Work with numbers that show their work.
Every benchmark, target range and rule in Sightline comes from a published source or a complete public dataset, and the product shows a gap where none exists rather than a guess. These are the studies behind the figures — what each number is, how it was built, and what it does not claim.
What a college team actually looks like, by level.
The measured 2026 population every Team Strength Profile is scored against. Lower divisions hit for a higher average and give up more runs than Division I, which is exactly why a national average is the wrong yardstick.
The spread of team batting average across every college level, 2026. Lower divisions hit for a higher average than Division I — which is why scoring a team against a national average, instead of against its own level, flatters some programs and punishes others.
The same picture from the pitching side. A 5.50 team ERA is a below-average Power-4 staff and an above-average one almost anywhere else — the single clearest reason a recruiting board needs level-aware benchmarks.
Below the NCAA, most programs publish no one to write to.
Sightline's directory covers every college program in the country, so the share that publishes a coaching email is something we can simply count. It is near-total across the NCAA divisions and thins out sharply below them — the levels where a recruit usually needs the most help finding a name.
Counted from the live table on every page load, not stated once and left to go stale.
Open the directory →Counted live from Sightline's directory: 1,021 of 1,623 programs publish at least one coaching email. Coverage is near-complete across the NCAA divisions and falls away below them — which is where a recruit typically has the least help finding anyone to write to. A published address is a fact about the program, not a sign it is recruiting anyone.
How the research is held to account
Every figure on this page is derived in code from the same constant the product runs on, and a test fails the build if a study's headline number stops matching it. Recruit targets are locked to their research file row by row. Where a source gives a floor with no ceiling, the product shows a floor. Where nothing was published, it shows a gap.
The valuation engine's own method — inputs, level handling, confidence, and its disclosed limits — is written up separately. How valuation works →