READ THE EVIDENCE.
Advanced football analysis in fan language, with every number labeled by what it knows, where it came from, and what it does not claim.
Four labels.
Four different claims.
The label travels with the number so recruiting reputation, observed production, projections, and market context never collapse into one misleading grade.
Observed game, drive, play, or player production from a completed season.
A published preseason role or lineup estimate. It is not an official Michigan depth chart.
A current roster, schedule, staff, or status fact published before games begin.
External context such as recruiting consensus or a sourced market price—not game performance.
What the numbers mean.
These are public-language definitions. The underlying analytics system and metric registry remain the calculation source of truth.
Success rate
The share of plays that gained enough yards for the down and distance. Higher is better.
Rush success rate
Success rate on designed rushing attempts. Quarterback scrambles are not silently treated as called runs.
Explosive run rate
The share of rushing attempts gaining at least 10 yards. Higher means more breakaway production.
Points per opportunity
Points scored per possession that reaches the opponent's 40-yard line. It measures whether promising field position becomes points.
Points per resolved possession
Points per possession with a known competitive result. It is broader than scoring-opportunity efficiency.
Position percentile
A player's standing among the published national FBS comparison cohort at the same position and above the required usage threshold.
Production grade
A display grade derived from measured 2025 usage and production within a positional cohort. It describes the season; it does not forecast the next one.
Prospect tier
A display translation of recruiting consensus. It describes incoming pedigree, not college performance.
Market win chance
A historical calibration from a sourced point spread. It is market context, not a Michigan Football Focus game prediction.
What the site knows today.
2026 roster, staff, schedule and roles: Preseason evidence.
2025 team performance: Completed-season actuals used as historical context.
Recruiting: Consensus benchmarks, not college proof.
Markets: Sourced prices and historical spread calibration.
2026 in-season analytics: Begin only after observed games are published.
What we will not do.
We do not invent rankings, grades, player roles, probabilities, or historical comparisons when an artifact is missing.
Prior-season performance and staff context inform expectations; they do not guarantee that Michigan will reproduce another team's tendencies or results.
Public language may simplify a calculation, but it does not change its denominator, direction, or underlying definition.

