2017–2024 backtest · graded against real NFL careers · the receipts →

Next draftApril 22-24, 2027 · Green Bay (TBC)

The best draft board public data can build.

Every prospect, graded on the only question that matters: how good an NFL career do they actually go on to have? Backtested out of sample across 2017–2024, Draftanomics's board predicts real career value at 0.55 — ahead of the analyst consensus (0.45) and a step shy of NFL war rooms (0.58), who grade with the medicals, interviews, and private workouts no public site ever sees. No public model gets closer to the league.

That's the rank correlation between each board's order and players' eventual career value (weighted Approximate Value) — 0 is a coin flip, 1 is perfect foresight — scored only on classes the model never trained on.

Enter the draft room →2027 big board How the model works →

The lead · August 3

Camp Opens, the Board Shakes Out, and the Model Digs In on Adam Trick

A six-and-a-half-day window produced the board's ugliest slide of the summer and its widest model-versus-room gap yet.

Read the piece →

Track record

The receipts

Full backtest →

Rank correlation between each board's pre-draft order and players' real NFL career value, pooled 2017–2024 — every class scored by a model that never saw it. We publish the number even though NFL war rooms still lead.

Full backtest: 11 draft classes, 2015–2025 — every year graded by a model trained without it.

The editorial

What we're writing

All writing →

The class at a glance

Model run just now

2027 draft class

The model's elite and bustcalls are its reliable ones — the middle tiers are closer to coin flips, so they're shown muted.

Ranked prospects

on the consensus board

Elite bets

Better-than-even shot at a star career

Likely starters+

147

At least a 60% chance he becomes an NFL starter or better

Real bust risk

Bust probability 30% or higher — top 100 on the board, same number the position pages quote

Scarcest premium

2 starter+ grades

Position-group tier mix · the model's top 200

elitestarterrolebust

21prospects

34prospects2 elite

11prospects

9prospects

6prospects

EDGE

14prospects1 elite

22prospects

24prospects

11prospects1 elite

9prospects

ATH

18prospects1 elite

The 2027 class is deepest at WR (30 starter-grade or better) and thinnest at G (2). Teams with a top-half pick and a G need should consider reaching or trading up.

Full board →

The next class

2027 NFL Draft

43 prospects with analyst coverage

Open 2027 board →

Open the board and build your own read on the class — tag your sleepers, rank your guys, and see exactly where you break from the model. When you disagree, put two prospects side by side or stack your board against Draftanomics's on your big board.

Highest elite odds right now

Jeremiah Smithblend#1$14M

WR · 52% elite odds

Leonard Mooreblend#2$13M

CB · 50% elite odds

Trevor Goosbyblend#3$12M

OT · 46% elite odds

Watchouts — high ceiling, highest bust risk in the class

Major Everhart

RB39% elite53% bust· rough estimate

Will Dixon

TE22% elite41% bust· rough estimate

Jaylen Westerblend#131$1M

LB38% elite41% bust

Da'Shawn Davis

DL27% elite41% bust

What we ship that no other draft site does

Three honest moves the consensus draft sites don't make.

Where we're wrong

Named failure modes

Every model has blind spots. Ours can barely tell which interior linemen will bust — those calls are no better than a coin flip. We publish the failures — the alternative is selling overconfidence.

Read the failure modes →

Risk × reward

The boom-bust quadrant

Every prospect plotted by bust probability against elite ceiling. Sleepers in the top-left, boom-busts top-right, fades bottom-right. The picture the consensus board hides.

Open the quadrant →

The room

Scout-synthesis prose

Claude Sonnet reads every public scouting report + analyst board on a prospect and writes Draftanomics's synthesized take. 320 prospects covered, citing 25-28 sources each.

See an example dossier →

The Tuesday tab

Get the model dissents in your inbox.

One email per week during the cycle (combine through May). Where the model breaks from consensus + the receipt to back it up. Monthly off-cycle. Free, no spam, easy unsubscribe.

Premium · founders' beta

Free where it counts. Paid where it matters.

The full 2027 board, mock-draft simulator with film, comp-anchored AAVs, and the consensus tracker — for less than half the price of PFF.

Reserve your spot →

Email-only signup right now — no card needed until Premium opens at the Combine.

How the model has done

These years are the backtest — each was graded by a model that never saw it, so the numbers show how the model would have actually performed if it had been used on draft night. Click a year to see the picks the model nailed and the ones it whiffed on.

11 years backtested