Polysimplr vs
SimpleFunctions.

SF's thesis system accepts a plain-language claim — 'the Fed will cut rates before September' — and decomposes it into a causal tree of testable sub-claims. Each node is mapped to tradeable contracts on Kalshi and Polymarket, probabilities are propagated through the tree, and an evaluation heartbeat runs continuously: news scan, price refresh, milestone check, LLM evaluation, and confidence update. You can inject new signals via /api/thesis/{id}/signal and fork any public thesis. An AI chat answers questions; the thesis system maintains a living, probabilistic model of a claim.

Yes. SF ships 56 MCP tools accessible via claude mcp add simplefunctions --url https://simplefunctions.dev/api/mcp/mcp. The server integrates with Claude Code, Cursor, and any MCP-compatible client. Tools cover market search, cross-venue price lookup, orderbook depth, thesis creation and signal injection, trade ideas, calibration data, and the autonomous trading interface. Unlike browser-based analytics tools, the MCP server is designed for agent-to-agent communication: your AI agent queries SF's world model and acts on the result.

Portfolio Autopilot is an autonomous trading agent that uses a 1M-context LLM and 13 data sources to evaluate positions continuously. Before any execution, the system runs a 7-gate risk cascade: kill switch check, position limits, drawdown gate, regime check, and additional safeguards. The agent synthesises thesis confidence scores, cross-venue pricing, orderbook depth, and indicator signals before sizing and placing orders. It is not a rule-based bot; it reasons over the full context window each cycle and can be halted via a kill switch at any time.

Based on Polysimplr's public-facing description, it is a Polymarket interface with an AI chat layer. We have not observed public documentation indicating Kalshi coverage. SimpleFunctions normalises both Kalshi and Polymarket into a single API, covering 48K+ active contracts across both venues, with cross-venue matched pairs available at /api/public/cross-venue/pairs?preset=arb for users who want arbitrage opportunities between the two markets.

SF publishes live Brier scores at /api/calibration, broken down by venue, category, and price bucket. Current figures: Kalshi 0.20, Polymarket 0.12 on T-24h price, computed over the past 90 days. Brier score measures probabilistic forecast accuracy — lower is better. Most prediction market data vendors claim accuracy in marketing copy; SF makes the underlying scores publicly queryable so you can verify them with a single curl request before trusting any model output or autopilot recommendation.

SF pre-computes six indicators across 48K+ active contracts, updated continuously: IY (implied yield — annualised return to resolution), CRI (cliff risk index — sensitivity to near-term adverse events), LAS (liquidity availability score — executable size at fair value), EE (event overround — total book edge above 100%), τ-days (time to settlement), and a regime label (adverse-selection classification). These are available at /screen and via the MCP server, without requiring you to derive them from raw price and volume data.

No authentication is required for read operations: cross-venue prices, orderbook depth, computed indicators, trade ideas, calibration data, the world snapshot, and the 56-tool MCP server are all openly accessible. The thesis system and intent execution (autonomous trading) require authentication and are free up to 15M tokens, with pay-per-token beyond that threshold. The CLI is MIT-licensed and available via npm i -g @spfunctions/cli; read operations work without a key.