| Rank | Candidate | Team | Position | Model award probability | Games | Pass yards | Pass TD | INT | Rush yards | Rec. yards | Rush / rec. TD |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Brock Purdy |  SF | QB | 29.0% | 14.6[9.0–17.0] | 3,975[2,312–5,180] | 35.0[22.0–45.0] | 9.7[4.0–15.0] | 423[176–749] | 0[0–0] | 2.4 / 0.0 |
| 2 | Jared Goff |  DET | QB | 12.4% | 14.6[9.0–17.0] | 3,691[2,249–4,771] | 31.9[20.0–41.0] | 3.9[1.0–7.0] | 185[51–366] | 0[0–0] | 1.5 / 0.0 |
| 3 | Lamar Jackson |  BAL | QB | 10.7% | 14.5[9.0–17.0] | 4,006[2,287–5,257] | 26.1[15.0–35.0] | 8.3[4.0–13.0] | 636[254–1,146] | 0[0–0] | 3.9 / 0.0 |
| 4 | Josh Allen |  BUF | QB | 7.9% | 14.5[9.0–17.0] | 3,897[2,313–5,055] | 21.0[12.0–28.0] | 10.6[5.0–15.0] | 618[254–1,075] | 1[1–1] | 11.9 / 0.0 |
| 5 | Patrick Mahomes |  KC | QB | 6.5% | 14.5[9.0–17.0] | 3,675[2,206–4,764] | 28.9[17.0–38.0] | 10.3[5.0–15.0] | 360[121–678] | 0[0–0] | 2.7 / 0.0 |
| 6 | Matthew Stafford |  LA | QB | 6.3% | 14.6[9.0–17.0] | 3,823[2,359–4,906] | 33.0[19.0–43.0] | 12.9[8.0–18.0] | 136[38–265] | 0[0–0] | 1.0 / 0.0 |
| 7 | Joe Burrow |  CIN | QB | 5.7% | 14.5[9.0–17.0] | 3,456[2,077–4,474] | 31.9[19.0–42.0] | 6.7[3.0–10.0] | 220[75–411] | 0[0–0] | 0.9 / 0.0 |
| 8 | Dak Prescott |  DAL | QB | 4.0% | 14.5[9.0–17.0] | 3,529[2,080–4,611] | 28.3[17.0–37.0] | 6.9[3.0–11.0] | 307[122–550] | 0[0–0] | 1.6 / 0.0 |
| 9 | Trevor Lawrence |  JAX | QB | 2.8% | 14.6[9.0–17.0] | 3,398[1,996–4,425] | 28.3[17.0–37.0] | 10.8[6.0–16.0] | 332[111–615] | 0[0–0] | 1.8 / 0.0 |
| 10 | Jalen Hurts |  PHI | QB | 2.4% | 14.5[9.0–17.0] | 3,307[1,939–4,327] | 25.7[15.0–34.0] | 11.8[7.0–17.0] | 504[197–904] | 0[0–0] | 3.7 / 0.0 |
NFL · BBMI projections
NFL Award Projections
Our NFL season simulations project the award races. We publish each update and keep the record.
Through Week 3 · 10,000 simulated seasons · 48 observed games · 224 simulated games
Statistics cutoff: September 29, 2026 at 3:00:00 AM CDT. Week 4 games and the rest of the regular season were simulated.
Publicly published: . Development baseline at freeze. Generated after Week 4 began; this was not a forecast issued before Week 4.
Roster / availability: completed-week inputs; a separate roster timestamp was not archived for this original snapshot.
Most Valuable Player
Last updated:
Draws generated October 4, 2026 at 7:12:43 PM CDT · Award scored October 4, 2026 at 7:12:43 PM CDT
Primary: Blend. Supporting statistics use the primary projection.
Percentages are not rescaled to the displayed subset.
Forecast dates and frozen version
- Statistics cutoff
- September 29, 2026 at 3:00:00 AM CDT
- Primary draws generated
- October 4, 2026 at 7:12:43 PM CDT
- Primary award scored
- October 4, 2026 at 7:12:43 PM CDT
- Comparison draws generated
- October 4, 2026 at 7:14:05 PM CDT
- Comparison award scored
- October 4, 2026 at 7:14:05 PM CDT
- Publicly published
- October 4, 2026 at 8:47:53 PM CDT
- Frozen version
- 20261005T001509Z_mvp_v4
- Snapshot
- 2026-w03-20261005T010116596489Z
Methodology, validation and limitations
Award probabilities have not yet been validated for calibration.
These are estimated probabilities of receiving an award, not projected vote shares. The primary and comparison models are frozen. We score OPOY and OROY on the same archived player draws as the primary MVP forecast; MVP’s comparison retains its separately archived recent-usage draws.
Each player award is scored within each simulated candidate field, then averaged across 10,000 seasons. Player availability, production and team outcomes remain connected. Completed games stay fixed. Only games and statistics through the stated completed-week cutoff enter observed inputs.
Historical award-selection tests use final-season statistics with fitting and tuning on earlier seasons. Reused development seasons and conditional player tests do not establish calibrated in-season award forecasts. Full OPOY/OROY voting histories are incomplete.
Running-back yardage intervals under-cover, return-from-absence projections remain uncertain, and rookie projections without observed usage rely on generic position priors. OPOY’s comparisons across positions need further validation.
Coach of the Year’s performance measure is separate from its award forecast. Projected wins above talent equals schedule-adjusted projected wins minus the full-season wins implied by roster talent. Surplus uses an average-schedule basis; the voter models use actual-schedule wins and playoff outcomes.
Tables are ordered by unrounded primary probability. Percentages are displayed to one decimal without rescaling the displayed field. A displayed 0.0% is not proof of impossibility. CSV downloads preserve full precision and the full field. Supporting statistics describe the primary projection; brackets show its 10th–90th percentile range where provided.
We preserve cutoff, draw-generation, award-scoring and public publication times separately. Dated publications are immutable. Failed or incomplete captures leave the last successful publication and its original dates in place.