Alternative · AI agent
PolyOracle vs
SimpleFunctions.
Same Polymarket feed. PolyOracle runs multiple large language models in parallel and surfaces consensus signals on the most active markets. SimpleFunctions ships the full agent layer above it: causal-tree thesis system, autonomous trading with a 7-gate risk cascade, calibrated world model, computed indicators across 48K contracts spanning Kalshi and Polymarket, and a 56-tool MCP server that drops into any MCP client in one command.
Verified 2026-04 · public sources only · live SF data from /calibration
Verdict
Pick the one that fits how
you actually work.
Choose SimpleFunctions if
You are building agents, autonomous trading systems, or research pipelines that need more than consensus signals — calibrated probabilities with public Brier scores, causal-tree thesis modelling that decomposes any claim into testable sub-theses with auto-evaluation cycles, regime classification across the full 48K-contract universe on both Kalshi and Polymarket, computed indicators (implied yield, cliff risk, liquidity availability score, event overround), and a 56-tool MCP server that integrates with Claude Code or Cursor in a single command.
Choose PolyOracle if
You want a focused multi-LLM consensus layer specifically on Polymarket's most active markets and prefer a purpose-built voting approach for surfacing directional signals — PolyOracle is designed for that workflow and does not require integrating a broader cross-venue data stack.
PolyOracle runs multi-LLM consensus on Polymarket's top markets. SimpleFunctions ships the full agent layer: world model, causal theses, indicators, autopilot, MCP, cross-venue. Different scopes.
At a glance
Three things that
actually differ.
Everything PolyOracle gives you — LLM-derived consensus signals on Polymarket's most active markets — SimpleFunctions also gives you, on the same Polymarket feed and across Kalshi, covering 48K+ contracts.
On top of that, SF ships a causal-tree thesis system, an autonomous trading agent (Portfolio Autopilot, 1M-context LLM, 7-gate risk cascade), and 56 MCP tools that no current prediction market data product exposes.
SF also publishes live Brier scores for itself at /api/calibration — Kalshi 0.20, Polymarket 0.12 on T-24h price, past 90 days. PolyOracle does not publish calibration data for its consensus outputs.
Side by side
9 dimensions · verified 2026-04SimpleFunctionsKalshi and Polymarket normalised, 48K+ active contracts indexed across both venues.
PolyOraclePolymarket-only, focused on the most active markets on that venue.
SimpleFunctionsCausal thesis system — POST /api/thesis/create decomposes a sentence into a testable sub-claim tree, propagates probabilities, and runs a continuous evaluation heartbeat (news scan, price refresh, LLM eval, confidence update).
PolyOracleMultiple large language models run in parallel and vote to reach a consensus signal on selected markets.
SimpleFunctionsImplied yield, cliff risk index, liquidity availability score, event overround, τ-days, and regime label pre-computed across all 48K contracts at /screen.
PolyOracleConsensus signal output; no mention of derived indicators or pre-computed metrics.
SimpleFunctionsGET /api/public/market/{ticker}?depth=true returns full bid/ask ladder, spread, and slippage estimates.
PolyOracleNot documented in public-facing materials.
SimpleFunctionsLive Brier scores at /api/calibration — by venue, category, and price bucket, updated continuously.
PolyOracleAccuracy of consensus outputs is not published.
SimpleFunctionsCausal-tree decomposition with signal injection at /api/thesis/{id}/signal and public fork-ability; no competitor exposes this.
PolyOracleNot in scope — PolyOracle produces consensus signals, not decomposed causal trees.
SimpleFunctionsPortfolio Autopilot uses a 1M-context LLM and a 7-gate risk cascade (kill switch, position limits, drawdown gate, regime check) before any execution.
PolyOracleNot in scope.
SimpleFunctions56 tools via claude mcp add simplefunctions --url https://simplefunctions.dev/api/mcp/mcp; works with Claude Code, Cursor, and any MCP client.
PolyOracleNo MCP server published.
SimpleFunctionsPublic REST, MCP, and CLI reads require no authentication. Authenticated thesis and intent execution is free up to 15M tokens, then pay-per-token.
PolyOraclePricing structure not publicly stated in available documentation.
