Draft Strategy Lab

What actually wins fantasy drafts — measured, not vibes. 1.04M simulated drafts over 15 real seasons, published with the failures included.
1.04M
simulated drafts
15
seasons of real ADP + results (2010–2024)
17
construction levers tested
3
formats (PPR · Standard · 2QB)
What controls your fantasy season
8% Repeatable strategy edge
23% Strategy value that reshuffles every season
46% Draft & injury luck
23% Other (schedule, season effects)

Only ~8% of the spread between fantasy teams is strategy you can bank on in advance. Three times that much is strategy value that reshuffles from season to season — and nearly half is luck no draft plan touches. Anyone selling you a season-winning “build” is selling you the amber and red parts of this bar.

The levers, gradedeffects in all-play win percentage points vs. a 12-team field
LeverVerdict
Take market value — never reach

The one edge that survived everything: rooms full of strategic drafters, standard scoring (+7.7), and title simulation. Its size scales with how sloppy your room is.

EDGE
Early-round WR lean (Zero-RB direction)

Real in PPR — but it vanished entirely in standard scoring. Zero-RB is a product of the reception point, not of RB fragility. Know your format before you punt RB.

DEPENDS
Chasing ceiling / upside picks

Best-ball lineups already harvest spike weeks; paying draft capital for variance double-counts it. A mild floor lean actually graded best.

AVOID
Handcuffing your RB1

The insurance costs more than it pays: a late pick spent on a backup who's only valuable if your best pick busts.

AVOID
Your draft slot

Across a million drafts, slot explained essentially nothing. Stop sweating the draft-order lottery.

MYTH
Avoiding injury-prone players

Actively fading durable players is disastrous, but chasing durability adds nothing — the market already prices it.

NO EFFECT
QB/TE timing (1-QB leagues)

In 1-QB formats, when you take your QB or TE barely matters. In 2QB/superflex it flips hard: waiting on QB costs −4.3 — and a third QB is still dead weight.

NO EFFECT
Stacking QB + receiver

In 1-QB season-long formats the correlation bonus is noise. (Tournament best-ball with weekly advancement is a different game.)

NO EFFECT
Rookie / veteran tilts

Neither chasing rookies nor avoiding them moved outcomes. A mild youth lean was the only age signal, and it's small.

NO EFFECT
The part nobody publishes

We held out 2023 and 2024, predicted them from the earlier seasons, and tested ourselves once. Every prediction hit in one season and missed in the other — exactly what the variance split above says must happen. The clearest example: in 2024, disciplined drafting meant taking the consensus #1 pick in every draft. That pick was Christian McCaffrey. He played four games. Discipline is an edge in expectation, not a guarantee — any fixed plan concentrates you into a specific handful of players, and single seasons are decided by how that handful turns out. That is the honest limit of draft strategy, and we'd rather publish it than pretend otherwise.

Method

Design-of-experiments over 17 roster-construction levers: simulated 12-team snake drafts against a field calibrated to real cross-draft ADP variance, graded best-ball on the actual weekly results of 2010–2024. Strategies could only use draft-day-knowable information. Screening designs, response surfaces, and championship simulation via schedule resampling; pre-registered predictions; 2023–24 held out and touched exactly once.

Use it

The practical playbook: know your room's prices, take value, never reach. Our Draft Board shows the market's price next to our projection for exactly this reason — use the gap to spot value, not to fight the room. Model receipts live on the Accuracy page, scored season by season.