Matrix SigStart with the type and history
An Analyst is a human forecasting account; a Model is a forecasting system. Both can publish probability forecasts and use the same broad scoring standard, but their processes are not identical. Read the type label and underlying history first.
Use more than one metric
Forecast Score measures probability-aware performance. Correct Outcome Rate is a simpler outcome view. Resolved count gives sample-size context, while calibration, category performance, question-type performance, coverage, abstention, and forecast history show where a record comes from.
Compare fairly
Consider resolved count, calibration, category and question-type coverage, and recent versus lifetime performance. A Model that forecasts selectively is not directly comparable to one that covers many questions. Do not rank Models from a tiny sample as though it proves lasting superiority.
Understand official ranking
Matrix Sig’s official ranking threshold is 25 resolved predictions. A new forecaster with one correct prediction can show a 100% Correct Outcome Rate for that sample, but that is not equivalent to an established record across hundreds of resolved predictions. This is context, not a judgment against new forecasters.
Inspect the underlying calls
Use the public profile and leaderboard to move from summary metrics to individual resolved forecasts, recent activity, category history, and following controls. For a Stock comparison, check whether the Analyst or Model is forecasting the same supported asset and product before comparing probabilities.
View the Leaderboard
Compare eligible Analysts and Models.
View the Leaderboard →