Troy
at James Madison
The score estimate, the unit production and the questions to take into the film room.
75.9% home win estimate · 50.6 total
Read the range. The nominal 80% home-margin range is -11.0 to +36.5 points. It includes a win by either team. The range and probability curve were calibrated on 2024 outcomes.
Know the inputs. The production model uses team identities, prior scores and home field. The unit and player statistics below supply context; they do not alter this prediction. Transfers, injuries, weather and coaching changes are not explicit model features.
Put the production side by side.
Play-weighted source rates against each team’s own opponents. These are unadjusted historical observations, not a head-to-head forecast or a new input to the score model. Defensive figures describe opponent production; lower EPA allowed is better.
2025 team statistics · FBS opponents only · Source retrieved Sep 9, 2026.
Troy vs James Madison
| Measure | TroyOffense produced | James MadisonDefense allowed |
|---|---|---|
| EPA / play | -0.08513 games · 843 recorded plays | -0.13813 games · 783 recorded plays |
| Pass EPA / play | -0.06313 games · 462 recorded plays | -0.15313 games · 429 recorded plays |
| Rush EPA / play | -0.08213 games · 378 recorded plays | -0.11613 games · 352 recorded plays |
| Yards / play | 4.6713 games · 843 recorded plays | 4.6713 games · 783 recorded plays |
| Explosive play share | 7.1%13 games · 843 recorded plays | 7.2%13 games · 783 recorded plays |
| Stuffed rush share | 20.1%13 games · 378 recorded plays | 16.5%13 games · 352 recorded plays |
Troy: 13 team records from 13 scheduled finals; 0 finals missing team data.
James Madison: 13 games with opponent records from 13 scheduled finals.
Each measure uses its own available plays. A missing rate is not zero. Stuffed-run share uses source-classified stuffed carries; higher is better for the defense.
James Madison vs Troy
| Measure | James MadisonOffense produced | TroyDefense allowed |
|---|---|---|
| EPA / play | 0.16413 games · 895 recorded plays | -0.00213 games · 860 recorded plays |
| Pass EPA / play | 0.13013 games · 377 recorded plays | -0.03913 games · 387 recorded plays |
| Rush EPA / play | 0.21613 games · 514 recorded plays | 0.05013 games · 470 recorded plays |
| Yards / play | 6.5113 games · 895 recorded plays | 5.8013 games · 860 recorded plays |
| Explosive play share | 8.3%13 games · 895 recorded plays | 8.1%13 games · 860 recorded plays |
| Stuffed rush share | 12.1%13 games · 514 recorded plays | 12.8%13 games · 470 recorded plays |
James Madison: 13 team records from 13 scheduled finals; 0 finals missing team data.
Troy: 13 games with opponent records from 13 scheduled finals.
Each measure uses its own available plays. A missing rate is not zero. Stuffed-run share uses source-classified stuffed carries; higher is better for the defense.
Metric definitions and film questions
EPA / play
Offensive expected points added divided by recorded scrimmage plays. Source EPA totals are rounded; this is unadjusted production.
On film: Which downs and field positions account for the efficiency? Review the game log before assigning a coverage.
Pass EPA / play
Passing EPA divided by source pass plays; these are not necessarily official pass attempts.
On film: Separate protection, pressure and coverage on film. Team passing EPA does not identify which player or scheme created the result.
Rush EPA / play
Rushing EPA divided by source rushing plays.
On film: Check box counts, blocking and quarterback run involvement. These totals do not establish a front or gap tendency.
Yards / play
Source offensive yards divided by scrimmage plays. Do not substitute passing plus rushing yards for the published offensive total.
On film: Separate sustained gains from a few long plays, and inspect opponent quality before comparing averages.
Explosive play share
Source-classified explosive plays divided by scrimmage plays. EPA_explosive is a count, not EPA. This is our stated denominator, not a claim to reproduce every source rate or a fixed yardage threshold.
On film: Review the source-classified explosive plays on film. This measure does not establish a fixed yardage threshold.
Stuffed rush share
Source stuffed carries divided by rushing plays. Lower is favorable for an offense; higher is favorable for the defending team.
On film: Check where runs break down and which fronts were involved. A team rate is not an individual blocking grade.
Who supplied the production?
Up to two qualified players per program and category, ordered by total source EPA. Minimums: 100 passing, 50 rushing or 30 receiving plays. These 2025 affiliations do not verify current rosters, eligibility, availability or starting roles. Player totals include all their recorded opponents and do not follow the team-sample filter above. Never add EPA across overlapping categories.
