Model accuracy & track record

Every number here is a Bet The Line model checked against real results. Graded, not claimed. The market is the accuracy ceiling; where we quote it, we say so.

Weekly projections

2025 season, every player

Every projection re-run and checked against the real box score. The players this model ranked highest averaged 20.8 points a week vs 6.6 for the field, and it beats a recent-form baseline at every skill position:

QB
0.55
form 0.47
RB
0.70
form 0.68
WR
0.64
form 0.63
TE
0.64
form 0.62
DST
0.30
form 0.07
K
0.17
form 0.11

r = correlation to actual · higher beats the recency-form baseline · graded on DraftKings scoring

A correlation says the order is broadly right. It does not say how wrong one row usually is, which is what you are actually risking when you start somebody. So here is the spread, measured:

// What a RB rank here has actually been worth

Where players we ranked in each band really finished, across the 2025 season. The bar is the middle half of outcomes, the notch is the median. Wide bars are the honest part: this is a weekly game and the order is a lean, not a promise.

RB1-RB3
RB2.25 to RB11
RB4-RB6
RB8 to RB18.5
RB7-RB12
RB7 to RB22.25
RB13-RB24
RB11 to RB30
RB25-RB48
RB25 to RB48

Measured on the 2025 holdout, the season nothing was fitted on. Sample sizes run 54 to 432 player weeks per band.

// What a WR rank here has actually been worth

Where players we ranked in each band really finished, across the 2025 season. The bar is the middle half of outcomes, the notch is the median. Wide bars are the honest part: this is a weekly game and the order is a lean, not a promise.

WR1-WR3
WR3 to WR18
WR4-WR6
WR8.5 to WR29
WR7-WR12
WR9 to WR34
WR13-WR24
WR12 to WR42.25
WR25-WR48
WR26 to WR59

Measured on the 2025 holdout, the season nothing was fitted on. Sample sizes run 54 to 432 player weeks per band.

// What a QB rank here has actually been worth

Where players we ranked in each band really finished, across the 2025 season. The bar is the middle half of outcomes, the notch is the median. Wide bars are the honest part: this is a weekly game and the order is a lean, not a promise.

QB1-QB3
QB4 to QB12.75
QB4-QB6
QB4 to QB14
QB7-QB12
QB5 to QB15.25
QB13-QB24
QB11 to QB22.25

Measured on the 2025 holdout, the season nothing was fitted on. Sample sizes run 54 to 192 player weeks per band.

// What a TE rank here has actually been worth

Where players we ranked in each band really finished, across the 2025 season. The bar is the middle half of outcomes, the notch is the median. Wide bars are the honest part: this is a weekly game and the order is a lean, not a promise.

TE1-TE3
TE2 to TE12.5
TE4-TE6
TE6.25 to TE22.75
TE7-TE12
TE8.75 to TE26
TE13-TE24
TE9.75 to TE33
TE25-TE48
TE25 to TE52

Measured on the 2025 holdout, the season nothing was fitted on. Sample sizes run 54 to 432 player weeks per band.

// What a DST rank here has actually been worth

Where players we ranked in each band really finished, across the 2025 season. The bar is the middle half of outcomes, the notch is the median. Wide bars are the honest part: this is a weekly game and the order is a lean, not a promise.

DST1-DST3
DST4 to DST10
DST4-DST6
DST4 to DST14
DST7-DST12
DST6 to DST16
DST13-DST24
DST10.75 to DST21.25

Measured on the 2025 holdout, the season nothing was fitted on. Sample sizes run 54 to 108 player weeks per band.

Draft tier track record

2019-2024

How the players in each draft tier have actually finished, graded leave-one-season-out in this scoring (a season’s tiers never see its own results). Three things the record shows:

  • · Elite running backs are the riskiest premium pick. The top RB tier finishes ~RB13 on median and busts 19% of the time, worse than tier 2.
  • · First-round receivers are the safest. The top WR tier finishes ~WR9 on median with a 45% league-winner rate and just 16% busts.
  • · The cliffs are real. League-winner odds fall from ~30-45% in tiers 1-2 to about 5% by tier 4; bust rate crosses 50% around tier 5.

Running back

TierDraft costMedian finishLeague-winnerBustSample
T1RB1-6RB1331%19%32
T2RB5-15RB1128%12%57
T3RB15-24RB2312%28%25
T4RB16-32RB265%35%40
T5RB19-41RB345%47%43
T6RB24-50RB432%58%57
T7RB33-62RB452%70%57
T8RB42-68RB770%85%59

Wide receiver

TierDraft costMedian finishLeague-winnerBustSample
T1WR1-7WR945%16%31
T2WR5-19WR1427%18%33
T3WR8-22WR209%18%55
T4WR10-38WR383%41%74
T5WR24-45WR376%38%52
T6WR39-57WR602%64%58
T7WR43-60WR523%64%39
T8WR55-65WR770%84%38

Quarterback

TierDraft costMedian finishLeague-winnerBustSample
T1QB1-3QB720%10%10
T2QB2-4QB443%0%7
T3QB3-8QB937%21%19
T4QB5-13QB1611%52%27
T5QB9-18QB146%39%31
T6QB14-24QB223%69%32
T7QB19-32QB260%71%31
T8QB25-34QB270%82%22

Tight end

TierDraft costMedian finishLeague-winnerBustSample
T1TE1one player deep, no tier record6
T2TE2-5TE636%36%14
T3TE4-8TE117%43%14
T4TE5-11TE1315%35%20
T5TE9-15TE135%45%20
T6TE12-26TE220%74%31
T7TE14-32TE235%67%39
T8TE19-34TE290%83%30

One tier here covers a single positional rank, so the same player held it most seasons and there is no tier-level record to report. Everything else on this page pools several players a season.

median finish = middle actual positional finish · league-winner = an elite year · bust = outside startable · sample = players across 2019-2024. Quarterback and tight end top tiers are small samples. Kicker and defense are streaming positions and are graded but not shown here.

Game & win model

2019-2025 holdout

The engine behind Winners, Survivor, and Pick 6. Graded straight-up against the closing market on 1,865 games.

Straight-up accuracy
66.5%
Model-only accuracy
64.5%
covers unpriced weeks
Brier vs market
0.210
market 0.210

lower Brier is better · the market moneyline is the ceiling · the blend matches it on priced weeks and covers the rest

Weekly projections: team environment x volume shares x efficiency x xTD allocation (DraftKings), graded every player-week against the box score. Draft tiers: frozen preseason consensus, clustered per position, refit leave-one-season-out and scored on the real season. Game model: a QB-value adjusted Elo graded against the de-vigged market. Numbers refresh as new results land.