Polyseer vs
SimpleFunctions.

POST /api/thesis/create takes any English sentence — for example, 'The Fed will cut rates before September' — and decomposes it into a tree of testable sub-claims. Each node carries a probability estimate that propagates up to the root. An evaluation heartbeat then runs on each active thesis: news scan, price refresh across Kalshi and Polymarket, milestone check, LLM evaluation, and confidence update. You can inject external signals via /api/thesis/{id}/signal and fork public theses. No other prediction market platform exposes structured causal reasoning at this depth.

Portfolio Autopilot is SimpleFunctions' autonomous trading agent. It runs a 1M-context LLM that ingests 13 data sources — current positions, live market prices, thesis confidence scores, regime labels, cross-venue pairs, and more — and passes every potential trade through a 7-gate risk cascade before execution: kill switch, position limits, drawdown gate, regime check, and others. It is designed for operators who want autonomous execution with auditable risk controls, not manual trade-by-trade entry.

Yes. Both platforms aggregate data from Kalshi and Polymarket. SimpleFunctions normalises prices across 48K+ active contracts and exposes them through a unified REST API, MCP server, and CLI. Cross-venue matched pairs — including arbitrage opportunities — are available at /api/public/cross-venue/pairs. Polyseer applies Bayesian multi-agent aggregation on the same venues to produce research reports. The venue coverage overlaps; the output format and intended consumer differ.

Yes. Polyseer is published as an open-source project, meaning you can audit the code, run it on your own infrastructure, and modify it. SimpleFunctions is a managed SaaS API — the CLI is MIT-licensed, but the core inference and thesis engine run server-side. If self-hosting and full code auditability are hard requirements for your research workflow, Polyseer's open-source model is a genuine advantage.

SimpleFunctions ships 56 MCP tools accessible via: claude mcp add simplefunctions --url https://simplefunctions.dev/api/mcp/mcp. The tools cover market queries, thesis creation and signal injection, orderbook depth, cross-venue pair matching, calibration retrieval, world snapshot, trade ideas, and more. Any MCP-compatible client — Claude Code, Cursor, and others — can consume them without additional configuration. Polyseer does not publish an equivalent MCP server.

/api/calibration returns SimpleFunctions' own Brier scores broken down by venue, category, and price bucket. Current figures: Kalshi 0.20, Polymarket 0.12 on the T-24h price, computed over the past 90 days. This lets you audit SF's predictive accuracy independently before relying on it. Polyseer does not publish equivalent calibration benchmarks publicly, so accuracy comparisons are not possible from the outside.

If your goal is to produce natural-language research reports on Polymarket or Kalshi events using a Bayesian multi-agent pipeline you can inspect and run yourself, Polyseer is purpose-built for that. SimpleFunctions' thesis system produces structured causal trees with probability propagation — output optimised for machine consumption and agent pipelines, not human-readable prose. The two tools serve different output formats and different downstream consumers.

Yes. Public REST endpoints — including /api/public/markets, /api/public/cross-venue/pairs, /api/public/ideas, /api/calibration, and orderbook depth — require no authentication. Thesis creation, signal injection, and intent execution require an account and are free up to 15M tokens, then billed pay-per-token. Polyseer is open-source and self-hosted, so access is controlled by whoever deploys the instance.