Projected win totals and ratings, fixed before the season — they do not move as games are played. 2026 rankings as they stand today →
Preseason rating = last season's SP+ regressed toward the mean, nudged by returning production, recruited talent and recruiting-class strength — the weights LEARNED from 10 seasons by walk-forward, not guessed. The projected-win range is a per-team ±1σ band: game randomness plus our season-long rating error (a shared shock across all of a team's games) — so teams with more coin-flip games get a wider band.
We tested recruiting-class strength and transfer-portal movement on top of the base (last-year rating + returning production + talent). Recruiting IMPROVED the backtest (8.08→7.91 rating RMSE) and is included; the transfer portal did NOT add signal beyond talent + recruiting (8.08 RMSE) and is honestly left out.
| Feature set tested | Corr | RMSE |
|---|---|---|
| base (carry+ret+talent) | 0.784 | 8.08 |
| + recruiting ✓ chosen | 0.792 | 7.91 |
| + portal | 0.784 | 8.08 |
| + recruiting + portal | 0.792 | 7.92 |
| # | Team | Outlook | Rating | Projected wins | Proj | '25 SP+ |
|---|---|---|---|---|---|---|
| 51 |  Western Kentucky | Bowl team | +3.4±7.7 | 6.2 10.0 | 8.1 ±1.9 | +1.6 |
| 73 |  Louisiana Tech | Bowl team | -0.2±7.8 | 4.9 9.6 | 7.2 ±2.3 | -1.3 |
| 91 |  Jacksonville State | Bowl team | -4.3±7.9 | 5.5 9.9 | 7.7 ±2.2 | -6.7 |
| 95 |  Delaware | Bubble | -5.1±7.7 | 4.2 8.8 | 6.5 ±2.3 | -10.9 |
| 102 |  Liberty | Bowl team | -7±7.9 | 4.5 9.2 | 6.8 ±2.3 | -9 |
| 109 |  Kennesaw State | Bubble | -7.6±8 | 3.9 8.6 | 6.2 ±2.3 | -5.4 |
| 112 |  Florida International | Bubble | -8.3±7.9 | 4.0 8.7 | 6.3 ±2.4 | -10.5 |
| 115 |  Missouri State | Bubble | -9.6±7.9 | 3.0 7.3 | 5.1 ±2.1 | -10.7 |
| 125 |  Middle Tennessee | Bubble | -12.1±7.9 | 2.8 7.4 | 5.1 ±2.3 | -16 |
| 129 |  New Mexico State | Rebuild | -14±8 | 2.1 6.6 | 4.3 ±2.2 | -15.5 |
| 130 |  UTEP | Rebuild | -14.8±8.1 | 2.6 6.7 | 4.6 ±2.0 | -17.5 |
| 134 |  Sam Houston | Rebuild | -17.8±7.8 | 1.7 5.8 | 3.7 ±2.0 | -27.8 |
Method: next-season rating is last year's SP+ regressed toward the mean (carryover 62%, learned from 2019+), plus returning production and recruited talent when published. The projected-win range is a per-team ±1σ band over the real 2026 schedule: game randomness plus our season-long rating error, which moves all of a team's games together — so a team with more toss-up games gets a wider band, and a clear favorite (or a clear also-ran) gets a tighter one. The rating ± is the walk-forward residual spread; it's the same for every established team because next-season predictability doesn't reliably differ between them (we tested — it will diverge for data-poor teams and once returning production publishes). Backtest is strictly walk-forward (fit on prior seasons only). This is a season-level outlook, not a per-game spread. Source: CollegeFootballData (SP+, returning production, talent, schedule).
