Learn
How the numbers are made and how to use them.
How the projections are made
Each player's projection blends the betting market (the player's props, the game total and spread, the team's implied total) with a usage model built from play-by-play: snaps, targets, carries, red-zone work, and what changes when a team-mate is out. The two are weighted by how much each has been worth on past seasons, and the blend is re-fitted above each position's roster floor.
Ownership is projected from salary, projection, position and slate size, calibrated on real contest standings, and re-projected at the size of the contest you are building for.
Every build runs 10,000 simulated slates with player correlation (a quarterback and the receivers rise and fall together), and lineups are ranked by how they finish against a simulated field, not by projection alone.
What the tiers mean
Core plays are high projection and high projected ownership: the field is right about them. They are the chalk piece of a tournament lineup, and the write-up says how much of each to carry; the cash call beside a core play is made from the numbers (a floor that holds 60% of the projection, or a 68% favourite in a fight), not from the label. When the write-up says a core play is the least certain of the tier, that is the recommendation.
Value plays are top-band points per dollar at the position with ownership below what the projection deserves. That is where leverage lives.
Spend-ups are expensive players whose ceiling, not projection, justifies the price. The page shows both numbers so a trap at the price is visible too.
Captains (Showdown) are the players whose captain seat the field leaves emptier than their flex seat deserves.
Every play comes with the reasons in numbers and the risks: an injury designation, a role that depends on a team-mate staying out, a simulated bust rate, chalk.
Shortlist, boost, lock, exclude
Shortlist picks a player out in the optimizer's pool and changes nothing about the model: the row is highlighted and ready to lock, group or boost by hand.
Boost leans the optimizer toward a player: the projection is raised, nobody is removed, and the whole set is one undoable research selection you can see and clear in the optimizer. The page says "+projection" beside it so the modelling change is never hidden.
Lock puts the player in every lineup. Exclude keeps the player out of all of them. Both are your own edits on the player and count toward DraftKings' rule that tool-built lineups carry at least two adjustments of yours.
DraftKings and contest entry
Clarity never enters a contest or handles money. You export a CSV and upload it on DraftKings yourself. Late swap is supported for NFL where DraftKings allows it: locked players stay, open slots are rebuilt.
After the games, upload the contest standings to see your entries against the field. Clarity's own uploads also grade the public pages.
Clarity in Claude
The Clarity connector lets Claude open a slate, read the plays, make your changes and build lineups by conversation. How to connect it