Draft Decision Grade — Methodology

What we publish

Each pick receives two qualitative labels and no letter grade. The labels are:

• Board Value band — one of Steal, Good Value, Slight Value, Fair Value, Slight Reach, Reach, Significant Reach, or Off Consensus Board. This compares the pick's actual slot to where the consensus analyst community had the prospect ranked pre-draft.
• Position Premium tier — one of Elite Premium (tier 5), Premium (tier 4), Standard (tier 3), Value (tier 2), or Discount (tier 1). This reflects how the NFL cap market prices the position, derived empirically from top-of-market AAV at each position.

Letter grades (A+ through F) are withheld for the 2026 draft. The research panel signed off on publishing the component signals only, pending a v2.1 calibration study that requires historical consensus-rank data for 2010–2020 draft classes. Letter grades will publish after that calibration lands and the panel formally approves them.

Board Value — what it measures

Board Value is the difference between the consensus pre-draft rank and the actual draft slot. Positive slide (consensus rank higher than actual slot) means the player fell past where the market valued him and the drafting team captured surplus. Negative slide means the drafting team reached past the consensus.

The consensus rank is built by aggregating seven-plus public analyst boards into a single 257-prospect ranking. Sources: NFL Draft Scout, Drafttek, Walter Football, NFL Mock Draft Database, plus major-outlet consensus boards.

Bands are set in uniform slide buckets: Steal (≥20), Good Value (≥10), Slight Value (≥4), Fair Value (≥–3), Slight Reach (≥–9), Reach (≥–19), Significant Reach (<–19). Off Consensus Board fires when the player was not in the 257-prospect ranking at all.

Position Premium — what it measures

Position Premium tiers reflect the relative cap-market value of each position, based on top-of-market AAV across the most recent five completed league years. The intuition is that a hit at a Premium position captures more surplus value on a rookie contract than a hit at a Discount position, because Premium-position free-agent salaries are larger.

Tier assignment is empirical — we fit it to observed top-of-market AAV data, not to subjective “skill position” folklore. Positions move between tiers as the market moves.

What the formula predicts

The underlying ExpectedPS(slot) function was fit on 1,000 NFL draft picks (2005–2025 classes) and validated non-circularly against career award tier (MVP/OPOY/DPOY, All-Pro, Pro Bowl, Rookie of the Year). Correlation: r = 0.273.

That r value sits at the upper bound of what published pre-draft forecasting literature achieves (Massey & Thaler 2013, Berri & Simmons 2011, Hendricks/DeBrock/Koenker 2003, Mulholland & Jensen 2014, Wolfson/Addona/Schmicker 2011) — all of which report max in-sample R² of 0.20–0.35 for pre-career career-production prediction regardless of methodology.

Translated plainly: a pick flagged as Significant Reach by the model is more likely than a Good Value pick to underperform its slot, but individual outcomes vary widely. The formula is a directional market-deviation signal, not a career-outcome predictor. Anyone who claims otherwise is overselling.

What the formula does NOT do

• It does not predict individual NFL careers. r=0.273 means roughly 7.5% of career-outcome variance explained.
• It does not account for team context (scheme fit, coaching, roster around the player).
• It does not account for positional roster need. A reach for a position the team needs is still a reach by Board Value, even if it's defensible on fit grounds.
• It does not account for rookie contract structure, 5th-year options, or compensatory-pick implications. Those are v2.1 work.
• It does not replace scouting. It surfaces where a pick deviates from the public market consensus, which is a different question than whether that deviation is correct.

Panel & versioning

The formula was reviewed by FanVerdicts' internal research panel (econometrics, labor economics, machine learning, NFL data) and signed off as CONDITIONAL — components only, letter grades withheld until v2.1 calibration completes. The panel signoff document is stored in the project's research archive.

Scheduled v2.1 work: position-stratified ExpectedPS fits, 2010–2020 historical consensus-rank backfill, rookie wage scale integration, 5th-year-option calibration, and out-of-sample holdout validation on 2021–2025 classes.