Matchup analysis
Before kickoff our model projected Western Michigan 32.6, Buffalo 18.6 at Broadview Stadium in Week 5. That is a solid edge: Western Michigan wins about 82% of the time by our estimate, with a projected margin of 14 points and 51.2 combined points.
Western Michigan is 2-2, scoring 24 and allowing 20 points per game (+4 per game). Buffalo is 2-2, scoring 21.3 and allowing 33.3 points per game (-12 per game).
The most lopsided unit battle: Western Michigan's below-average offense (90th of 138 in our opponent-adjusted ratings, worth about 24.2 points against an average defense) against Buffalo's struggling defense (133rd). Western Michigan's middle-of-the-pack defense (56th) should limit Buffalo's struggling offense (124th of 138 in our opponent-adjusted ratings, worth about 19.2 points against an average defense).
College football is noisy: even a well-rated projection misses the final margin by about 14 points on average. Based on how real results have scattered around our numbers, the final margin should land somewhere between Western Michigan by 37 and Buffalo by 9 about 80% of the time.
Buffalo gets about 2.1 points for playing at home. Western Michigan is still favoured on the road, so the ratings gap is bigger than the value of playing at home. The game is on ESPN+.
Result: Western Michigan 20, Buffalo 17. The model had the right winner; the margin was off by 11 points and the total by 14.2.
Market lines
Market lines · Source: The Odds API, ESPN, Action Network · Updated Oct 29, 9:25 AM UTC. Shown for information; not an input to our projections.
| Provider | Spread (open → now) | Total (open → now) | Moneyline (Western Michigan / Buffalo) |
|---|---|---|---|
| Consensus | Western Michigan -15.75 → Western Michigan -13.25 | 50 → 46.25 | -592.5 / +425 |
| DraftKings | Western Michigan -16.5 → Western Michigan -13.5 | 51.5 → 46.5 | -535 / +400 |
| Bovada | Western Michigan -15 → Western Michigan -13 | 48.5 → 46 | -650 / +450 |
This game's lines: CollegeFootballData.com · read Oct 22, 6:37 PM UTC.
| Projection | Margin | Total | Buffalo win |
|---|---|---|---|
| Our model | Western Michigan -14 | 51.2 | 18% |
| Market consensus | Western Michigan -13 | 46.3 | 18% |
| Blended projection | Western Michigan -13.5 | 47.6 | 18% |
Blended projection = 0.16 x our margin + 0.84 x the market margin, weights fitted on past seasons (see methodology).
Model vs market
- Buffalo win probability
- model 18% · market (no-vig) 18% · difference -0 pts
- Margin
- model Western Michigan -14 · market Western Michigan -13.25 · difference -0.7 pts (Buffalo view)
- Total
- model 51.2 · market 46.25 · difference +5.0
Model margin is 0.7 pts more toward Western Michigan than the market.
21+ only where legal. Gambling problem? Call 1-800-GAMBLER.
Original projection vs new formula
| Projection | Score | Win prob | Margin miss | Total miss |
|---|---|---|---|---|
| Original projection | Western Michigan 32.6, Buffalo 18.6 | Western Michigan 82%right winner | 11.0 | 14.2 |
| New formula | Western Michigan 33.2, Buffalo 19.4 | Western Michigan 82%right winner | 10.7 | 15.6 |
Final: Western Michigan 20, Buffalo 17. Original = the number we posted before kickoff. New formula (2026-10-21 matchup terms) = the current model re-run with only the games played before this one.
Western Michigan vs Buffalo: stat comparison
Season to date before kickoff: these numbers are frozen at kickoff and do not include this game or anything after it.
| Stat | Western Michigan | Buffalo |
|---|---|---|
| Points per game | 24.0#98 | 21.3#117 |
| Points allowed per game | 20.0#46 | 33.3#126 |
| Yards per play | 5.14#123 | 4.39#137 |
| Yards per play allowed | 5.27#62 | 6.51#122 |
| Rushing yards per game | 166.3#76 | 166.8#74 |
| Net passing yards per game | 138.3#131 | 161.3#121 |
| Rushing yards allowed per game | 153.0#82 | 173.5#107 |
| Net passing yards allowed per game | 191.0#59 | 264.5#127 |
| Turnovers per game | 0.25#26 | 1.75#131 |
| Takeaways per game | 0.25#132 | 0.25#131 |
| Sacks per game | 0.75#134 | 1.00#124 |
| Sacks allowed per game | 1.50#68 | 1.75#77 |
| Stat | Western Michigan | Buffalo |
|---|---|---|
| EPA per play | +0.00#110 | -0.09#130 |
| EPA per play allowed | +0.11#103 | +0.04#69 |
| Success rate | 39.9%#110 | 33.8%#131 |
| Success rate allowed | 44.2%#109 | 36.1%#26 |
| Explosive play rate | 9.4%#130 | 8.9%#135 |
| Explosive plays allowed | 10.3%#43 | 13.6%#105 |
| Havoc rate | 7.6%#138 | 13.6%#81 |
| Havoc allowed | 10.8%#26 | 12.7%#65 |
| Points per scoring opportunity | 4.57#40 | 4.45#50 |
| Points per opportunity allowed | 4.35#108 | 4.14#91 |
| Average starting field position | own 32#42 | own 27#112 |
| Opponents' starting field position | own 26#22 | own 32#116 |
| Offensive plays per game (pace) | 59.3#121 | 74.8#9 |
| Stat | Western Michigan | Buffalo |
|---|---|---|
| Rushing EPA per play | +0.12#50 | +0.01#105 |
| Passing EPA per play | -0.16#130 | -0.18#131 |
| Rushing EPA allowed | +0.08#100 | -0.01#51 |
| Passing EPA allowed | +0.14#99 | +0.07#87 |
| Rushing success rate | 42.0%#94 | 34.6%#129 |
| Passing success rate | 36.9%#116 | 32.2%#131 |
| Run stuff rate | 14.2%#34 | 16.7%#72 |
| Passing-down success rate | 31.3%#72 | 23.9%#126 |
| Passing-down success allowed | 40.3%#132 | 27.2%#55 |
| Third-down conversion rate | 41.2%#77 | 36.1%#107 |
| Third-down rate allowed | 46.3%#124 | 33.9%#40 |
| Red-zone touchdown rate | 75.0%#24 | 66.4%#67 |
| Red-zone TD rate allowed | 61.5%#85 | 67.2%#101 |
# = rank among 138 FBS teams. Efficiency numbers exclude garbage time. Per-play numbers of teams with a weak schedule are pulled toward their opponent-adjusted values (fully at one standard deviation below an average schedule). Source: play-by-play (cfbfastR / sportsdataverse), aggregated by us.
More: Western Michigan schedule · Buffalo schedule · all Week 5 projections
Share this prediction
Projections are statistical estimates for entertainment and analysis only. They are not betting advice and nothing here guarantees any result.
Free weekly projections email
One email a week in season: the closest projected games, the biggest mismatches and how the model graded out last week. Email address only — no spam, unsubscribe any time.