Monday, August 17, 2026
Why Nine Games Should Not Weigh Like Seventeen
Regression separates durable 2025 evidence from small-sample scoring that deserves more skepticism in 2026 drafts.
Bobby's AI Desk·written from our own draft-market and 2025 production data·how this works
Takeaways
- → Check games played before trusting any per-game average.
- → Prefer sturdier full-season evidence when similarly priced ranges overlap.
- → Treat rookie ADP as projection, not NFL production evidence.
In this piece
Last season matters, but every scoring average should not carry equal weight. The useful approach takes a player's 2025 per-game scoring, pulls it toward his positional average based on how many games we actually saw, then adds a range around that adjusted evidence.
This is not a 2026 forecast. It is an honest accounting of what 2025 showed, with less trust placed in smaller samples.
Small samples lose authority
Brock Purdy averaged 19.71 PPR points per game across 9 games last season. That is strong on its face, but the limited sample gets pulled down hard by the weighting process and produces a wide 9.84-25.78 range.
Compare that with Matthew Stafford, who averaged 20.61 across 17 games. His weighted 2025 evidence lands in a 13.19-25.20 range. The averages were similar, but Stafford's full season creates a narrower range and a higher low end. That does not make Stafford a safer 2026 pick by itself. It means we have more 2025 evidence behind his scoring rate.
Sample size is easy to miss when scanning the Draft Kit. Put the games played beside the average before treating two players as statistical equals.
Full seasons hold their shape better
Drake Maye supplied 20.70 points per game over 17 games, leaving a weighted range of 14.63-23.90. Josh Allen posted 22.79 over 16 games, with an 11.91-29.66 range.
The ranges still acknowledge substantial weekly uncertainty, but neither average rests on a brief run. The same principle applies outside quarterback. Jaxon Smith-Njigba averaged 21.17 in 17 games, producing an 11.36-24.23 range. Christian McCaffrey's 24.51 across 17 games becomes 13.53-27.68 after regression and uncertainty are applied.
Even complete seasons get pulled toward normal positional scoring. Regression is a guardrail against assuming an exceptional average automatically repeats.
Draft price can disagree with the evidence
ADP, or average draft position, is the typical spot where a player is selected. Value means comparing that 2026 price directly with what the player's 2025 production supports, not claiming he could have been drafted later last year.
Stafford's 7.09 ADP comes after Maye's 5.02, even though their weighted 2025 ranges overlap: 13.19-25.20 for Stafford and 14.63-23.90 for Maye. Trevor Lawrence costs 8.08 after averaging 19.89 over 17 games, with an 11.09-26.20 range. Patrick Mahomes goes at 9.03 after 20.41 across 14 games, with a 12.13-25.51 range.
Those are not automatic bargains. They are reasons to investigate the scoring inputs in the Usage Report before paying the market price.
Know where the method stops
Rookies have no NFL sample, so their prices are projections rather than regressed evidence. Jeremiyah Love at 3.05 and Jordyn Tyson at 8.10 must be evaluated differently from veterans with 2025 games.
Weekly volatility also needs separate treatment. Boom/bust describes how often a player reached a high score or fell below a low one. Joe Burrow scored 20-plus in 50% of his 2025 games but finished under 8 in 25%. That distribution adds context beyond his 16.8 per-game average.
How to use this: Favor full-season evidence when similarly priced players have overlapping ranges, and demand a discount for small samples. Use the Anytime TD Board for weekly touchdown pricing, not as a substitute for season-long evidence.
Get the ADP movers by email
One email a week through draft season: who's climbing, who's falling, and which players the market is pricing furthest from what they actually produced last year. Free — no account needed.
Weekly through Week 1. No spam, unsubscribe any time.