Methodology

Outlier's core engine is a Monte Carlo simulation. For every game on the slate, the model generates fifty thousand simulated outcomes built from team-level efficiency, recent form, contextual factors, and real-time inputs. The output isn't a prediction — it's a probability distribution.

That distribution is then compared against the market's implied odds at major sportsbooks. The gap between the model's probability and the market's implied probability is the edge. When the edge is meaningful — and only when it's meaningful — a play is released.

A model is only as good as the discipline applied to it. Outlier applies hard guardrails on every play: minimum edge required, position sizing tied to confidence, and overrides for situational factors that the model alone can't see — back-to-back fatigue, key injuries, and starting pitcher news among them. Plays that don't clear every check don't get released.

Every play is logged the moment it's released, with the line and the edge captured at release. After the game closes, two things are recorded: the result, and the closing line. Closing line value — the difference between the line at release and the line at close — is the long-run signal that the edge was real. Outlier publishes both.

Outlier is not a lock service. It's not “guaranteed winners.” Sports betting involves variance, and any honest analytical product loses some weeks. The point of the model is positive expected value across thousands of plays, not a perfect record across ten.

Read the track record and decide for yourself.