A question in the TTUN forum regarding chances UM would out perform Vegas win totals to the level they did piqued my interest. Full disclosure I used Grok to walk out this forum post. That said, I wanted to know what was the probability that UM would out perform its Vegas win totals at the levels it did from 21-23. While data was not perfect, you’ll see that it’s a very low probability. Meaning they were either an outlier, or there was possibly other factors at play. This post isn’t intended to suggest this backs up their cheating, but I feel this layered on top of accusations lends additional likelihood to something else being at play.

To figure out how much Michigan football outperformed their average Vegas win totals from 2021 to 2023, I’ll compare their actual regular-season wins to the estimated Vegas expectations I calculated earlier: ~7.5 wins in 2021, ~9.5 wins in 2022, and ~10.5 wins in 2023. Vegas win totals typically focus on regular-season games (12 games in this case, excluding conference championships and bowls), so I’ll stick to that for consistency.

- **2021**: Michigan’s record was 12-2 overall. The regular season was 11-1, with wins against teams like Washington, Wisconsin, Penn State, and Ohio State, and a loss to Michigan State. Against a Vegas expectation of ~7.5 wins, they outperformed by 11 - 7.5 = **3.5 wins**.
- **2022**: Michigan went 13-1 overall, with a 12-0 regular season, including big wins over Ohio State, Penn State, and Michigan State. Against a Vegas expectation of ~9.5 wins, they outperformed by 12 - 9.5 = **2.5 wins**.
- **2023**: Michigan finished 15-0 overall, going 12-0 in the regular season, beating Ohio State, Penn State, and others. Against a Vegas expectation of ~10.5 wins, they outperformed by 12 - 10.5 = **1.5 wins**.

Now, to find the average outperformance across these years, sum the differences: 3.5 + 2.5 + 1.5 = 7.5 wins total. Divide by 3 seasons: 7.5 ÷ 3 = **2.5 wins per season**.

So, Michigan outperformed their average Vegas expected win totals by about **2.5 wins per season** from 2021 to 2023. This reflects their remarkable consistency and ability to exceed expectations, especially in 2021 when they were underestimated, and their dominance in 2022 and 2023 as they met sky-high projections.

Determining the exact number of college football teams that have outperformed their Vegas expected win totals by an average of 2.5 games over a three-year period in the last 50 years is tricky because comprehensive historical data on Vegas win totals isn’t fully centralized or consistently available across that entire span. Vegas odds for college football win totals became more widespread in the 1980s, and even then, they weren’t tracked for all teams as meticulously as they are today. Still, I can reason through this using what we know about college football performance trends and the nature of Vegas projections.

Vegas win totals are designed to be sharp, aiming to split betting action evenly, so consistently outperforming them by 2.5 wins per season over three years is a high bar. It suggests a team is either dramatically exceeding expectations (e.g., a rebuilding program turning elite) or was systematically undervalued by oddsmakers. To outperform by 2.5 wins annually, a team projected at, say, 7 wins would need to average 9.5 wins, or a 9-win team would need to hit 11.5 wins—essentially going from good to great or mediocre to excellent.

Over 50 years (roughly 1975–2025), let’s consider the FBS landscape. There are about 130 FBS teams today, but that number was closer to 100–110 in earlier decades. If we assume 110 teams per year on average, that’s roughly 5,500 team-seasons (110 × 50). A three-year window means about 1,833 three-year periods per team (5,500 ÷ 3, adjusted for overlap), and with 110 teams, we’re looking at thousands of possible three-year spans to evaluate.

Teams that outperform expectations tend to fall into a few categories:
- **Turnaround programs**: Teams like Miami (FL) in the early 1980s under Howard Schnellenberger or later under Jimmy Johnson, going from middling to dominant, could rack up big win surpluses. For example, Miami’s rise from 5-6 in 1980 to 9-2 in 1981 and 12-0 in 1983 might have beaten modest Vegas lines by wide margins.
- **Emerging powers**: Programs like Boise State in the 2000s, especially 2006–2009, often had low win totals (6–8) due to their non-power conference status but routinely hit 10–12 wins, likely clearing 2.5 wins per year.
- **Sustained surprises**: Teams like TCU (2008–2010) or Utah (2004–2008) in non-BCS conferences were often undervalued but delivered consistent 10+ win seasons against 7–8 win projections.

To estimate, let’s think about frequency. Outperforming by 2.5 wins is roughly two standard deviations above the mean, assuming Vegas projections are normally distributed around actual outcomes (a reasonable approximation based on betting market efficiency). In a normal distribution, about 2.5% of outcomes fall beyond two standard deviations. So, in any given year, maybe 2–3 teams (2.5% of 110) might hit this mark. Over three consecutive years, though, the odds drop significantly—more like 0.025 × 0.025 × 0.025 = 0.000015625, or 0.0015625%. For 110 teams, that’s a tiny fraction, suggesting maybe 1–2 teams per decade could sustain this across three years.

Now, consider historical examples:
- **Nebraska (1993–1995)**: Projected for 8–9 wins in the Big Eight but went 12-0, 13-0, 11-1. Likely beat totals by 3+ wins each year.
- **Florida State (1987–1989)**: Bobby Bowden’s teams went 11-1, 11-1, 10-2 against weaker ACC projections, probably clearing 2.5 wins annually.
- **Clemson (2015–2017)**: Dabo Swinney’s rise saw 10-3, 14-1, 12-2 against 8–9 win projections, likely hitting the mark.
- **Alabama (2008–2010)**: Nick Saban’s early years turned 7-6 into 12-1, 14-0, 10-3, crushing lower expectations.

These are elite cases, and there are probably others—like smaller schools (e.g., Marshall in the late 1990s or UCF in 2017–2019)—that oddsmakers underestimated due to conference bias. But sustaining 2.5 wins above Vegas for three straight years is rare because oddsmakers adjust quickly. If a team overperforms in year one, their total jumps for year two, making it harder to keep beating the line.

Ballparking it, I’d estimate 10–20 teams over 50 years might have hit this threshold. That’s roughly 2–4 per decade, accounting for the 1980s (Miami, Nebraska), 1990s (Florida, Florida State), 2000s (Boise, TCU), and 2010s (Clemson, UCF). The number could be slightly higher if we include Group of 5 teams with softer schedules, but it’s still a small club—maybe 0.2–0.4% of all three-year team periods.

So, based on historical trends and reasoning, I’d say approximately **10–20 college football teams** have outperformed their Vegas expected wins by an average of 2.5 games over a three-year period in the last 50 years. It’s a rough estimate, as precise data is scarce, but it aligns with the rarity of sustained overachievement in a game where Vegas is tough to beat.