DAWG Index

More than the score.

Vegas sets the bar for every game. We grade who cleared it.

This Week

The week's biggest stages, ranked. Each team's rank is where it stands at living up to expectations this season, out of every Power 4 team. Green top third · gold middle · red bottom third.

Every game is graded Sunday. Then come back to see who showed up.

Compare Two Teams

Side by side this season. No predictions, just how each has played.
vs

Green marks the better side. Everything here describes how each team has played this season; none of it predicts the game.

What is this?

Vegas sets the bar for every game. We grade who cleared it.

  1. What it is. Every Power 4 game is graded against the closing line. DAWG is how many points better or worse than expected a team played, with garbage time thrown out. 0 = exactly what Vegas expected · +12.3 or better = a DAWG game · −12.3 or worse = SOFT
  2. Why it matters. A record says who won. This says who played above or below what was expected of them. PLAY checks the offense and defense against the same expectations, in the same points.
  3. Why trust it. Every Power 4 game since 2019, every grade public, and the formula is frozen and versioned. We show what holds up and what doesn't.

Where does your team rank?

Highlights

DAWG: 0 = played like Vegas expected · +12.3 or better = DAWG game · +24 or better = Certified DAWG · −12.3 or worse = SOFT

DAWG Rankings

Hover a header for its meaning · tap to sortTap a header to sort · tap a team for its page

Season DAWG = average game DAWG, in points vs. what the closing line expected · PLAY = offense + defense vs. expectations · not a power ranking

#TeamConfW–L

Scoreboard vs. Offense + Defense

This season so far. Up = the scoreboard beat expectations (DAWG) · Right = offense + defense did (PLAY)

Swipe the chart sideways to see every team →

Poll Check

CFP rank next to DAWG rank: where each team ranks at living up to expectations, out of every Power 4 team.

Above Vegas ranked and a top-12 DAWG team · Below Vegas ranked, but in the bottom half of the DAWG Rankings · Unranked, above Vegas unranked, but a top-15 DAWG team

CFP RankCFPTeamW–LDAWG RankDAWGDAWGPLAYVerdict

Biggest Blown Leads

This season · the biggest leads that still turned into losses
TeamGameLed byFinal

Top 10 Games

Single-game DAWG

The season's best single-game performances against expectations. DAWG is points better than the closing spread said to expect, garbage time out. PLAY is the offense and defense vs. expectations, from expected points added (EPA), in the same points; special teams aren’t in it. Tap a game for the full card.

#TeamGameScoreSpreadExpectedMarginexcl. garbage timeDAWGPLAY

Conferences

Is the conference actually better, or just louder?

Results: Power 4 teams playing Power 4 teams from other conferences (nonconference games, bowls, the playoff). Expectations: how each conference's teams have played against what was expected. A conference can win head to head and still sit near zero on expectations, because Vegas already expects it to win.

ConferenceSince 2019Avg marginAvg DAWGWinning yearsThis yearMargin

Results: head to head, 2019 to now

Against expectations, this season

Average season DAWG and PLAY of each conference's teams · tap a conference to filter the rankings and chart

#ConferenceTeamsDAWGPLAYDAWG gamesSOFT games

Year by year

Season DAWG Season PLAY Current Coach's Tenure

OpponentScoreDAWGPLAYDate · SiteLineExpectedMarginexcl. garbage timeCash-InAnswersKnife TwistStarting QB

Tap a game for the full card: what the line expected, what happened with garbage time out, and how the offense and defense graded (PLAY).

Best Performances Since 2019

GameDateScoreDAWGPLAY

Worst Performances Since 2019

GameDateScoreDAWGPLAY

Quarterbacks

Points per game each QB adds compared with an average Power 4 QB

Every pass, sack, run and turnover counts, adjusted for the defense he faced. Unlike team scores, this one sticks from year to year, even after a transfer.

#QB
Loading quarterbacks…

Rating pulls small samples toward average; Raw is the number with no pulling. Tap a QB for his year-by-year line and game log. This season needs 3+ games; career needs 10+.

Team Tendencies

Season projection · through Week 11 · hover a header for its formula · tap to sort

The game grade tells you how a team played. These are habits that tend to stick from one year to the next, so they say more about how a team will play. Early in the season each number leans on last year (less so with a new coach) and moves as this season's games come in.

