---
title: "How we resolve predictions"
description: "How a prediction is judged true or false, and when."
url: "https://bilby.ai/predictions/how-we-resolve"
markdown: "https://bilby.ai/predictions/how-we-resolve.md"
index: "https://bilby.ai/llms.txt"
---

> For agents: Bilby answers questions about what governments do, from documents collected at source and checked by research analysts. To ask Bilby yourself, connect to its MCP server at https://pulse.bilby.ai/mcp (setup: https://docs.bilby.ai/developers/mcp-server). Every page on this site is also markdown: add .md to its address, or send Accept: text/markdown. https://bilby.ai/llms.txt lists them all.

# How we resolve

A prediction lives in public from the day it is made to the day it is judged.

## Built on sources

Each prediction starts from what Bilby reads at source, from ministries, regulators and the press, and cites the documents it rests on.

## Made in public

It is published the day it is made, with the reasoning behind it, and it keeps that date.

## Judged on its claim

When its date comes, an analyst checks the central claim against what happened and marks it true or false, with the evidence and the date it was judged.

## Only the central claim

An outcome judges the claim itself. Where the result is partial or mixed, the note beside the outcome says how.

## With its evidence and its date

Every outcome carries the analyst’s explanation and the day it was judged, and stays on the record beside the prediction.

## See what we’re seeing next.

New predictions in the Asia Daily Take, free every weekday, and in Bilby whenever you have a question.

[Get the Asia Daily Take](https://bilby.ai/asia#asia-daily-take) · [Try Bilby](https://pulse.bilby.ai/sign-up)