TL;DR
Forezai has published Polybot, an MIT-licensed open-source trading-bot experiment for Polymarket that tests whether an AI agent can form probability estimates that differ from market prices. The project is framed as a research tool, not a trading recommendation, with repeated warnings about legal limits, automated-trading risk and possible total capital loss.
Forezai has released Polybot, an MIT-licensed open-source experiment that compares AI-generated probability estimates with Polymarket prediction-market prices and, in limited cases, can act on the gap, according to material published by Thorsten Meyer AI.
The project, available through forezai.com/polybot.html and GitHub, is described as a trading bot for Polymarket and a research tool for studying whether an AI agent can disagree with a live market price using public information. The source material says Polybot records the reasoning behind each estimate, so a decision can be reviewed after the fact.
Forezai says the system is designed to skip most markets by default. In the example provided, Polybot compares a market-implied probability with an AI estimate, checks whether the difference clears costs and confidence thresholds, and only then considers a small, risk-capped trade. The figures cited in the material are described as illustrative, not a record of past performance.
The announcement is heavily caveated. Thorsten Meyer AI says Polybot is experimental software, carries no guarantee of accuracy or profit, and should not be treated as financial, legal, tax or investment advice. The material also states that prediction-market access is restricted or prohibited in some jurisdictions, including for U.S. persons.
Polybot — when the AI disagrees with the odds
A prediction market puts a price on the future. Polybot asks: can an AI’s own estimate diverge from that price for real — and should it ever act on the gap?
Not financial, investment, legal or tax advice; not a recommendation or solicitation to trade, invest or use any software. Forezai · Polybot is experimental open-source software (MIT), provided “as is” without warranty of accuracy or profitability. Trading and automated trading carry a substantial risk of loss including total loss of capital; past or backtested performance does not indicate future results. Prediction-market participation is restricted or prohibited in some jurisdictions (including for US persons) — you are solely responsible for compliance with applicable law. Consult a licensed professional before any financial decision. Produced with AI assistance under human editorial oversight; independent commentary, the author’s own views. Product and company names are trademarks of their respective owners; mention does not imply endorsement.
AI Forecasts Meet Market Prices
Polybot matters because it applies AI forecasting to a market that already converts trader beliefs into live prices. A prediction-market contract trading at 62 cents is commonly read as a market-implied probability of about 62%, although fees, liquidity and market structure can affect that reading.
The project tests a narrow question: whether an AI system, using public information, can produce an estimate that differs from the market enough to justify attention. Forezai’s own framing is cautious. The material says the gap between an AI estimate and a market price is a hypothesis, not proof of an edge.
For readers following AI agents, open-source trading tools or prediction markets, the release is relevant less because it promises performance and more because it exposes the risk controls, assumptions and audit trail behind an automated decision process.
Use Claude to Build an AI Trading Bot: 90 Days with Stocks and Prediction Markets (AI Trading Bot Series Book 1)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Day 13 Adds Markets
Polybot is presented as Day 13 of a 19-day Built in Public series by Thorsten Meyer AI. The release opens the portfolio’s Markets family, joining other products in the operator portfolio listed in the source material.
The project’s stated design themes are local-first operation, provider-agnostic model use and inspectable forecasting. Forezai says the forecasting model can be swapped, and the system is not built around treating any single AI model as an oracle.
The source material repeatedly frames markets as difficult to beat because prices already combine information, trader judgment and capital. That framing makes Polybot’s central claim narrower than a promise of market-beating returns: it is positioned as a way to study when an AI estimate diverges from a live, adversarial market.
Use Claude to Build an AI Trading Bot: 90 Days with Stocks and Prediction Markets (AI Trading Bot Series Book 1)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Profitability Is Not Established
It is not clear from the source material whether Polybot has produced profitable live results, how much live trading has occurred, or how it performs across market types, liquidity conditions and resolution categories. The example shown is explicitly illustrative.
It is also unclear which model or models were used in any specific run, how often estimates are refreshed, what exact execution rules apply in production, or how the tool handles market manipulation, thin order books, stale information and resolution ambiguity.
Legal access remains a separate issue. The source material says prediction-market participation is restricted or prohibited in some jurisdictions, including for U.S. persons, and places responsibility for compliance on users.
open-source trading bot for Polymarket
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Open Code Faces Testing
The next test is whether developers and researchers inspect the code, run their own evaluations and examine whether the recorded reasoning is useful after market outcomes are known. Because the project is MIT-licensed, outside users can study, modify and audit the software, subject to applicable law and platform rules.
Future credibility will depend on evidence that separates live performance from illustrative examples, includes trading costs and slippage, and reports losses as plainly as wins. For now, the confirmed development is the open-source release, not a proven trading record.
AI + Prediction Markets: The New Edge: How to Use Artificial Intelligence Tools to Research, Scan, and Win in Prediction Markets (Markets Intelligence Series)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is Polybot?
Polybot is an open-source Forezai experiment for Polymarket that compares AI probability estimates with prediction-market prices and records the reasoning behind those estimates.
Is Polybot financial advice?
No. The source material says it is not financial advice and is not a recommendation to trade, invest or use the software.
Does Polybot prove AI can beat prediction markets?
No. The material frames the project as an experiment. It does not provide a confirmed profitable track record, and it warns that markets are hard to beat.
Can U.S. users trade with it?
The source material says prediction-market access is restricted or prohibited in some jurisdictions, including for U.S. persons. Users would need to verify what is lawful where they are before taking any action.
What happens when the AI disagrees with the market?
According to the project description, most cases are skipped. A trade is considered only when the gap appears large enough after costs and confidence checks, and even then the material says it can still be wrong.
Source: Thorsten Meyer AI
