Sports Analytics -- Draft Strategy

ADP rankings aren't wrong.
They're just not using the right variables.

Julian Batto-Hokson ~4,000 player-season observations, 2021 to 2025 Python, Statsmodels, FantasyPros March 2026

Executive Summary

Consensus ADP rankings are built from aggregate draft behavior -- which means they encode the biases of the average drafter, not the optimal draft strategy. This analysis identifies where those biases create exploitable inefficiencies across five seasons of NFL data.

Three patterns are consistent and predictable. Running backs in rounds 1-2 underperform their ADP by 15 to 20 points on average. Wide receivers in rounds 3-5 outperform ADP by 10 to 20 points. Targets are the single strongest predictor of fantasy production across all positions (correlation 0.72 to 0.81) -- yet ADP rankings weight prior-season points more heavily than target share.

A draft strategy that exploits these three inefficiencies is estimated to add 25 to 40 fantasy points over a full season compared to consensus ADP drafting. In a 12-team league, that margin is the difference between making the playoffs and missing them.

-15 to -20
Avg ADP overperformance gap for RBs drafted in rounds 1-2
+10 to +20
Avg ADP outperformance for WRs drafted in rounds 3-5
0.72-0.81
Target share correlation with fantasy production -- strongest predictor
R2 = 0.71
QB predictability -- highest of any position
R2 = 0.58
TE predictability -- lowest, most volatile year-to-year
25-40 pts
Estimated season-long gain from data-driven draft vs. consensus ADP

Five patterns in the data -- and what each one suggests about consensus ADP

01Early-round RBs underperform -- injury rates and committee backfields are the mechanism

Running backs drafted in rounds 1-2 underperform their ADP expectations by 15 to 20 fantasy points on average across the 2021-2025 period. The drivers are structural: injury rates of 15-20% per season, usage volatility from game-script dependencies, and NFL teams increasingly deploying committee backfields that reduce any single back's target share.

Draft strategy implication

The only exception to avoiding early-round RBs is a true workhorse back with demonstrated pass-catching volume -- one who maintains relevance even if the team falls behind. For most RBs in rounds 1-2, the ADP reflects upside scenarios rather than median production. Waiting until round 4 or later for RBs gives equivalent production at lower opportunity cost.

02Mid-round WRs outperform -- stable systems and target volume are the explanation

Wide receivers in rounds 3-5 consistently beat ADP expectations by 10 to 20 points. The position shows lower injury volatility than RB, more stable offensive systems, and -- critically -- target share is more predictable year-over-year than RB usage. The R2 for WR production is 0.68, second only to QBs.

Draft strategy implication

Concentrating 3 to 4 WR picks in rounds 3-5 extracts more value per pick than any other strategy. The WR position in this range has the best combination of predictability, depth, and ADP discount. Drafters who fill RB slots in rounds 1-2 and pivot to WR heavy in the middle rounds leave value on the table.

03Targets predict production better than any other variable -- and ADP underweights them

Target share correlates 0.72 to 0.81 with fantasy production across RB, WR, and TE positions. Prior-season fantasy points, which ADP rankings weight heavily, explain only 45-54% of next-season variance. The gap between what predicts production and what ADP uses to set prices is where draft value is extracted.

Pre-draft research implication

The correct pre-draft ranking process: identify target share projections by position, flag players whose ADP ranks them below their projected target share rank, and prioritize those players in the relevant draft windows. Players whose ADP was depressed by a down prior season but whose target share held stable are the clearest value targets -- the market is pricing recent points, not the underlying opportunity.

04QB streaming from round 10 or later matches early-round QB production

Quarterbacks drafted after round 10 match first-tier QB production on a point-per-game basis when matchup-selected weekly. QB scoring shows the highest year-to-year stability (R2 = 0.71) -- which means the drop-off from QB1 to QB12 is smaller than ADP pricing implies.

