There are a lot of differences between forecasting and prediction platforms, such as how they let users make predictions, how they figure out what "actually happened," what technology stack they employ, how they calculate rewards, and so on.
Trepa is a "precision forecasting" platform that is now in public testing. It enables players wager on continuous or numerical forecasts (like the inflation rate or bond yields) instead of binary yes/no predictions. It also awards partial correctness and uses official data sources.
This article develops a classification of such projects, drawing on Trepa as the primary case study while contrasting it with established alternatives like Polymarket, Augur, Metaculus, and UMA's Optimistic Oracle.
Using these, we can group existing platforms (including Trepa) into categories.
Below are categories based on different combinations of those dimensions, with Trepa highlighted and compared to other platforms that already exist. For each category I analyze the strengths & weaknesses.