to look at on Monday, not for what to believe.
Taught in: Chapter 28; the screen table in Chapter 20; the play-action splits in Chapter 19; the boot table in Chapter 29; Y Cross in Chapter 30; the shots box in Chapter 18; Inside Read in Chapter 12; third downs and the red zone in Chapter 26.
H.13 Public Data Sources
You don't need a team's budget to count. These sources are named so you can find them; definitions and access change, so check each one's current documentation before you compare numbers across them.
| Source | What it offers, generally | Watch for |
|---|---|---|
| nflverse (the community behind the nflfastR play-by-play data) | Free, open-source play-by-play going back many seasons, with expected points, EPA, win probability, and a completion-probability field for CPOE; community members have also published fourth-down models | It's a model; its decimals will differ from other models' decimals |
| NFL Next Gen Stats | The league's player-tracking measures, including completion-probability-based numbers and timing measures for quarterbacks and receivers | Tracking-based definitions differ from play-by-play ones |
| Pro Football Focus (PFF) | Charting of every snap and player grades; largely subscription | Grades are judgments by charters, not counts |
| Sports Info Solutions (SIS) | Detailed charting data on alignments, coverages, and play types | Charting definitions are specific to the provider |
| Football Outsiders' DVOA (historical) and FTN's continuation | An efficiency measure adjusted for situation and opponent | It isn't EPA; don't mix the two in one table |
| Broadcast fourth-down models | Live go / kick / punt recommendations on screen | You see the recommendation, not the assumptions |
| The official NFL Rulebook; NCAA and NFHS rules | The rules behind the arithmetic: where the two-point try and the extra point are snapped, clock rules | Rules change; check the current edition |
Three cautions for every source:
- Definitions differ. Success rate has two common definitions (H.4); explosive thresholds vary by source (this book uses runs of 12+ and passes of 20+); "pressure" and "blitz" mean different things in different charting. Before comparing two numbers, make sure they count the same thing.
- Broadcast film can't see everything. Tag "?" when the camera cuts away; a guess written as a fact is worse than a blank.
- Public numbers describe the league, not your team. A league-average fourth-and-1 rate assumes a league-average call. Build your own counts — six games of a manager's hand chart beats any league table for knowing your offense.
H.14 Reference Card: Every Calculation in Words
| To find… | Do this | Worked |
|---|---|---|
| EPA of a snap | EP after − EP before (a turnover's "after" is the opponent's EP with the sign flipped) | Hitch on 3rd-and-7: +2.0 − 0.8 = +1.2 |
| Success (distance rule) | 1st down: gain ≥ 40% of distance; 2nd: ≥ 60%; 3rd/4th: all of it | +3 on 1st-and-10: needed 4 → no |
| Success (EPA rule) | EPA above zero | — say which rule you used |
| Success rate | Successful snaps ÷ snaps, written as a count first | "26 of 58" |
| Explosive rate | (Runs of 12+ + passes of 20+) ÷ plays (or per game) | 71 in 17 games ≈ 4.2 a game |
| CPOE | Actual completion % − expected completion % | 70% − 65% = +5 (illustrative) |
| Early-down pass rate | Early-down passes ÷ early-down snaps (say what you excluded) | — |
| PROE | Actual pass rate − expected pass rate | 56% − 52% = +4 (illustrative) |
| Third-down rate by band | Conversions ÷ attempts in short / medium / long / long-long | Long: 19 of 58 |
| A tell | ≈70% or more on 10+ snaps — both parts | 11 of 17 = 65%: a lean |
| Fourth-down break-even | (kick value − fail value) ÷ (convert value − fail value) | (−0.2 + 2.3) ÷ (1.7 + 2.3) ≈ 53% |
| Two-point vs. kick, on average | 2 × conversion chance vs. 1 × kick chance | About 1 vs. a little under 1: the score decides |
| How far luck moves a rate | 20 snaps ≈ ±20 points; 100 ≈ ±10; 400 ≈ ±5 | "60% on 20" could be 40% or 80% |
When should the number override your gut? Four questions, in order:
- Is the sample big enough? Counts first. Under about 20 snaps, it's a lean.
- Is this the situation the model assumes? League-average play, kicker, defense, healthy players.
- Can you name what it can't see — and price it? Most of the time that changes the call, not the decision.
- Is your objection about the play, or about the blame? If you can't name it, the number wins.
Taught in: Chapter 28, with the third-down cliff in Chapter 26, the scouting counts in Chapter 23, and the "a call is a bet on a distribution" idea in Chapter 25.