The ATH Score: Athleticism, Measured Properly
Data Dictionary · combine testing without the highlight-reel bias
ATH is the athleticism piece of DMX. It's the combine and pro-day testing — 40, vertical, broad, 3-cone, shuttle, bench, height-adjusted weight — each z-scored within the prospect's position and rolled into four buckets (Speed, Agility, Power, Strength) and one composite. It treats athletic tools the way the data says you should: as one input of three, not the answer.
The uncomfortable truth about combine numbers
On their own, individual combine drills are weak predictors of fantasy careers. My correlation analysis (over on the Analytics page) says it plainly: the best single combine test anywhere is the TE 40, at a correlation of just −0.252 with 5-year VBD. QB broad jump comes in around +0.202. And the most-hyped number in the building — the 40 time for WRs and RBs — is basically zero on its own.
That doesn't mean athleticism is useless. It means you can't read the drills raw. Speed matters depending on a guy's size, his position, and whether the tape shows that speed actually showing up in football. A composite catches that. A stopwatch doesn't.
How ATH is built
- Position-normed z-scores. A 4.55 forty is ordinary for a WR and exceptional for a TE. Every metric is scored within the position cohort, so “fast” always means fast for the job.
- Height-adjusted weight. Mass matters relative to frame; the size input is adjusted rather than raw.
- Four sub-components. Speed (ATH-SPD), Agility (ATH-AGIL), Lower-body Power (ATH-LP), and Strength (ATH-STR) — visible on player detail panels, so you can see which kind of athlete a prospect is, not just how much.
- One composite. The sub-components blend into a single ATH z-score that enters the DMX composite alongside DPOS and AWP.
The workout-warrior trap
The way DMX is built is the guardrail against the classic combine mistake. An elite ATH score sitting next to weak age-adjusted production and modest draft capital is one loud vote against two quiet ones, and the model treats it that way. The guys who hit are almost always the ones where the independent signals agree. A pure testing spike with no production or capital behind it has always been the lowest-yield profile in the dataset.
The opposite mistake matters just as much: a slow 40 on a guy with elite production and real draft capital is mostly noise. The data's telling you the league already watched the tape.
Coverage and noise
ATH uses combine and pro-day testing together, so a guy who skips a drill at the combine is usually still covered. Here's the honest caveat, and it's in the known limits: small-school prospects who barely tested carry noisier ATH scores, so do a little extra homework before you lean on the number.
Frequently asked questions
Which combine drill predicts fantasy success best?
None of them, really, on their own. The best single test in my analysis is the TE 40 at a correlation of just -0.252 with 5-year value; the WR and RB 40 times are near zero by themselves. The real signal is in the composites of position-normed scores.
Does a fast 40 time matter for a WR?
By itself, barely — the correlation for WR 40 times with career value is near zero. Speed matters in context: relative to position and size, and backed up by production. That's what the ATH composite is built to capture.
What are the four ATH sub-components?
Speed (ATH-SPD), Agility (ATH-AGIL), Power (ATH-LP), and Strength (ATH-STR) — each built from position-normed z-scores of the underlying drills, and you can see them on a player's detail panel.
What if a prospect didn't test at the combine?
Pro-day results are in there too, so most guys are covered. A player who barely tested — common for small-school prospects — gets a noisier ATH score, which is a limitation I document and you should account for.