NovaKingSports
NovaKingSports
College FootballCollege BasketballNFLNBA
  • Home
  • Rankings
  • Teams
  • Players
Menu

Sport

College FootballCollege BasketballNFLNBA

Theme

  • Home
  • Rankings
  • Teams
  • Players
  • Scores

    • Scores & Schedule
    • What to Watch
    • Standings
  • More

    • Compare
    • Analytics
    • Games
    • Film Room
    • Polls
    • Articles
    • Videos
    • About

NovaKingSports — data driven ratings.

RankingsTeamsPlayersCompareAnalyticsGamesFilm RoomPollsArticlesAboutFAQMethodologyVideos
Privacy PolicyTerms of ServiceContactCookie Preferences

College Football · Analytics

Does a Tough Non-Conference Schedule Hurt?

Coaches say scheduling up gets punished: lose to a good team and the poll drops you, beat a cupcake and it never notices. Every final ranking and every non-conference loss, against teams just as strong.

Share

The answer · FBS, 2022-2025

No. A tough non-conference schedule moves a team's final ranking by less than one spot either way, in the AP poll and in the playoff committee's. What moves a team up is beating good teams: each win over a top-25 team is worth about 1.6 committee spots.

Does a tough schedule cost you?

Each row is the change in a team's final ranking from one of the softest non-conference schedules to one of the toughest, against teams just as strong. The shaded band is -1 to +1 spot; a bar is highlighted when its whole 95% interval sits inside it.

Final rankings

Final AP pollevery FBS team-season
+0.03
Committeefinal regular-season ranking
-0.12
AP, same recordlosses held equal
+0.55

Final AP poll: +0.03 spots (95% CI -0.62 to +0.72). Less than one spot either way - neither a penalty nor a reward.

Committee ranking: -0.12 spots (95% CI -0.73 to +0.54). Less than one spot either way - neither a penalty nor a reward.

AP poll, same record: +0.55 spots (95% CI -0.07 to +1.22). Unresolved - the interval reaches past one spot, so the data can't rule a small effect in or out.

MethodOrdinary least squares (OLS) regression with one row per FBS team-season (534): its points in the last regular-season AP Top 25 (0 if unranked) - or, for the committee, 26 minus its rank in the last Playoff Committee Rankings (0 if unranked) - on its final v4-state-space rating, strength-tier dummies (the team's final v4-state-space rank: 1-5, 6-10, 11-15, 16-25, 26-40, with 41+ as the baseline), a power-conference indicator (SEC, Big Ten, Big 12 and ACC, the Pac-12 through 2023, and Notre Dame) and its non-conference schedule strength (the average final v4-state-space rating of its non-conference opponents, an FCS opponent counted at that season's lowest FBS rating). "At the same record" adds its regular-season losses.

MethodEach schedule effect is read across the 10th to 90th percentile of non-conference schedule strength (38.7 to 49.7 rating points), in poll spots - AP points over 60.6 a spot. "Neither a penalty nor a reward" means that whole 95% interval sits inside -1 to +1 spots, a bound set after the first run, which found nothing; "unresolved" means the interval reaches past it.

Who you play, or who you beat?

Week to week, AP voters weigh who a ranked team played. By the end of the season, what's left is credit for the good teams it beat.

Week to week

After a loss

+100 pts

kept per 10 rating points of the winner

After a win

+48 pts

gained per 10 rating points of the loser

Losing to a better team costs less: for every 10 rating points of the team that beat it, a ranked team keeps 100 more AP points, about 2 spots (95% CI +16 to +185, p = 0.003, q = 0.006).

Beating a better team earns more: for every 10 rating points of the team it beat, a ranked team gains 48 more AP points, about 0.8 spots (95% CI +34 to +64, p < 0.001, q = 0.003).

MethodOLS regression with one row per non-conference game played by a team ranked in that week's AP Top 25: its points in the next poll (0 if it fell out) on its points before, the margin, the venue (+1 home, -1 away, 0 neutral) and the opponent's final v4-state-space rating, per 10 rating points. Losses (63) and wins (329) are fitted separately.

By the end of the season

Each top-25 win

+1.6 spots

in the committee's ranking

Blue-blood name

-0.4 spots

once wins are counted

Each win over a team in the top 25 is worth +1.6 committee spots (95% CI +0.99 to +2.08, p < 0.001, q = 0.002) and +97 AP points, about 2 spots (95% CI +60 to +127, p < 0.001, q = 0.002) - with strength, record and schedule held equal.

By season's end the name adds nothing clear once wins over good teams are counted: -0.4 committee spots (95% CI -1.83 to +1.08, p = 0.407, q = 0.407) and -31 AP points, about 0.5 spots (95% CI -122 to +49, p = 0.278, q = 0.334).

AP, with wins countedschedule, soft to tough
+0.52
Committee, with wins countedschedule, soft to tough
+0.35

Schedule, with wins counted (committee): +0.35 spots (95% CI -0.19 to +0.90). Less than one spot either way - neither a penalty nor a reward.

