UNOWNEDFANTASY
21+ · NOT GUARANTEED · 1-800-GAMBLERGet the slate email

How it works

Everything on this site is computed from public data by a pipeline that runs every morning. Here is what it does, in the order it does it, and what the numbers mean.

1. Data

Nothing here comes from a paid feed. The nightly job pulls:

2. Projections

Each player's DraftKings points are projected from a blend of his last 5, last 10, season and prior-season averages, then adjusted for the opponent's defense against his position (shrunk toward average when the sample is small), the implied team total from the DraftKings line, and home or away. Players listed Out or Doubtful are zeroed. The output is a median, a floor (roughly the 20th percentile) and a ceiling (roughly the 85th), plus the chance of returning 2x, 3x, 4x and 5x salary.

3. Ownership and leverage

A gradient boosted model trained on more than 6,000 real DraftKings contests predicts each player's ownership in a large field tournament, a small field tournament and cash games, from salary, projection, value, slate size and contest shape. On the most recent 15% of slates held out of training, it misses by about 3.7 ownership points per player in NFL, 6.9 in NBA and 6.8 in NHL.

Leverage is a player's share of the slate's total ceiling divided by his projected ownership. Above 1.2 the field is under-rostering his upside, below 0.6 you pay full ownership price for it. The Unowned ladder chart on every slate draws exactly this.

4. Props

Every prop line is scored two ways. The model gives a probability from the player's own distribution of that stat (normal for yards and high counts, Poisson for low counts), adjusted for matchup and environment. The market gives one by removing the vig from the two sides of the line. We show a blend, 35% model and 65% market, because the market prices late news and role changes that box scores have not caught up with. When the model and the market disagree by more than 25 points we label the line "check news" and keep it out of the leans, because that gap almost always means something has changed.

Confidence labels: pass under 2 points of edge, lean 2 to 4, solid 4 to 6, strong above 6. Where a book has posted a line but no price yet, the lean is "model only" and shrunk toward a coin flip.

5. Plays and fades

Plays are ranked within position by a blend of ceiling (50%), points per dollar (30%) and leverage (20%). Core is a high ceiling player who will be popular and is worth it anyway. Pivot is similar upside at lower ownership. Punt is a cheap price that unlocks the stars. Players flagged questionable or day to day are excluded from plays.

Stay-aways are popular or expensive players with a specific, named problem: injury risk, chalk without the ceiling, a tough matchup, a shadow corner, minutes trending down, a goalie facing a loaded offense, or a price the role does not support. The reason is written out from the numbers every time.

6. Payouts and prices

On the Props & Lines page every price carries what a $100 bet wins, profit only. Sportsbook odds are American. Polymarket prices are cents per share and the price is the probability, so a 40¢ share wins 60¢ on a dollar, $150 on $100. Pick'em payouts (DK Pick 6, PrizePicks, Underdog) are fixed by the number of picks and listed on that page, labeled as standard payouts because the apps run boosts.

7. What we do not do

We do not guarantee anything. We do not sell picks. We do not take affiliate money that changes what we show, and if we ever add affiliate links they will be labeled. We do not scrape anything behind a login or a paywall. And we do not delete a bad call: the Record page grades every play, fade and lean against the box score overnight.