#TeamConf

Trackers

2026 season · through Week 11 · hover a header for its formula · tap to sort

Who lets games get away after building a lead, and who makes it count when the defense gets a stop.

Blown Leads

Team

Cash-In After Stops

Team

How the Grade Works

Two numbers for every team in every Power 4 game since 2019, kept separate on purpose. Tap any heading below for the full rule.

1

DAWG

The scoreboard, in points vs. Vegas. How many points better or worse a team's scoring margin was than Vegas's pregame expectation, with garbage time removed. Special teams count. A season grade is the per-game average.

2

PLAY

Offense and defense vs. expectations. Expected points added (EPA) on every competitive offensive and defensive snap, compared with what the spread implied, in the same points. Special teams aren't in it.

3

Labels

+12.3 or better is a DAWG game, +24 a Certified DAWG. −12.3 or worse is SOFT, −24 Certified SOFT. Labels come from DAWG only.

Then: a team's season number is the plain average of its games. No extra weight for road games, night games or ranked opponents, and DAWG and PLAY are never added together.

DAWG: the scoreboard, in points vs. Vegas

Take the final margin with garbage time removed. Compare it with the margin teams usually produce at that closing spread. A 7-point favorite usually wins by about 7.6 once garbage time is out; a 35-point favorite by only about 29, because blowouts get called off early. DAWG is the difference, in points. So the expectation is the spread adjusted for how games at that spread usually end, not the spread itself.

How close is that to simply beating the spread? Close: in about three of four graded Power 4 games the two land within 3 points of each other. They differ most when points were scored in garbage time and when the spread is big.

DAWG = margin (no garbage time) − expected margin for that spread

Garbage time is out

A play is garbage time when the lead before the snap is more than 38 in the 2nd quarter, 28 in the 3rd or 22 in the 4th. That's the CollegeFootballData rule. The 1st quarter and overtime are never garbage time. If the lead shrinks back under the line, plays count again, so a collapse still shows up. DAWG and PLAY use the same window.

PLAY: offense and defense, in points vs. Vegas

Every competitive offensive and defensive snap is valued by how much it changed the expected score (expected points added, or EPA). Kickoffs, punts and field-goal tries are special teams and aren't counted. A team's number is its offense's points added per play minus what its defense allowed per play. PLAY compares that with what the spread implied and turns it into points, scaled by how many competitive snaps the game had, so it reads on the same scale as DAWG.

PLAY = conversion × (net EPA per play − expected for that spread) × (competitive snaps ÷ average snaps)

It's shown to one decimal, like DAWG, straight from the unrounded grade. That's not a claim of precision: one game's PLAY is noisy, and about 95% of single-game PLAY results land within roughly ±21 points, with a standard deviation of about 11. A season of games is much steadier.

Known lean. PLAY runs a little low for clear favorites. From 2019 to 2026, teams favored by 14 to 21 averaged −1.7 PLAY and teams favored by 21 to 28 averaged −2.0, where a fair grade would average 0 (DAWG for the same games averages within half a point of 0). That's small next to how much a single game's PLAY varies (roughly ±21 points for 95% of games), but over a season a big favorite's PLAY reads about 2 points low. V3's formula is frozen, so this isn't adjusted mid-version; read a heavy favorite's PLAY with that in mind.

DAWG and PLAY are separate

They usually agree. When they don't, it's because PLAY counts only offensive and defensive snaps: kickoffs, punts, field goals, kick and punt return touchdowns and how the big plays happened to bunch together move the scoreboard but not PLAY. We show both and let the disagreement speak for itself. We never add them, average them or rank by both.

Example. Oklahoma beat Texas 27–24 on Oct. 10, 2026, as a 7.5-point underdog: +11.2 DAWG but only +1.1 PLAY. Its scoreboard result beat expectations by far more than its offense and defense did. A blocked-punt touchdown counted on the scoreboard, but special teams aren't in PLAY.

The labels

The cutoffs are the 5th, 20th, 80th and 95th percentiles of every Power 4 game from 2014 to 2022, rounded to the same distance on both sides, then frozen. About 1 game in 20 is Certified DAWG and 1 in 20 Certified SOFT; most games are As expected.