Troy
| Player / category | Plays | Total EPA | EPA / play |
|---|---|---|---|
| Tucker Kilcreasepassing · FBS rank 105 | 215 | 6.07 | 0.024 |
| Goose Crowderpassing · FBS rank 140 | 166 | -26.66 | -0.139 |
| Tucker Kilcreaserushing · FBS rank 227 | 85 | 2.60 | 0.031 |
| Tae Meadowsrushing · FBS rank 357 | 133 | -10.92 | -0.082 |
| Tray Taylorreceiving · FBS rank 147 | 46 | 31.55 | 0.686 |
| RaRa Thomasreceiving · FBS rank 170 | 68 | 29.42 | 0.433 |
James Madison
| Player / category | Plays | Total EPA | EPA / play |
|---|---|---|---|
| Alonza Barnett IIIpassing · FBS rank 57 | 338 | 38.87 | 0.107 |
| Wayne Knightrushing · FBS rank 36 | 198 | 35.96 | 0.182 |
| Alonza Barnett IIIrushing · FBS rank 41 | 101 | 33.65 | 0.333 |
| Landon Ellisreceiving · FBS rank 72 | 53 | 43.08 | 0.813 |
| Nick DeGennaroreceiving · FBS rank 246 | 48 | 21.80 | 0.454 |
Source receipts and linked data editions
SportsDataverse bulk releases, publisher-stated CC BY 4.0. Silvermine aggregates the team rates and selects historical player leaders. Neither collection time nor this brief’s publication is backdated to the games described.
Team edition: 9b220906aa39e5ffafc171a15995b28beba495cfcf257902d7dc9f2e747dcdcb
Player catalog edition: e1999e227eb03908787ef9af2761a62d35f5fee930102a9140b46d2afba3cce0
Download 2025 player statistics ↗ · SHA-256: cc68ee5060ff517c90e79389bd5d901c6f1b0bc1648b8b5735bd8f62f051500d
- 2025 / schedule · Retrieved 2026-09-09T00:21:12.578400Z
bc8fb0df43bf5e3b2f8a1973a78ca909901b4370e3a690cab6e2c62cf8290aca - 2025 / team_advanced · Retrieved 2026-09-09T00:21:22.052664Z
63d100ace111c00e0171a958974d2f6f589c0cfcef4834a113232b0a734be77f - 2025 / teams · Retrieved 2026-09-05T05:20:51.123196Z
5446f91a8a0d27e112fa7369ba0e887ed27aa6eec61130e00f20339ef50e7abf - 2026 / schedule · Retrieved 2026-09-12T13:29:39.244014Z
5b6f62a42001c37b2471e094b5c251c7ee6d99f6e93c8ff452fc4d90e4e9bcf2 - 2026 / team_advanced · Retrieved 2026-09-12T13:29:47.262896Z
debe1ef4fbb03863d319980bc87ec69c9ae0e198b72848c1ec8c2a1b32a953b2 - 2026 / teams · Retrieved 2026-09-05T05:21:00.957219Z
6d3ec4bd262943dfa3777e4c31ece0a20eac2bb6194ef20a84c165b7cd2240fd - 2025 / box · Retrieved 2026-09-09T00:21:14.853561Z
0f41a1ce9cb76fdaefc87d694d15b3255db02c569ac0e8ff2de7c9c998208cc2 - 2025 / passing · Retrieved 2026-09-09T00:21:17.096058Z
b8dab38e8485c565b871fb5e7fe5338a70dc39bdcafb95b5941089cd71aa9b38 - 2025 / receiving · Retrieved 2026-09-09T00:21:19.043912Z
302c53ac4c2316f1d9611e80e06f98186f176e4feb8a4531ef4290841ee6aac5 - 2025 / rushing · Retrieved 2026-09-09T00:21:18.042034Z
c4ed1fb90a293b29477046419f59f8a1a6d41adc21a79cc11b255b48b0e3ed30 - 2025 / schedule · Retrieved 2026-09-09T00:21:12.578400Z
bc8fb0df43bf5e3b2f8a1973a78ca909901b4370e3a690cab6e2c62cf8290aca - 2025 / teams · Retrieved 2026-09-05T05:20:51.123196Z
5446f91a8a0d27e112fa7369ba0e887ed27aa6eec61130e00f20339ef50e7abf
Check the record before the edge.
The game record keeps registered forecast versions, source-state observations and qualifying line comparisons. The prospective scorecard selects the earliest eligible registration; that can be an older model version than this brief.
The separate imported archive has no line for this game. See the dated game record for any qualifying feed observations; an absent quote is not replaced with a guessed price.
A model home margin of +5 and a home spread of −3 differ by +2 points toward the home side. This describes disagreement, not a betting return.
Retained reading snapshots of this brief →
How much weight to give this forecast
The ridge-team-calibrated-v2 model was evaluated on 784 games from 2025, fitted using earlier seasons. Mean absolute margin error was 14.2 points. These are retrospective test results, separate from the prospective record.
Model: ridge-team-calibrated-v2-b4a3a7280415 · Forecast cutoff: 2026-09-12T13:29:53.576737Z. Schedule and source statistics: SportsDataverse, CC BY 4.0. Silvermine supplies the independent model, aggregations and generated commentary.
Model methodology · Weekly model experiment · All football matchups