Am I better than the superforecasters? The prediction markets? The think tanks? The honest answer includes where each of them beats me. A map of the forecasting landscape by the only axis that matters — whether a stranger can grade you against the public record.
The fair question, asked often, is: are you actually better than the superforecasters, or the prediction markets, or the big strategic-intelligence shops? It deserves a straight answer, and a straight answer has to include the places where each of them beats me. So this is a map of the forecasting landscape, sorted by the one axis that actually separates these worlds — not who is smartest, but who can be graded by a stranger against the public record.
The field has roughly four kinds of players. Superforecasters, in the Good Judgment Project tradition, are tournament-scored over large numbers of questions and post a published Brier benchmark of 0.149. Prediction markets — Polymarket, Metaculus, Kalshi — turn a crowd's money or reputation into a live probability, and they are genuinely excellent on liquid, popular, binary questions. Think tanks like RAND and Stratfor produce deep, narrative analysis that is rarely expressed as a graded probability. And private data and asset shops — Bridgewater, Palantir — may forecast superbly, but their calls are proprietary and never publicly falsifiable.
Sort all of them by accountability and the picture clarifies. Can an outsider grade you against the public record? Superforecasters: yes, inside their tournaments. Markets: yes — prices resolve to a settled outcome. Think tanks: usually not — prose with no fixed rubric and no pre-event seal can always be read, after the fact, as having been sort of right. Private firms: no, by definition. That axis, not raw cleverness, is what most 'who's the best forecaster' arguments are really about.