What this is,
and what it isn't.
Short version: this is an automated experiment that publishes machine-generated guesses about the future. Some of them are wrong. Do not make decisions that matter on the strength of a number you read here.
This is information, not advice
Nothing on this site is financial, investment, trading, betting, legal, tax, medical or professional advice of any kind, and nothing here is a recommendation or solicitation to buy, sell or stake anything. The probabilities are produced by language models answering a prompt. They are not research, they are not a forecast from anyone qualified to give one, and no human reviews them before they publish.
If a number here would change what you do with your money, your vote or your health, treat that as a sign to go and read a primary source instead.
Accuracy is not promised
The whole point of the scoreboard is that the models are often wrong, and we publish the misses as prominently as the hits. Questions are rewritten by a model, which can subtly change what is being asked. Outcomes are decided by exchange settlement, by a panel of models reading cited sources, or by hand — and each of those can be mistaken.
The site is provided as is, without warranty of any kind, express or implied, including merchantability, fitness for a particular purpose and non-infringement. To the fullest extent the law allows, the maintainers are not liable for any loss or damage arising from use of this site or reliance on anything published here.
If you think a published outcome is wrong, say so in public. Corrections are part of the design, not an embarrassment.
Not affiliated with anyone named here
Market prices are read from the public APIs of Polymarket and Kalshi. Model answers are obtained through OpenRouter. This project is independent of those services and of every AI lab whose model appears on the panel. No endorsement in either direction is implied, and all trademarks belong to their owners.
Prediction markets are regulated differently from one country to the next, and in some places they are not available to you at all. Whether you may use them is your responsibility to determine, not ours. We take no position on it and link to them only as the source of a price.
Links and third parties
Pages link out to news sources and exchanges. We do not control that content and are not responsible for it. Following a link takes you under someone else's terms.
What we collect
Google Analytics measures traffic in aggregate, and counts four things people do here: voting, sharing, scrolling far enough to reach the model answers, and opening a question's resolution criteria. Each is recorded against the forecast it happened on, never against you — no account is required and no profile is built. Votes are additionally counted per question against a short-lived fingerprint, so one browser cannot stuff the tally. If you open an issue or submit a suggestion, that goes to GitHub under their terms and is public.
Which site these terms cover
This one. The source is MIT licensed, so other people may be running their own copies; those are their sites, with their questions and their answers, and nothing here speaks for them.
Changes
These terms may change; the current version is always the one on this page, and its history is in the public repository like everything else.
Written to be read rather than to be airtight. It is not legal advice and has not been reviewed by a lawyer — if you are relying on this project for something consequential, that is already the mistake this page is warning you about.