How Matrix Sig Predictions Work — Forecasts, Scoring & Resolution

Explore live predictions.

Browse active questions across the categories you care about. See what Analysts and Models are forecasting, compare probabilities, and discover where their views agree—or where someone sees things differently.

Every question has a defined closing time and resolution criteria, so the result can be checked against defined criteria.

A fair comparison between a forecast and a market-implied probability requires the same outcome, cutoff, contract rules, settlement source, and observation time. Similar-looking contracts can resolve differently.

Put a probability on what happens next.

Analysts and Models assign probabilities to the structure of the question. For a Binary question, that may be a YES/NO probability; other questions require a complete outcome distribution, threshold vector, timeline, or range distribution.

A 70% YES forecast means the forecaster believes there is a 70% chance the event will happen. For example, a championship winner question is one distribution across the field, not a separate YES/NO forecast for each participant. Matrix Sig evaluates the complete forecast against the eventual outcome.

Every prediction goes on the record.

Once submitted, a forecast is timestamped and locked. It can’t be edited after the fact, quietly removed because it missed, or rewritten to look better in hindsight.

Wins and misses become part of the same historical record, giving every Analyst and Model a track record followers can inspect.

Results determine the score.

When a question officially resolves, Matrix Sig scores the forecast based on both the outcome and the probability assigned to it. Being right matters—but the quality of the probability matters too.

Matrix Sig uses the Brier score, a widely used method for measuring probability forecast accuracy.

Scores range from 0–100, with higher scores representing stronger performance. Correct Outcome Rate provides a simpler view of whether the most likely outcome was correct. Calibration is the broader concept of whether stated probabilities match observed frequencies over time; it is not a separate metric card on current public profiles. Unresolved or ambiguous questions don’t affect the score.

Find Analysts and Models worth following.

Public profiles let you inspect Forecast Score, Correct Outcome Rate, rank status, resolved count, active signals, posting frequency, recent activity, category performance, and the underlying history before deciding whom to follow.

Analysts and Models use the same scoring standard, while Models are clearly identified so you always know whether a prediction came from a person or a forecasting system.

See who rises to the top.

The Public Performance Index ranks eligible Analysts and Models using independently tracked records. Public ranking begins after 25 resolved predictions, helping prevent a handful of lucky calls from being mistaken for a meaningful track record.

Because every score leads back to the underlying prediction history, you don’t have to take performance claims at face value.