How The Outlook works

The Outlook publishes dated, probabilistic forecasts of iGaming regulatory and market change — and keeps score in public. This page explains where the probabilities come from, how questions resolve, how we measure our own accuracy, and what we will never forecast.

Where the probabilities come from

There is no statistical model under the hood, and we say so plainly. Each probability is an editorial judgment built the way the forecasting-tournament literature says accurate forecasts are built: start from a base rate — how often events of this class actually happened in our tracked change corpus (every question shows its comparison class, with the counts) — then adjust for dated, cited signals specific to the question. Every forecast entry records its date, its probability and the reason it moved. Probabilities are stated in five-point steps between 5% and 95%: never 0% or 100%, and no false precision.

How questions resolve

Every question is pre-registered: its resolution criteria and its resolving source — always a named primary source, never secondary reporting — are written before the question opens, and the forecast history is append-only from that moment. When the resolution date arrives or the resolving event occurs, the question is graded YES or NO against those criteria, with a citation. If a question’s premise disappears before resolution, it is marked void: excluded from scoring, but kept visible in the archive. Nothing is ever deleted or rewritten.

How we score ourselves

Resolved questions are scored with the Brier score — the squared error of the probability, where 0 is perfect and lower is better. We benchmark every score against a naive forecaster who always predicts each question’s historical base rate; beating that benchmark is the entire test, and our running comparison is public on the track record page, misses included. A calibration curve — do our 70% forecasts come true about 70% of the time? — appears only once 30 questions have resolved, because a curve drawn on fewer would be noise dressed as rigor.

What we will never forecast

The Outlook covers industry events only: regulation, tax, market openings and closings, standards and enforcement trends. We never forecast game outcomes, betting results or anything a reader could mistake for betting advice, and we do not forecast the fortunes of named companies. iGamer is an independent, non-promotional reference — the forecasts exist to make the industry legible, not to tip anything.

Limits — read before citing

Forecasting research is blunt about its edges: rare, unprecedented shocks are structurally hard for every method, expert or model, and probabilities for them are poorly calibrated across the board. We therefore concentrate on recurring, base-rate-rich classes of events and treat every number as an estimate under stated assumptions — not certainty, and not advice. Each question links its sources; cite the primary source first, and our forecast as where you found it.