Data through Week 6
The per-play view, and every number is adjusted for who the team actually played. Higher is always better, defence included — a big offensive rating scores, a big defensive rating suppresses. EPA is expected points added per play; success rate is the share of plays that stay on schedule. For the same teams expressed in whole points rather than per play, see the Off and Def columns on Advanced.
Data through Week 6
Off and Def are Fathom's own split, in points: what the offence is worth above an average one and what the defence suppresses below an average one. They are opponent-adjusted, higher is better in both, and they add up to the team's Fathom rating exactly — the split says how a team is built, the rating says how good it is. Everything after them is raw per-game rate stats, not adjusted for opponent: a team that has played three overmatched opponents looks better on those than it is. For those columns lower is better on defence, higher on offence.
2026 recruiting cycle — preseason, not updated during the season
247Sports composite team talent and this cycle's recruiting class. Nothing on this site is built on these numbers. We tested whether adding talent to the Fathom rating improves early-season prediction: it gained 0.165 points of error, p = 0.19 — not significant — while talent on its own was 1.45 points worse than the rating (p = 6e-06). The rating gap and the talent gap already correlate at r = 0.617, so most of this is priced in by last season's results. It is here because it is interesting, not because it predicts.