Methodology

How DraftBoard builds your personal board and the community consensus.

Roadmap

Pick Grader — unlocks after Round 1. Once the picks are in, compare every selection head-to-head and earn an A+ to F− grade for each pick. Talent matters, but so does value — a late-round steal can outgrade a first-round overdraft.

Mock drafts, coming soon. The mock draft runner is being rebuilt to use live board ratings alongside team needs and positional value. Focus on stacking your personal board — that data feeds the mock engine when it ships.

Historical redraft, coming soon. Pre-draft vs actual comparisons and biggest-steals views are on deck. You can already pit current prospects against past picks in Historical mode in the arena.

Pairwise comparison engine

Every prospect starts at the same baseline rating (1500) with a relatively large uncertainty (sigma of 350). When you pick one player over another, both ratings move using an Elo-style update. The K factor decays as players get more votes, so early choices move the needle more than late ones.

Sigma decays slowly toward a floor over time. Combined with the number of comparisons you've made for a player, it drives the confidence label you see on each board row: volatile, forming, medium, high, or locked in.

Matchup selection

The selector prioritizes prospects with the highest remaining uncertainty, then pairs them with a player ranked close to them so every vote is informative. It avoids pairs that showed up in your last 25 matchups and occasionally revisits older matchups to refine rankings.

The arena progresses through four stages automatically: top players at each position, full same-position board, cross-position comparisons, and finally current prospects vs actual picks from 2018–2025. You can also lock to a specific mode at any time.

Pick Grader

After Round 1, every pick in the 2026 class becomes comparable — not just as players, but as selections. The Pick Grader runs a separate Elo system on drafted players: you compare two picks and decide which gave the team better value.

Grades (A+ to F−) are assigned by percentile rank within the class, not by raw Elo. This means roughly the top 8% earn A or better, and only the worst 4% fall to F territory. A third-round steal can absolutely outgrade a first-round bust.

Community consensus

Community ratings are a parallel Elo run where every non-skipped vote from any user updates both players. Ranks within a class are recomputed after every batch of votes so the live board stays consistent. We use the same scale as personal boards so you can compare ratings directly.

Data sources

Historical picks and combine measurables come from nflverse via nfl_data_py — all 257+ picks per class from 2018–2025. Headshots use ESPN's CDN (NFL bucket for drafted players, college-football bucket for prospects). The 2026 declared prospect list is sourced from public declarations as of April 2026.

All player data is for research and illustrative purposes. DraftBoard is an independent project and is not affiliated with the NFL, ESPN, or any team.