Prediction markets tell you what might happen; Polyseer tells you why. This project provides systematic, AI-driven analysis for prediction markets like Polymarket and Kalshi, aiming to deliver rigorous, evidence-based insights beyond simple speculation. It targets developers, researchers, and users seeking a structured understanding of market drivers. By leveraging multiple AI agents and Bayesian probability, it synthesizes information from academic papers, news, and market data to forecast outcomes, offering a data-driven alternative to gut feelings.

How It Works

The core of Polyseer is a multi-agent AI architecture that orchestrates specialized agents (Planner, Researcher, Critic, Analyst, Reporter) to conduct deep analysis. It utilizes the Valyu search network to gather and classify evidence from diverse sources, including academic papers, web intelligence, and market data. This evidence is then aggregated using sophisticated Bayesian probability models, incorporating log-likelihood ratios and correlation adjustments, to produce objective probability assessments and a detailed analytical report. This systematic approach aims to mitigate confirmation bias and provide a data-driven perspective on market questions.

Quick Start & Requirements

• Installation: Clone the repository (git clone https://github.com/yorkeccak/polyseer.git), navigate into the directory (cd polyseer), and run npm install (or pnpm install).
• Prerequisites: Node.js 18+ and a package manager (npm, pnpm, or yarn).
• API Keys: Requires OPENAI_API_KEY (for GPT-5 access) and VALYU_API_KEY (for Valyu Search Network). An optional POLYMARKET_API_KEY can enhance data fetching.
• Modes: Development mode (default) uses personal API keys, requires no signup, and offers unlimited usage. Production mode requires Supabase for database/authentication and Polar for billing, enabling monetization and user management.
• Hosted Version: A hosted version is available at polyseer.xyz.

Highlighted Details

• Systematic research across academic, web, and market data sources.
• Evidence classification (Types A-D) with quality scoring.
• Mathematical probability aggregation via Bayesian updating and log-likelihood calculations.
• Multi-agent AI architecture for deep, multi-faceted analysis.
• Technology stack includes Next.js 15.5, React 19, GPT-5, Valyu JS SDK, Supabase, and Polar for billing.

Maintenance & Community

The project welcomes contributions with a defined development workflow, including forking, feature branching, testing, and pull requests. Code style is enforced via TypeScript strict mode, ESLint, Prettier, and Conventional Commits. Specific community channels (e.g., Discord, Slack) or details on notable contributors and sponsorships are not detailed in the README.

Licensing & Compatibility

This project is licensed under the MIT License. This permissive license allows for broad use, modification, and distribution, including integration into commercial and closed-source projects without significant restrictions.

Limitations & Caveats

Polyseer is explicitly stated as "NOT FINANCIAL ADVICE" and is intended for "entertainment and research purposes only." Its core functionality is dependent on external API keys (OpenAI, Valyu), which may incur costs and are subject to associated rate limits. Setting up the production mode requires significant configuration of Supabase and Polar services for database, authentication, and billing.