Slick Labs LLC

Everyone can show you the best player left. That’s the easy part.

The hard part is knowing whether he’ll still be there at your next pick, and what it’ll cost you if you’re wrong.

What it is

A draft-day assistant for people who take their league seriously and would rather not spend August reading rankings.

It watches your draft

You give it your league’s rules once — teams, roster slots, scoring. From then on it follows along pick by pick.

It thinks a round ahead

Before recommending anyone it plays out what the other managers are likely to do between now and your next turn.

It explains itself

Why this player, what the alternative was, and how close the call was — in plain English, every time.

Who it is for. People in a home league who want to draft well without turning it into a second job. It is not built for professional daily-fantasy players, and it does not try to be an encyclopedia. It is built for the ninety minutes of a draft, and for the hour before it.

How it is different

The big platforms are excellent at running a league. Draft advice is not what they are for.

What it does	The big platforms	SlickDrafts
Simulates what the other managers will do	No	Before every pick
Tells you the cost of the alternative	No	Yes
Publishes how accurate its projections are	No	Yes — just below

The projections are our own. They are built from public NFL history rather than bought in, and we publish how they score against a public baseline — including the seasons where they did worse. Almost nobody in this industry publishes a backtest.

Does it actually work?

The projections are measured against seven seasons the model never saw, and against the baseline that quietly beats most published projections: assume every player repeats last year.

Predictor	QB	RB	WR	TE
"Just use last season"	0.493	0.577	0.609	0.509
SlickDrafts	0.699	0.697	0.689	0.616

Rank correlation between projected and actual finish, higher is better. Zero would mean we knew nothing; one would mean we ordered every player perfectly.

We publish the misses too. Our accuracy report breaks the result out by season and by position, including the seasons where the naive baseline beat us, and it ships with the product rather than a single averaged headline. A mean with the bad years averaged out is an advertisement, not a measurement. Tight end is the weakest of the four and the margin there is not yet large enough to be sure it is real — improving it is current work.

Nothing here was scored on data the model had seen. Constants are fitted on the earliest seasons, and everything reported is measured on later ones that were held out.

What’s next

A draft room worth showing to a real league — then ten people from one, drafting with it live and telling us whether the advice is any good.