Who rises when the game gets big, and who shrinks.
Blank = use this season's betting ratings. Negative spread = home team favored.
The big numbers are the most likely real final score near the prediction, nudged toward scores football actually produces (24, 31, 17, 14 and 27 are common; 22 is six times rarer than 24). The averages underneath are the raw prediction, and close matchups land in the 20s. Real games miss the line by about 15 points either way, which is why one in five games has a team scoring 40+ and one in five has a team held under 14.
Every number is points of final margin, added to whichever team is named. "Covered" means how a team did against the betting spread: +2 means it beat the spread by 2 points on average. "Trust" grows with sample size: games ÷ (games + 30), so 30 games is trusted halfway. Tap the i on any row to see its math.
The DAWG Index grades every Power 4 game on whether a team showed up: beat the spread, finished drives, kept pulling away with a lead, cashed in after defensive stops and answered when the other team scored. Garbage time doesn't count, and road, night, packed-stadium and ranked-opponent games count more.
Download every graded game to check the work. It describes how teams played. It doesn't predict who wins or covers. For how a team is likely to play, see Tendencies. Tap any game to see how it graded.
Right = better than expected in routine games · Up = better in big games
Zero is a typical team in that kind of game that season, since big games are harder for everyone. Bigger dots played more big games. The scale adjusts to each season; full seasons sit closer to zero because a dozen-plus games average out the extremes.
Coach Career follows the current head coach, not the school. It blends every season he's been a head coach since 2019, including previous jobs (Jon Sumrall's Troy and Tulane years count for Florida). Recent seasons count more (each year back is worth 80%), and short careers are pulled toward zero. Zero is a typical current Power 4 coach, since most of them were hired after beating expectations somewhere else. Dawg and Soft start at ±0.8 in this view instead of ±3.
| # | Team | W–L |
|---|
| # | Conference | Teams | DAWG | Big Games | Routine | Big-Stage Drop | Dawgs · Soft |
|---|
The best single-game performances of the selected season. The game grade rewards beating the spread, finishing drives, growing a lead, cashing in after stops and answering when the opponent scores, counts only competitive plays (no garbage time), and scales up for road, night, full-house and ranked-opponent games. Tap a game to see how it graded. Hindsight swaps the closing spread for one rebuilt from every other result that season, then re-grades the game; the number in parentheses is the change. The spread is only one of five parts and the grade isn't in points, so a 10-point change in the spread moves the grade about 1 to 2 points, not 10. Early-season hindsight is noisy.
| # | Team | Game | Score | Spread | Hindsight Spread | Game Grade | Hindsight Grade |
|---|
The game grade tells you how a team played. These tell you how it's likely to play. Each trait made the list only because it carries over from one season to the next (tested on 2021–2024, then checked on 2025). Each number starts from last season, discounted by how well that trait carries over (more for a new coach), then updates with this season's games.
| # | Team |
|---|
Carryover is the correlation between a trait one season and the next (0 = none; team quality is about 0.5). "Predicts" is whether the trait told us anything about the next season's results beyond team quality.
| Trait | Carryover 2021–24 | Checked on 2025 | Same coach | Predicts next season | Verdict |
|---|
Two plain counts behind the DAWG idea: who lets games get away after building a lead, and who takes advantage when its defense gets a stop.
| Team |
|---|
| Team | Game | Led by | Final |
|---|
| Team |
|---|
| Game | Date | Score | Game Grade |
|---|
| Game | Date | Score | Game Grade |
|---|
| Opponent | Date · Site | Stage | Score | Line | vs Spread | Finishing | Knife Twist | Cash-In | Answers | Starting QB | Game Grade |
|---|
How each coach's teams have done against expectations, season by season. Career and At School averages use full seasons (8+ games); This Season shows the current year on its own. Read it as a résumé, not a forecast: we tested whether a coach's average or his trend predicts his next season, and neither does (r ≈ 0). It also measures expectations, not wins. A coach whose teams win a lot but keep falling short of the spread, like Lincoln Riley's early Oklahoma teams, grades lower than his record.
| Coach | School | At School Since | Seasons | Record | By season |
|---|
If grit is a trait, a team's grade should carry over: good this year, good next year. If it's mostly luck, it should reset. These are correlations across 2019–2025 (0 means no carryover, 1 means perfect).
Each game gets a stage score, the same for both teams except where noted: ranked opponent +15 (top 10 +25), ranked vs. ranked +10, rivalry +20, national TV (ABC, CBS, NBC, FOX, ESPN) +10, primetime on national TV +10 more, bowl, playoff or title game +30, neutral-site showcase +20, non-conference Power 4 matchup +10, packed stadium (90%+ of its biggest crowd, night crowds count 1.25×) +10, and an opponent that won 10+ games the season before +10. 25 or more is a big game, so it takes at least two of those. Contenders in big conferences play more of them, and that's intended.
Only competitive plays count. A play is garbage time when the lead before it is more than 38 in the 2nd quarter, 28 in the 3rd or 22 in the 4th. If the lead shrinks back under the line, plays count again, so a collapse still shows up.
Did they beat the expectation set before kickoff? Home field is already priced in.
competitive margin + closing lineDid the scoreboard match the yardage? The yards-to-points rate is fit to each season's games. Out-gaining a team and losing is a finishing gap.
competitive margin − expected margin from yardsOnce up 9 or more, does the lead grow? Net points from that moment until garbage time, compared with what's normal for that matchup.
Points on the very next drive after the defense gets a stop or a takeaway, compared with what's normal for a team favored by that much. The national average is about 45% of those drives scoring.
When the opponent scores to make it a one-score game, how often does the next drive score?
Each part is measured in standard deviations against every Power 4 game since 2019, so there's no ceiling: a 30-point cover counts for more than a 14-point one. Answers rests on a handful of drives, so it counts half as much as the other four. A game needs at least three of the five parts. The average is scaled so about two-thirds of games land between −5 and +5.
weighted average of part SDs × 7 × settingBigger stages magnify the grade, good or bad: road +10%, night +5%, full road stadium (90%+, night crowds count 1.25×) +10%, ranked opponent +10% (top 10 +20%), postseason +10%. A good grade against a Group of 5 team shrinks 15%, against an FCS team 40%; a bad grade against them is never softened.
A season's DAWG is the average game grade. Big-game and routine grades use the grade before the setting adjustment, so a big stage doesn't make a team look like it rose just because the stakes were higher. Those averages are pulled slightly toward the season average (as if one average game were mixed in), so a single game can't put a team at the far edge of the chart; the pull fades as teams play more games. Season DAWG is centered so the average Power 4 team is 0. +3 or better for a season is Dawg, −3 or worse is Soft.
big-stage drop = big-game grade − routine gradeWins minus expected wins, where expected wins come from each game's yardage. +2 means two more wins than the box scores earned. Teams with high luck tend to come back to earth.
Data: ESPN box scores, drive logs and closing lines (ESPN BET and DraftKings) for every game involving a current Power 4 team or Notre Dame since 2019, including bowls and playoffs. Coaches from Wikipedia's list of current FBS coaches. Attendance is missing for some games; those use a 75% default.
Check the work: download every graded game (CSV, one row per team per game, with each part of the grade). It's updated every Sunday with the site.