How it works

Built to beat the market, not to sound smart.

A proprietary model simulates every player’s season thousands of times. We re-test it against every season on record, using only what was knowable before Week 1. Nothing reaches the board because it sounds right. It reaches the board because it beat Market ADP.

Historical results are published from our own boards after each season ends.

What you get

Questions

How do you actually test this?

We rebuild an old board using only what existed before Week 1 that year, then grade it against Market ADP and the real finishes. Season after season, walking forward. A change ships only if it helped after the market was accounted for.

Where do the projections and ranges come from?

Every player’s season is simulated thousands of times, drawing on how comparable players in comparable situations actually finished. So we publish a range, not one number. A floor, a ceiling, and the odds of each.

Is the algorithm proprietary?

Yes. We publish the record, how it’s scored, and the reasoning behind every rank. You see why we disagree with the market before the season settles it.

Does it get better over time?

Every completed season is another year of evidence to test against. Signals that stop working get pulled. That is the process, not a flaw in it.

What happens to ideas that don’t work?

Most die, and we keep the receipts. A stat can be fascinating and still worthless for beating a draft price. See what didn’t survive →

Can I connect my actual league?

Sleeper connects with just a username — no password. Manual import works for any league. Connections are read-only: we can’t change your lineup, make moves, or post in your league.