Year One Tells You 60% of Everything You’ll Ever Know About a Coach

Fans want patience. Boosters want results. The math says the boosters are closer to right: a single 12-game season explains 60% of a coach’s true ability. By Year 3, the picture is 82% clear. The “5-year plan” is a $20 million way to learn what 12 games already told you.

By The Sports Page · Pre-Season 2026 · Simulation: 10,000 coaching careers

60.2%

Variance Explained by Year 1

r = .906

Correlation at Year 3

23.5%

Bad Coaches Who Look Good After Yr 1

The question every fan base asks when a new coach arrives: how long until we know? The conventional wisdom is five years — enough time to recruit “his guys,” install a system, and build a culture. Athletic directors write contracts around this assumption. Boosters fund facility upgrades based on it. And the data say it’s mostly wrong. We simulated 10,000 coaching careers using a Beta-Binomial model calibrated to the real distribution of college football coaching talent. Each coach has a “true” win rate drawn from a population where the average is .562 and the range spans .283 to .825. Each season is 12 games. The question: when does the observed record converge to the truth?

The answer is uncomfortable for the patience crowd: Year 1 alone explains 60.2% of the variance in a coach’s true ability (r = .776). By Year 2, the correlation reaches .867. By Year 3, it’s .906 — meaning over 82% of what you’ll ever know about this coach is already visible. Year 5 gets you to .940, but the marginal gain from Year 3 to Year 5 is just 6 percentage points. You’re paying $10 million per year for those 6 points. The 12-game college football season is a small sample — but coaching talent varies so enormously (from .300 to .900) that even 12 games is enough to separate the great from the terrible.

“The five-year plan is a statistical luxury no one can afford. By Year 3, the picture is 82% clear. The remaining 18% rarely changes the verdict.”

— The Sports Page, on coaching evaluation timelines

The Stabilization Curve: When You Know What You Have

Model: 10,000 simulated coaching careers True win rate: Beta(4.5, 3.5), mean .562 Season length: 12 games Career length: 15 seasons Correlation with true ability by year: Year 1 (12 games): r = .776 R² = 60.2% Year 2 (24 games): r = .867 R² = 75.2% Year 3 (36 games): r = .906 R² = 82.0% ← the sweet spot Year 5 (60 games): r = .940 R² = 88.3% Year 10 (120 games): r = .968 R² = 93.7% The jump from Year 1 to Year 3: +21.8 pts of R² The jump from Year 3 to Year 5: +6.3 pts of R² The jump from Year 5 to Year 10: +5.4 pts of R² Diminishing returns set in hard after Year 3.

The False Hope Problem: Bad Coaches Who Look Good Early

"Good coach" = true win rate ≥ .600 "Bad coach" = true win rate < .450 Misclassification rates (using .500 as the threshold): Good coach Bad coach looks bad looks good Year 1: 4.9% 23.5% ← DANGER ZONE Year 3: 1.2% 10.2% Year 5: 0.5% 6.6% After Year 1: • Only 5% of good coaches are below .500 — rare • But 24% of bad coaches are above .500 — common! → Year 1 rarely fools you about good coaches → Year 1 OFTEN fools you about bad coaches → This is why ADs give bad coaches extensions

The asymmetry is the key insight: if a coach goes 8-4 in Year 1, he’s probably good. But if he goes 7-5, he could easily be a .400 coach having a lucky year. Year 3 resolves most of this ambiguity.

The R² Stabilization Curve

Historical Parallels

Curt Cignetti — The Signal Was There From Day One

.795

Career win percentage across 4 schools, 15 seasons

IUP: .757. Elon: .609. JMU: .852. Indiana: .926. Cignetti has coached at four different schools across three divisions and his win rate has never dipped below .600. His Bayesian-estimated true win rate: .786, with a 95% confidence interval of .724 to .841. After just one season at IUP, the posterior already pegged him above .650. The signal was there from game one — just nobody outside Pennsylvania was looking.

True talent announces itself early

The 5-Year Extension Trap

$10M+

Average annual cost of a Power 5 coach

If Year 3 gives you 82% of the answer and Year 5 gives you 88%, the marginal cost of those extra two years at $10M+ per year is roughly $20M for 6 percentage points of certainty. That’s $3.3 million per percentage point. Athletic directors aren’t buying information with years 4 and 5 — they’re buying political cover. The data already told them the answer.

Years 4-5 are expensive confirmation bias

“Cignetti went 11-2 in his first year at Indiana — a program that went 3-9 the year before. That’s not luck. That’s a .795 career coach doing what .795 career coaches do. The only surprise was that anyone was surprised.”

— The Sports Page, on signal vs. noise in coaching hires