Daily fantasy is a quantitative problem. We treat it like one.
Empire Sports Analytics is a Bayesian DFS simulator: Monte Carlo lineup generation built on hierarchical posterior projections. It's the toolkit a working professional would build for themselves if they had a few months and a stats background. Bayesian projections, correlated Monte Carlo, integer programming: published methodology, audited inside the product.
ESA is the work of quantitative engineers who spent careers modeling real-money decisions in markets where being wrong is expensive. The same toolkit a desk would use to size a position is the toolkit we point at a DraftKings slate.
We don't have podcasts. We don't post winning screenshots. We spend our energy on the model and the audit trail. When we ship something, we publish what it actually does and how often it's right.
Each dot: one player, one slate. The line is a perfect prediction. We measure how far the dots fall from it, every week.
How a projection becomes a lineup.
Every component (singles, doubles, walks, strikeouts, shots on goal) is modeled as a hierarchical Bayesian distribution conditioned on player history, opponent, park and weather priors, and the Vegas market. We sample the posterior at slate-load and serve the median, the spread, and the simulation distribution. No black box, no ensemble of opaque ML wrappers.
Lineups are scored against thousands of joint scenarios. The unique players in your lineup share a single set of draws. If Mookie Betts hits a home run in trial 47, every lineup that rosters him sees the same outcome. Stacks behave like real stacks. Bring-backs land where they should.
The optimizer solves a constrained integer linear program (salary cap, position eligibility, exposure ceilings, stacking rules) over the posterior, not over a point estimate. The lineup that maximizes mean isn't usually the lineup that maximizes ceiling; the optimizer knows the difference.
Inputs feed a hierarchical model. The posterior is what the model believes. MC + ILP turn belief into a lineup.
Honest fit. Not every player needs the same tool.
ESA is built for a specific kind of DFS player. The toolkit, the voice, and the way we publish our work all point in the same direction. Some readers will recognize themselves immediately; others will find a better match elsewhere. Both outcomes are useful.
Two lanes in DFS tools. One is built on influencer voice and content. This one is built on data, methodology, and published accuracy. Pick the lane that matches how you make decisions.
And one thing we will.
We will not claim a guaranteed ROI.
Daily fantasy contains irreducible variance. A model that improves your decisions is not a model that prints money. Anyone who tells you otherwise is selling you something other than a model.
We will not name competitors and assert we beat them without a public backtest.
We have not committed to the out-of-sample work that would defend such a claim. Until we do, we publish our own numbers and let you compare.
We will publish our numbers and let you audit them inside the product.
Weekly MAE, rank correlation, top-K overlap. Per sport, per slate. The dashboard is in the app, not behind a support email.