No hindsight

Each season is graded with expectations built only from earlier seasons, back to 2014. The 2026 grades use what was learned from 2014–2025; the 2019 grades use 2014–2018. Once a season's numbers are set, they don't change.

Missing data

No closing line: no DAWG and no PLAY. If the scoring record doesn't match the final score, no DAWG. No usable play-by-play: DAWG comes from ESPN's scoring summary and PLAY isn't available. We check the play-by-play's scoring order against ESPN's scoring summary. If the two disagree, DAWG uses the summary and PLAY isn't shown, because the play-by-play's score columns are out of order. If the play-by-play's scoring doesn't add up and there's no summary to check it against, no DAWG. Fewer than 20 competitive snaps on offense or on defense: PLAY shows "too few plays." Overtime points count in DAWG, but overtime snaps aren't in PLAY, because the expected-points model doesn't cover overtime.

What's description only

Cash-in after stops, Answers, Knife twist, turnovers and the setting (road, night, packed stadium, ranked opponent) are shown as counts and tags. They describe a game; none of them changes a grade.

What changed from V2

The old grade averaged five scored parts and multiplied the result for tougher settings. That's gone, along with the big-game split, Luck and the Matchup Builder: the research didn't support them as lasting team traits. The old site is archived, unchanged, at dawgindex.com/v2.

Data: ESPN scores, play-by-play, drive logs and closing lines for every game involving a current Power 4 team or Notre Dame since 2019 (2014–2018 too, for building expectations), including bowls and playoffs. Expected points: the open-source cfbfastR model, pinned to one version (). Coaches from Wikipedia's list of current FBS coaches.

Check the work: download every game (CSV, one row per team per game, graded or not, with the reason when it isn't) and the frozen methodology numbers. Methodology v3.0 · config 6309ce0da7a16d54dd8870ee164eceead85d2176fec913476bdaed19a96cf7b4. Updated every Sunday with the site.

Does It Hold Up?

If playing above expectations were a trait, a team's season DAWG would carry over: good this year, good next year. Each number below runs from 0 (no carryover at all) to 1 (perfect). For reference, how good a team is carries over at about 0.5.

Which Habits Stick

Each team page shows five habits. These are the ones that carry over from one season to the next; DAWG itself mostly doesn't.

What made the cut, and what didn't

Carryover is how closely a habit one year matches the next: 0 is not at all, 1 is perfectly. For reference, how good a team is carries over at about 0.5. Habits that use expected points (Foot on the gas) are measured with the cfbfastR model, the same one behind PLAY.

TraitCarryover 2021–24Checked on 2025Same coachVerdict

Questions You'll Ask

Why isn't my unbeaten team ranked higher?

The rank isn't about wins. It's how a team has played compared with what was expected of it. A 5–0 team that was supposed to win big and keeps winning close can rank below a 3–1 team that keeps beating expectations. Records still matter for the playoff. This measures something different: who's showing up.

Is this a betting tool?

No. DAWG and PLAY describe games after they happen. A team's season DAWG barely carries into the next season (the numbers are in "Does It Hold Up?" above), so it isn't a forecast of anything.

Why don't DAWG and PLAY always agree?

DAWG is the scoreboard. PLAY is the offense and defense. Kickoffs, punts, field goals, kick and punt returns, and how the big plays bunched together count on the scoreboard but not in PLAY, so a team can win big on special teams with an ordinary offense and defense, or play well on both sides of the ball and lose the kicking game. Oklahoma's 27–24 win over Texas in 2026 was +11.2 DAWG and +1.1 PLAY; a blocked-punt touchdown was part of the gap. When they disagree, both numbers are telling you something true.

Why does a win sometimes grade negative?

Because DAWG compares against expectations. A team favored by 20 that wins by 10 didn't do what it was supposed to, and DAWG says so. Same the other way: a 7-point loss as a 20-point underdog grades well.

Does the formula change during the season?

No. The formula is frozen and versioned (now v3.0). Before each season, its expectations are built from earlier seasons only and then locked. Any other change ships as a new version with a dated entry below, and old grades keep the version they were made with. Data fixes are listed below too, so nothing changes quietly.

Changes

© 2026 DAWG Index · dawgindex.com