Brock Bowers finished last summer with an average draft position of 23. If ADP worked the way most draft prep treats it, he would be gone by the end of the second round in essentially every league. He wasn't. In 1 draft out of 10, he was still on the board six picks later. Josh Allen, priced at 21, lasted six picks past his ADP in 17% of drafts. Lamar Jackson did it in 18%.
We pulled every completed 12-team snake draft from our Sleeper crawl for the 2025 season and measured where players were actually taken: 21,893 drafts, about 3.9 million picks after removing keepers, and 204 players with at least 500 samples each. The average told us what ADP already tells everyone. The spread around it turned out to matter more.
A pick is a distribution, not a date
Every player's draft slot has a bell around his price, and the bell widens the deeper into the draft you go. A player priced at pick 6 lands within about 3 picks of it. A player priced at 70 scatters across 10. By pick 120, the same "price" covers two full rounds in either direction.
Fitting all 204 players gives a clean rule of thumb: the spread of a player's landing spot is about 2.25 + 0.11 × his pick number. Hover the chart above and you can read it directly. A player priced at 30 still has a 24% chance of being on the board six picks later. Nobody priced at 6 survives to 20.
Some of that spread is the season itself. ADP moved between May and September, and these drafts span the whole summer. That looseness is honest: your draft happens somewhere in that window too, and the room you sit in has read different news than the one next door.
Same price, different tail
Two players priced at the same pick do not fall the same way. The difference is position.
Running backs and receivers priced in the first two rounds barely slide. Of early RBs, 5.1% lasted six picks past their price. Receivers: 4.8%. Nico Collins, priced at 14, made it six picks further in 4.1% of drafts. When the market prices an early back or receiver, the market means it.
Tight ends and quarterbacks are a different market. The width of their early-round tail measures 6 to 7 picks at the base, against about 2.5 for backs and receivers. That is Bowers lasting to the third round in a tenth of drafts, and Allen and Lamar outlasting their price in nearly a fifth.
The curves above answer a question we kept getting about our draft plan simulator: you hold the 1.02, and Bowers is somehow still there at your second pick. Should you have planned for that? Yes. A tight end priced at 18 reaches pick 23 roughly twice as often as a running back priced at 18. If a sliding TE or QB tempts you, that temptation is a predictable event, and you can decide in advance what you'll do about it.
The whole league, one dot per player
Two things stand out when you plot all 204 players. The spread grows almost linearly with depth, and the purple and blue dots (quarterbacks and tight ends) sit above the trend line at every depth. The market prices them, then drafts them late anyway.
What this changes about how you draft
Treat ADP as a deadline for backs and receivers, and a soft opening bid for everyone else. If you want a specific early RB or WR, his price is roughly the last moment you can act. If you want a tight end or quarterback, the data says you can often wait a few picks past where the board tells you to panic.
Plan around the tail, not the average. The useful question before a draft is what you'll do when a faller reaches you, because somebody always does. Each individual slide is rare, but with a dozen players priced ahead of any given pick, one of them sliding to you is close to routine, and whether you can afford him depends on the rest of your plan, not on the size of his discount.
Later rounds are wider than your queue thinks. At pick 120 the spread is ±16. Queueing three names for a late pick means all three could be gone, or all three could still be there two rounds later. Queue by tier, not by name.
Where these numbers now live in STACKED
The fitted curves replaced the assumptions in our tools this week:
- The draft war room availability percentages now use the measured spread for each player's position and depth, so a tight end's "still available at your next pick" number reflects his real tail instead of a fixed window.
- The mock draft bots in the demo draft room now scatter their picks the way real rooms do, including letting quarterbacks and tight ends slide.
- The draft plan simulator prices every plan, deadline, and "can I afford the faller" verdict off these distributions.
Method notes. 12-team snake drafts only, 2025 season, primary completed drafts, keeper picks excluded, picks capped at 180. Spreads are quantile-based, (p90 − p10) / 2.563, because raw standard deviations get inflated by a handful of unusual leagues. The distributions are also skewed toward the back of the draft: players get reached for less far than they fall, so we fit the two half-widths separately, (p50 − p10) / 1.282 for the reach side and (p90 − p50) / 1.282 for the fall side, and our tools model each player as a two-piece curve around his median. Bowers again: his 2025 quantiles ran 16–23–30, a tight front and a long back. If you want to check our work, the fit tables and refit queries are documented in the repo behind the site.