NFL Showdown Scenarios
A showdown is six players from one game, one of them a captain who scores and costs one and a half times. Because every lineup is in the same game, asking which game to stack is meaningless here. The questions that actually separate two lineups are who captains, how the six split across the two sides, and what kind of game it turns into. So that is what this measures: it plays the game out thousands of times with players correlated, solves the lineup that would have won each one, and groups those winners.
These are win rates, not leverage. Every number here is measured from the simulation: the share of games whose best possible lineup had that shape. There is deliberately no field or leverage column, because a showdown field needs a model of captain ownership, which sums to 100% across the slate while flex ownership sums to 500%, and no such model exists here yet. Your payoff in a tournament depends on how many other people built the same thing, so until that denominator is measured rather than guessed it is left out instead of filled in. Read this page as which constructions win, not as which are uncrowded.
Captains
The captain slot is the one real lever in this format. A leveraged captain compounds, because you get one and a half times his score inside a small slice of the field, and no slot in a classic lineup does that.
Shapes
The captain's position together with the team split. 4-2 means four players from the captain's own side and two from the other. DraftKings requires at least one from each team, so 6-0 cannot be built.
Team split on its own, ignoring who captained:
Game script
With one game you cannot ask which game goes off, so you ask how this one goes off. Each simulated game is bucketed by how much the two sides produced and how far apart they finished, using this game's own distribution rather than fixed point totals, and the winning shapes are reported inside each bucket. This is the part that should change what you build.
How this is built
The solve is exact, which matters more here than on a main slate
A single-game pool is about two dozen players once the unprojectable deep roster is set aside, and a lineup is six slots. That is small enough to solve properly: every player is tried as captain, and every legal way of filling the other five slots is scored against the cap. Nothing is left to a greedy search.
That is not a detail. A captain is decided by a single point of score, so an approximate fill would hand the slot to the wrong player while still producing a plausible-looking table. The solver is checked against an independently written brute force over random games and agrees to within a rounding error, and it is separately checked to never break the salary cap, never duplicate the captain into a flex slot, and never return a lineup missing one of the two teams.
Why a showdown suits this method better than a main slate
On a twelve-game slate, saying a construction is uncrowded requires modelling what several hundred thousand entries look like across nearly six hundred players, and that model is only as good as its ownership projections. Here the pool is the game. All the correlation that matters is inside it, which is exactly what the simulation models best, and there are no cross-game independence assumptions left to get wrong.
What it does not tell you
It does not say which way the game will go. Nothing can. It says which six-man constructions win if the game breaks a particular way, and how often each way happens in the simulation, which is the part that is knowable before kickoff. It also does not yet say how many other people built the same thing, and that gap is stated on the page rather than papered over with a number.