Draft strategy implication

Spending a pick on a QB before round 10 in most formats costs a WR or RB selection in the high-value windows. The exception is a format with two starting QB slots or a superflex league -- in those formats, QB scarcity is real and early investment is warranted. In standard formats, streaming saves 6 to 8 picks of draft capital for positions where the value gap between early and late selections is larger.

05Only 2 TEs are worth early investment -- streaming is correct for every other TE slot

Tight end production has the lowest predictability of any position (R2 = 0.58) and the steepest drop-off after the elite tier. After the top 2 TEs by target share, streaming produces equivalent results to drafting depth early.

Draft strategy implication

If an elite TE (top-2 by projected target share) is available in rounds 4 to 6, take them. Otherwise, skip the position entirely until round 8 or later. Spending round 4 capital on the third-best TE is a common ADP-driven mistake -- the positional scarcity is real only at the very top of the position, not throughout it.

Recommended round-by-round approach

Rounds 1-2

Elite WR or pass-catching RB

Avg: 149-175 pts. Avoid standard RBs unless top-3 workhorse.

Rounds 3-5

WR focus -- 3 picks here

Sweet spot: +10 to +20 pts vs. ADP. Best value window in the draft.

Rounds 6-7

RBs and flex WR

RBs now offer better risk-adjusted value than rounds 1-2.

Round 8

Elite TE only -- or skip

Only invest if top-2 TE by target share is available. Otherwise defer.

Rounds 9-12

QB, backup RB/WR, Defense

Stream QB by matchup. Defense in round 12-13.

Late rounds

High-upside targets and handcuffs

Target players with stable target share whose ADP reflects down prior season.

Recommended draft adjustments and estimated impact

Highest Impact

Avoid RB in rounds 1-2 unless workhorse confirmed

Early-round RBs underperform ADP by 15-20 pts on average. Redirecting those picks to elite WRs preserves the same ceiling with lower injury and usage risk.

Estimated gain: 15-20 pts recovered vs. consensus RB-heavy drafting in rounds 1-2

Highest Value Window

Stack 3 WRs in rounds 3-5

Mid-round WRs outperform ADP by 10-20 pts. This is the single most consistent exploitable inefficiency across all five seasons of data.

Estimated gain: 10-20 pts per WR pick vs. ADP expectation in this range

Draft Capital Savings

Stream QB from round 10 or later

QB predictability (R2 = 0.71) makes the drop-off from QB1 to QB12 smaller than ADP implies. Streaming frees 6-8 picks for WR and RB windows where the value gap is larger.

Estimated gain: 6-8 picks of draft capital redirected to higher-value positions

Target Share Mispricing

Draft players with stable target share whose ADP reflects a down prior season

Prior-season points explain only 45-54% of next-season variance. Target share (0.72-0.81 correlation) is the stronger signal. ADP underweights it.

Mechanism: identify players where target share rank and ADP rank diverge by 2+ rounds

Overall estimated season gain

Implementing these four adjustments together is estimated to add 25 to 40 fantasy points over a full season compared to consensus ADP drafting. Derivation: 15-20 pts from RB avoidance in rounds 1-2, plus 10-20 pts from WR value extraction in rounds 3-5, minus overlap from players already at the top of both lists. In a 12-team league where playoff qualification typically requires finishing in the top 4, a 25-40 point margin over a 13-14 week regular season represents roughly 2-3 points per week -- a meaningful edge in close matchups.

Model results and data scope

PositionR2PredictabilityRMSE
QB0.71High -- stable passing volume42
WR0.68High when team situation stable38
RB0.64Moderate -- usage volatility45
TE0.58Low -- highest year-to-year variance52

Data Note

Analysis period is 2021-2025 (5 seasons, approximately 4,000 player-season observations, Full PPR scoring). Sources: FantasyPros for scoring datasets, Fantasy Football Calculator for ADP rankings. Tom Brady is excluded from the QB production analysis for seasons after his retirement following the 2022 season. Any player listed in top performers charts reflects their performance during the seasons they were active in the dataset. Tom Brady is excluded from the QB analysis for seasons after his 2022 retirement. The R2 values represent cross-season holdout predictability, not in-sample fit.