MethodThe two final-ranking regressions with regular-season losses, wins over teams in v4-state-space's final top 25 (conference games included), a blue-blood indicator (Alabama, Ohio State, Michigan, Notre Dame, USC, Texas, Oklahoma, Georgia, LSU, Penn State, Florida State, Tennessee, Florida, Ole Miss and Oregon) and schedule strength added, each of the last three tested with the others held fixed. These six tests get their own Benjamini-Hochberg adjustment.

Texas 2025: would a cupcake have helped?

Texas lost to Ohio State out of conference and finished #13 in the committee's ranking, outside the #10 cut for the playoff field.

Where each loss left them in the AP poll

  • Week 1, at Ohio State (NovaModel #2), lost by 7: #1 to #7, -259 AP points
  • Week 6, at Florida (NovaModel #48), lost by 8: #9 to unranked, -1,071 AP points
  • Week 12, at Georgia (NovaModel #7), lost by 25: #10 to #17, -397 AP points

A cupcake instead of Ohio State

Strength tiers

#13

+0.42 spots

Smooth curve

#13

+0.05 spots

Swap Ohio State for a win over a cupcake and Texas finishes #13 in the committee's ranking with the strength tiers (95% #12-#13) and #13 with a smooth rating curve (95% #13). It finished #13 as played. Neither fit puts Texas inside the #10 cut for the playoff in any of 2,000 refits.

Methodv4-state-space replayed week by week from 2022 through 2025, twice: as played (it reproduces the stored ratings exactly) and with Ohio State's side of that game made an FCS team and the result flipped. Each team's committee score (26 minus its rank) then moves by the change in its predicted score from the committee regression with losses added - fitted once with the strength tiers and once with a cubic in the rating in their place, since crossing a tier line moves a prediction a long way. Texas's new rank is counted among every team, refitted on 2,000 bootstrap resamples. The cut is the lowest rank with every team at or above it in the playoff field.

Who gets the chances?

A contender from the Group of 5 played 1.0 games a season against top-25 teams; one from the SEC played 4.7. You can't get credit for beating good teams you don't play.

Contenders fromSeasonsTop-25 gamesNon-conf.Wins
Group of 571.00.70.3
ACC122.71.01.5
Big 12132.80.51.4
Notre Dame43.03.01.3
Big Ten233.40.41.7
Pac-1293.40.31.7
SEC324.70.52.3

MethodAmong team-seasons in v4-state-space's final top 25: the average number of regular-season games against other teams in that top 25, how many of those were non-conference, and wins in them. Teams outside the power conferences are grouped as Group of 5. A description, not a test - some groups are a handful of team-seasons.

Does it hold up?

The three final-ranking questions on the model's top 40 only, and with FBS opponents only. AP points or committee spots per 10 rating points of schedule; highlighted when the interval clears zero.

QuestionEvery teamTop 40FBS opponents
Final AP poll+2-34 to +40-8-109 to +105-3-33 to +28
AP poll, same record+30-4 to +67+43-26 to +116+20-10 to +50
Committee ranking-0.11-0.66 to +0.49-0.20-1.71 to +1.52-0.14-0.63 to +0.32

How this was done

Enough to recreate every number on this page. Each result's own method line above lists its rows, outcome and controls.

  • Data: AP Top 25 and Playoff Committee Rankings ballots and game results from CollegeFootballData (CFBD) for the 2022-2025 regular seasons (pulled with games through 2026-10-10): 2,145 team-games out of conference, 485 of them against FCS teams. A game is non-conference when the two teams were in different conferences that season, either was an FBS independent, or the opponent wasn't FBS. Team strength is NovaModel's v4-state-space rating at the end of the regular season: how good a team turned out to be.
  • Effects: every number is a coefficient from an ordinary least squares (OLS) regression with an intercept - AP points or committee spots per 10 rating points of schedule or opponent strength, read in spots where the page says so. Each section's method line lists its outcome and every control.
  • 95% confidence intervals: percentile cluster bootstrap. Teams are drawn with replacement 2,000 times, keeping every row of each drawn team; the regression is refitted on each draw, and the interval runs from the 2.5th to the 97.5th percentile.
  • p-values: Freedman-Lane permutation test. The regression is fitted without the tested term, its residuals shuffled across rows 2,000 times and added back to its fit, and the full regression refitted each time; p = (shuffles whose coefficient is at least as far from zero as the real one, plus 1) / (shuffles plus 1), two-sided.
  • Multiple comparisons: Benjamini-Hochberg false-discovery-rate adjustment (q-values), once across the five main questions (final AP poll, at the same record, committee, after a loss, after a win) and once across the six resume tests. A result is called real only when its 95% interval excludes zero and its q is under 0.05.
  • No effect: a schedule is called neither a penalty nor a reward only when the whole soft-to-tough 95% interval sits inside -1 to +1 poll spots. That bound was set after the first run.
  • Checks, reported but never part of the verdict: the three final-ranking questions on v4-state-space's final top 40 only, since most team-seasons are unranked zeros; and on FBS opponents only, with the number of FCS games as a control, since the FCS rating is a stand-in.
  • Poll spots: the average AP-points gap between #1 and #25, divided by 24, over every AP Top 25 release in the data (60.6 points a spot). Committee scores are spots already.
  • Random seed 42, for the bootstrap and the shuffles. The limit: rankings record what voters did, not why, and no data here shows who tried to schedule whom.