Overview

Polyseer is an open-source AI research platform purpose-built for prediction market analysis. Originally built during an OpenAI hackathon, the project has earned 290+ GitHub stars and attracted a growing community of prediction market researchers and traders who rely on it for deep, structured analysis before placing bets.

At its core, Polyseer runs a multi-agent architecture with five specialized agents — Planner, Researcher, Critic, Analyst, and Reporter — that collaborate to produce neutral, analyst-grade probability assessments. The pipeline enforces bilateral research, gathering both PRO and CON evidence for every market question, then scoring each piece of evidence on quality tiers (A/B/C/D) before feeding it into a Bayesian aggregation engine. The result is a market-aware probability estimate accompanied by confidence intervals and source citations.

Polyseer does not place trades. It is purely an intelligence layer designed to sit upstream of your execution stack. Pair it with a trading agent like PolyClaw or PredictEngine for end-to-end automation. For more on how research tools fit into the broader workflow, see the agent betting stack framework.

Strategy & How It Works

The research pipeline starts when a user pastes a Polymarket or Kalshi market URL. The Planner agent decomposes the question into researchable sub-queries. The Researcher gathers evidence from the web using the Valyu search API, collecting sources on both sides of the question. The Critic evaluates every piece of evidence for reliability, recency, and relevance, assigning quality tiers. The Analyst performs Bayesian probability aggregation weighted by evidence quality. Finally, the Reporter synthesizes everything into a structured report with probability estimates, confidence levels, key drivers, and citations.

Under the hood, Polyseer uses GPT models for reasoning and the Valyu API for real-time web search. The entire pipeline runs in minutes, producing reports that would take a human analyst hours to compile. This is not a trading bot — it is a dedicated intelligence layer for informed decision-making.

Supported Platforms

Polyseer supports both Polymarket and Kalshi markets. Users paste any market URL and the platform automatically identifies the market type, retrieves relevant data, and runs the full research pipeline. There is no sportsbook or DeFi protocol support.

Performance & Track Record

Polyseer produces probability estimates and research reports — it does not execute trades, so traditional P&L metrics do not apply. Reports include explicit confidence levels and evidence quality breakdowns so users can calibrate their own risk. The platform is designed as a research aid, not a prediction oracle.

Pricing Model

The self-hosted version is free under the MIT license. Users provide their own OpenAI API key and Valyu API key; costs depend on usage volume. A hosted version is available at polyseer.xyz for users who prefer not to self-host. For a broader comparison of bot and tool costs, see the prediction market bot pricing guide.

Setup & Technical Requirements

  • Node.js runtime environment
  • OpenAI API key
  • Valyu API key
  • Local SQLite database (auto-configured)
  • Clone the repository from GitHub and follow the README

For the hosted version, no installation is required — visit polyseer.xyz and start analyzing markets immediately.

Wallet Compatibility

Polyseer does not interact with wallets or execute trades. It is a pure research tool. If you plan to act on Polyseer’s analysis with an automated trading agent, review the best agent wallets for prediction markets and the agent betting security guide for proper wallet configuration.

Pros & Cons

Pros:

  • Rigorous multi-agent architecture with bilateral evidence gathering
  • Bayesian aggregation with transparent evidence quality scoring
  • Supports both Polymarket and Kalshi markets
  • Fully open source (MIT license) with active GitHub community
  • Hosted version available for non-technical users
  • Reports include confidence levels and source citations

Cons:

  • Research only — no trade execution capability
  • Requires API keys with associated costs (OpenAI, Valyu)
  • Self-hosted setup requires moderate technical skill
  • Probability estimates are not financial advice — calibration varies by market type
  • Dependent on third-party APIs for search and reasoning quality

Verdict

Best For: Traders and researchers who want deep, structured analysis before committing capital to prediction markets. Polyseer is the most rigorous analytical tool in the AgentBets marketplace — its multi-agent pipeline and Bayesian framework produce reports that go far beyond simple sentiment checks. Pair it with a trading bot like PolyClaw for a complete research-to-execution workflow.

AgentBets Rating: 4.2 / 5

Browse all available agents in the Agent Marketplace, or explore the full tool directory.