The edge

Optimizers chase
a number.
We run the tournament.

A lineup optimizer answers "which nine players sum to the biggest projection?" That's the wrong question for a GPP, where 20% of entries get paid and the top 1% takes most of the pool. SimSlate answers the question you're actually betting on: how does this lineup perform against this field, in this payout structure, across every way Sunday can go?

Question
Typical optimizer
SimSlate
What is a player outcome?
One projected number, sometimes ± a random jitter
A fitted right-skewed distribution matching mean, ceiling, and boom/bust odds
Do outcomes move together?
No — stacking is a manual rule you toggle
Yes — a Gaussian copula correlates stacks, bring-backs, and DST matchups in every draw
Who are you playing against?
Nobody — lineups are scored in a vacuum
A 50,000-lineup field held to projected ownership within hard tolerance bands
What happens when chalk wins?
Full first-place prize, every time
Duplicated lineups split a rank band of prizes — the chalk tax is priced in
What gets maximized?
Projected points
Simulated ROI in your contest's actual payout curve
Design rules
what we refuse to do to your signal
No added
noise

Your inputs come back out undistorted

A common industry shortcut is to "add randomness" — jitter projections, normalize ownership to 100%, hand-tune fudge factors until output looks diverse. Every one of those moves injects noise between your research and your lineups.

  • Ownership is never normalized. A real NFL slate sums to ~900% (nine roster spots); rescaling it corrupts every exposure downstream.
  • Variance comes from the fitted model, not a dice roll. The spread in outcomes is each player's measured volatility — nothing is jittered.
  • Runs are reproducible. Pinned seeds give bit-identical results, so a difference between two runs is a difference in inputs, never in luck.
AI-calibrated,
statistically
governed

AI where it helps. Statistics where it counts.

SimSlate's models are built and calibrated with AI: boom/bust thresholds are learned from slate data instead of copied from folklore, distribution fits are auto-solved against five constraints per player, and the field generator self-calibrates through repeated passes until exposure fidelity is met.

But no AI touches your results at simulation time. The engine that scores 50,000 contests is closed-form statistics with proven invariants — conservation of prize money to the rake, negative ownership–ROI correlation within score deciles, positive semi-definite correlations. AI tunes the instrument; math plays it. You get the pattern-finding without a black box between you and your bankroll.

Built for
top-heavy
payouts

GPPs are won in the tail, so the tail is the product

Cash-game logic — maximize the median — actively loses money in tournaments. Everything in SimSlate is oriented to the payout curve's shape: skewed marginals preserve ceiling probability, correlations let stacks spike together, duplication modeling discounts the crowded routes to the top, and the final ranking is ROI in your contest's structure, from single-entry to milly-maker top-heaviness.

The result is a board that surfaces leverage — lineups with live paths to first that the field underweights — instead of re-serving you the chalk everyone already has.

The field is already set

Stop guessing what it looks like.

Simulate it — and enter the lineups the math likes, not the ones the crowd does.