NFL DFS Slate Board
Every player DraftKings is offering, with what he is likely to score, how often he reaches what the slate needs, and how much of the field will be on him. Proj is the median, Ceiling the 90th percentile and Super the one game in thirty that wins a tournament. Boom is how often he clears the score his position actually needs on this slate. Leverage is that boom rate minus his ownership, so positive means the field is under-rostering the outcome he genuinely produces. Value is his projection minus what his price normally returns, so zero is a player scoring like a typical starter at his salary.
Ownership here is a model, not a measurement. There is no NFL contest ownership data to train on yet, so these come from salary, value and ceiling with one hard constraint: a lineup has exactly one quarterback, so every quarterback's ownership must add to 100%, and every running back to about 230% once the flex is shared out. That arithmetic is enforced. The split between the three contest types is a documented assumption about how those fields behave, not three separately measured models. Every build is archived so that the moment real results exist, this becomes checkable.
How to read this board
Leverage is the number worth sorting on
A projection tells you what a player scores. It does not tell you whether rostering him is a good idea, because that depends on his salary and on how many other entries have the same player. Two lineups that both score well do not both win: the one nobody else had does.
So leverage compares how often a player reaches the score his position actually needs on this slate against how much of the field is on him. The threshold is taken from the slate rather than fixed, because the score that makes a quarterback worth playing on a twelve game slate is not the same as on a three game one.
Value is measured against the price, not divided by it
The usual value column is points per thousand dollars of salary, and it is mechanically misleading. Salary does not buy points at a constant rate, so dividing by it always flatters the cheapest players and the column ends up ranking mostly by price. Here a player is compared with what his own price typically returns, so zero means he scores like a typical starter at his salary and positive means he beats his price tag.
The curve that defines "typical at this price" is fitted across the slate, separately for each position, and only players with a real role count toward it. That exclusion is the part that matters. A backup quarterback priced at four thousand dollars who will not take a snap scores nothing, and letting those into the fit drags the cheap end of the curve toward zero, which then makes every cheap player who actually plays look like a miracle. Before that filter this page expected a four thousand dollar quarterback to score 1.6 points; with it, 14.3, which is what a cheap starter really does.
One honest difference from how others do this. The curve here is fitted on what this model projects for the current slate, which measures whether DraftKings has mispriced a player relative to the projections. Fitting it instead on what players at each price actually scored across past seasons would be the more robust version, and it is not possible yet because historical salaries were never archived. Every build now archives them, so that version becomes available in a few weeks.
Why ownership has to add up
Most ownership models are a list of guesses that do not reconcile. A DraftKings lineup contains exactly one quarterback, so the ownership of every quarterback on the slate has to total 100%. Two running backs and a share of the flex means the backs total about 230%, the receivers about 355%, the tight ends about 115%. That is arithmetic, and enforcing it stops the common failure where a model likes eight quarterbacks and quietly claims that 180% of lineups contain one.
What the model does inside that constraint is opinion, and it is stated plainly: players get rostered for value, with a pull toward expensive players who are obviously good, and for upside. How concentrated the field is depends on the contest, which is the only thing separating the three columns.
The three contest types
Single entry is the most concentrated, because a casual entrant with one lineup takes the obvious plays. High stakes is flatter and pays more for upside, because that field knows what the chalk is and is trying to get off it. Mass multi entry is flattest of all, because somebody filling two hundred lineups physically cannot put the same player in all of them. Those are assumptions about behaviour, and they are the honest part of an otherwise unvalidated model.
What is still missing
Defenses are not projected at all, so they appear with a salary and nothing else. Players whose betting lines have been pulled because of injury carry no projection either, and they are shown rather than hidden, because DraftKings is still offering them and a board that quietly drops them is lying about what the slate contains. There is no optimizer or contest simulation on this page yet.