Hit isn't one skill.
Bat-to-ball ability and swing decisions need to be modeled separately. A single 55 or 60 Hit grade can hide very different failure modes.
S80 Research Lab
Show the work.
Keep the misses.
S80 isn't built by picking weights that sound smart. We reconstruct what was known when a player was still a prospect, preserve the source and date, then compare the projection signals with what actually happened in Major League Baseball.
50players studied
2015–21draft cohorts
39mature cases
28/39tool-only model off ≥1 tier
Backtest #001
What broke the
five-tool average?
In our first diagnostic, a crude five-tool expected tier either beat or missed the realized outcome by at least one tier in 28 of 39 mature cases. That's not a finished statistical result—it is a warning that averaging tools hides important information.
Raw power isn't game power.
S80 now distinguishes the ability to generate raw strength from the ability to access damage against real pitching.
Position changes the bar.
A first baseman and a shortstop cannot be judged against the same offensive threshold. Defensive context changes both ceiling and floor.
Trajectory is information.
Grade movement, level progression and approach changes are signals. A prospect improving rapidly is not the same prospect he was twelve months ago.
Risk ≠ talent.
Injury and sample uncertainty now affect Confidence and the width of the projection band instead of simply subtracting talent points.
Bottlenecks matter.
A 70 arm and 65 power cannot average away a bat that may not survive MLB pitching. S80 v2 can cap a profile when contact becomes the limiting skill.
Source provenance
One player.
Many lenses.
S80's long-term database is designed to retain the source and timestamp for every important input. We want to know not only the grade, but who published it and what was knowable at that moment.
MLB Pipeline / MLB.com20–80 tool grades, rankings, scouting reports, draft evaluations.
Baseball AmericaTool grades, risk, role context and historical rankings when available.
FanGraphsFV, tool grades, analytical prospect work and historical valuation.
Baseball-ReferenceMLB outcomes: WAR, playing time, peak production, awards and longevity.
The Baseball Cube + MiLBDraft, amateur, level progression and minor-league history.
Baseball Savant / StatcastQuality-of-contact and MLB tracking data where available.
Download the evidence
50-player backtest workbook
The exact calibration workbook behind the first S80 DNA v2 changes is included with this release.
Next research milestone
50 → 200 → out-of-sample.
The next step is deeper multi-year prospect snapshots and a larger mature cohort. Only after that do we turn model signals into public star/regular/bust probabilities.