Football Lab: compare two offenses

Three inputs, one average

Ordinary YPP is the average positive gain below 20 yards. Breakaway rate is the percentage of all snaps assigned a 20+ yard gain. Negative play rate is the percentage assigned a loss. Zero-yard plays stay at 21%; the rest are ordinary gains.

Est. overall YPP estimates the yards you would actually record with these settings. After a slider change, 1,000 games run in the background against the fixed CFB average preset, with 12 possessions each and first possession alternating. Gains and losses stop at the goal lines. The readout is total recorded yards ÷ total offensive plays, including zero-yard plays and excluding kicks. Presets and manual settings use exactly the same calculation.

The same random seeds make repeated settings reproducible. The opponent is fixed so editing one team never changes the other team’s estimate. Actual games and batch results will vary with sampling and opponent-driven starting field position.

The underlying mean before goal-line clipping still follows directly from the sliders. With ordinary mean g, breakaway rate r, and negative rate n, expressed as fractions:

Uncapped mean = (0.79 − r − n)g + 38.81r − 3.88n

The overall readout uses simulated field positions to account for clipping; it does not just subtract a fixed number from this formula.

Fixed play shapes

Ordinary gains use a beta distribution on 0–20 yards with a fixed 4.5-yard SD. This approximates the ordinary-gain variability in the CFB sample. Breakaways retain the fitted CFB positive tail conditional on reaching 20 yards, capped at 99 yards; their average is about 38.8 yards. Recorded gains stop at the goal line.

Negative plays draw from a fixed empirical distribution of 8,583 text-verified CFB losses. About 68.7% lose 1–4 yards, 23.8% lose 5–9, and 7.5% lose 10 or more. The mean loss is 3.9 yards. Every profile uses this same size mix; only its frequency changes. Rare large losses remain in the sample, with no fitted extrapolation beyond the observed maximum of 45 yards.

A CFB baseline, with distinct styles

CFB average uses 6.0 ordinary YPP, 7.8% breakaways, and 8.4% negative plays. The five generic presets have ordinary gains reduced by 0.5 yards from their prior settings. The empirical reference remains 6.5 ordinary YPP; the rate inputs are unchanged. The reference comes from 59,708 own-half offensive snaps across 136 FBS teams in 2025, excluding kneels, spikes, no-play and offsetting records. Own-half plays reduce goal-line truncation.

The other presets deliberately strengthen the differences. Explosive doubles breakaways, keeps baseline losses, and lowers ordinary YPP to 5.0. Steady halves breakaways, keeps baseline losses, and raises ordinary YPP to 7.0. Boom or Bust doubles both rates: 15.6% breakaways and 16.8% negative plays, with 6.0 ordinary YPP. Its breakaway rate matches Explosive, but ordinary gains are stronger and losses are twice as frequent. Undisciplined keeps baseline breakaways, doubles losses, and raises ordinary YPP to 7.0.

The 1-yard ordinary-gain adjustments for Explosive and Steady make the short-gain versus medium-gain contrast visible. Boom or Bust keeps the adjusted baseline ordinary YPP, one yard above Explosive, providing an upside to its extra losses. Undisciplined receives a 1-yard boost: it has a capable offense interrupted by frequent setbacks. These choices do not target equal overall YPP or a scoring order. Multipliers refer to the CFB preset, so editing either team never changes the other.

The amplified profiles are illustrative experiments; their elevated rates extend beyond typical team ranges. Zero plays remain fixed at 21%, ordinary SD at 4.5 yards, and the positive and negative tail shapes are shared. Undisciplined describes frequent negative plays; actual penalties and replayed downs are not implemented. Applying one distribution at every down and field position is still a simplification.

2025 play-by-play · CFBD / SportsDataverse ↗ · Empirical baseline & team benchmarks ↗

Michigan season profiles

These are calibrated game representations informed by Michigan’s official statistics and public CFBD play-by-play, rather than literal measurements of each slider. 2023 emphasizes consistent gains and few mistakes; 2025 depends more on breakaways; 2026, through the Minnesota game, has fewer breakaways and more setbacks. The 2023 ordinary-gain boost stands in for consistency and finishing ability omitted by the simplified rules. The higher 2025 and 2026 negative rates also represent penalty-driven setbacks; actual penalties and replayed downs are not simulated. All three use the same fixed play shapes and zero rate as the generic presets. They do not reproduce every season statistic or team scoring average.

Official Michigan statistics: 2023 ↗ · 2025 ↗ · 2026 ↗

The house rules

• Start at own 25. Gain 10 yards in three offensive downs; first downs reset the count.
• On fourth down at the opponent’s 35 or closer, kick an automatic field goal for 3 points.
• Otherwise, from midfield or closer, punt to the opponent’s 10. From behind midfield, punt 40 yards. No return.
• Touchdowns earn 7 points. The opponent starts at own 25.
• A safety awards 2 points and possession at own 25 to the opponent.
• Exactly 12 possessions per team. No clock, overtime, turnovers, actual penalties, or defense effects; ties stand.

Using the controls

Both teams’ profile dropdowns and three sliders sit together beside the field on desktop. Play mix, box score, and experiments expand on demand. Related number boxes and overall YPP update live. Selecting a profile affects only that team. Editing a slider marks it Custom. Changing settings, including profiles mid-game, starts a new game.

Higher variance is not guaranteed to improve scoring. Consistent gains can prevent stalls; breakaways can rescue drives. The experiment measures the trade-off under these rules.