When you press SIMULATE GAMES, the engine plays every game on this slate from kickoff to the final whistle — one play at a time. Drives stall or extend, quarterbacks scramble, red-zone trips turn into touchdowns, field goals, or turnovers, clocks get managed, and teammates' outcomes rise and fall in a correlated manner - the way they do in real football. Then it plays the whole slate again, thousands of times over. Each complete playthrough is one possible version of the week, so instead of a single guess, every player ends up with a full range of outcomes — that's what fills Sim Proj, Ceiling, and Sim GPP%, and it's why the run takes a moment.
The contest tools — Rank Lineups, Sim Target Own %, and the Portfolio scorers — drop your lineups into a realistic modeled field of opponent's lineups inside those same simulated weeks, with the current contest's payout structure accounted for. That's how a lineup earns a win rate and an ROI prediction instead of just a projection, and how Target Own % can tell you where the winning lineups tend to be overweight or underweight versus the crowd.
Your numbers stay the boss. Most tools sim around their own projections. Ours wrap the full range of outcomes around your projections (or Kritikal Hit's, at the flip of a toggle) — the engine adds realism and correlation, it never quietly overrides your read.
One shared universe. Every tool on this site — lineup ranking, target ownership, portfolio scoring — reads from the same set of simulated weeks. Numbers here agree with each other, because they all come from the same worlds.
We price regression, not just projection. The Luck engine and the Leverage Lens estimate how much of a hot or cold stretch will tend to carry forward — and every one of those weights is re-tested against real league-wide results, every week. If a signal stops proving itself, it loses its influence and our models adjust accordingly.