Methodology
Verified 2026-04 from public sources only — PolyOracle's documentation, public website, and publicly observable behaviour. We never claim non-public information about PolyOracle's internals. SimpleFunctions claims on this page are computed live from /api/calibration, /api/public/cross-venue/pairs, and /api/public/markets — you can re-verify them yourself with curl.
Use cases
Same data, different
best fit per scenario.
Scenario 01
Building an AI agent that needs cross-venue market data, indicators, and a structured world model.
SimpleFunctions · best fit
SF's /api/agent/world returns an ~800-token world snapshot covering both Kalshi and Polymarket. Computed indicators (implied yield, cliff risk, LAS) are pre-fetched so the agent does not need to derive them. The 56-tool MCP server exposes all of this natively to any MCP client.
PolyOracle
PolyOracle focuses on Polymarket-only consensus signals and does not publish a world-model endpoint or cross-venue data feed. Agents needing Kalshi coverage or structured indicator access would need to source that separately.
Scenario 02
Getting a fast directional read on Polymarket's most active markets using LLM voting consensus.
SimpleFunctions
SF provides LLM-driven thesis evaluation and indicators across both venues, but its design is oriented toward structured causal analysis rather than rapid-consensus voting on a ranked list of top markets.
PolyOracle · best fit
PolyOracle is purpose-built for this: multiple LLMs vote in parallel on the most active Polymarket markets and surface a consolidated directional signal. For teams that specifically want a lightweight consensus layer on the top-of-book markets, PolyOracle is the focused tool.
Scenario 03
Decomposing a complex macro or geopolitical claim into tradeable prediction-market positions.
SimpleFunctions · best fit
POST /api/thesis/create takes a plain-language sentence, decomposes it into a causal sub-claim tree, scans Kalshi and Polymarket for edges on each node, and runs a recurring evaluation heartbeat — news, price refresh, and LLM confidence update — indefinitely. Public theses are forkable.
PolyOracle
PolyOracle generates consensus signals on existing markets but does not decompose user-defined claims into causal trees or scan for cross-venue arbitrage edges between sub-claims.
Scenario 04
Auditing a prediction model's historical accuracy before relying on it for portfolio decisions.
SimpleFunctions · best fit
SF publishes live Brier scores at /api/calibration, broken down by venue, category, and price bucket — Kalshi 0.20, Polymarket 0.12 on T-24h price over the past 90 days. The endpoint is public and re-verifiable with curl.
PolyOracle
PolyOracle does not publish calibration or accuracy metrics for its consensus outputs, making external audit of its historical performance difficult.
Live data
The SimpleFunctions claims on this page are not marketing copy. Brier scores, market counts, and cross-venue pair counts are computed live from /calibration, /screen, and /api/public/cross-venue/pairs. All public, all free, all CC-BY-4.0.
FAQ
What does SimpleFunctions add beyond multi-LLM consensus?+
PolyOracle surfaces a voted directional signal on top Polymarket markets. SF builds a persistent causal model above those signals: a thesis system that decomposes any claim into a tree of testable sub-claims, propagates probabilities across them, scans both Kalshi and Polymarket for edges on each node, and runs a continuous evaluation heartbeat. On top of that sits Portfolio Autopilot (autonomous execution with a 7-gate risk cascade) and a 56-tool MCP server — none of which PolyOracle exposes.
How does SF's causal thesis system work?+
POST /api/thesis/create takes a plain-language sentence and decomposes it into a causal tree of testable sub-claims. SF scans Kalshi and Polymarket for contracts that correspond to each node, propagates probability estimates through the tree, and starts an evaluation heartbeat — periodic news scans, price refreshes, milestone checks, and LLM re-evaluations that update confidence scores over time. You can inject new signals at any point via /api/thesis/{id}/signal. Public theses are forkable by other users.
Does PolyOracle cover Kalshi markets?+
Based on publicly available information, PolyOracle focuses on Polymarket's most active markets. SimpleFunctions indexes both Kalshi and Polymarket — 48K+ active contracts across both venues — with normalised prices, computed indicators, and cross-venue matched pairs available at /api/public/cross-venue/pairs.
What is Portfolio Autopilot and how does it manage risk?+
Portfolio Autopilot is SF's autonomous trading agent. It uses a 1M-context LLM and 13 data sources — including live prices, calibration data, computed indicators, and active theses — to evaluate positions. Before any execution, a 7-gate risk cascade checks: kill switch status, position limits, drawdown gate, regime classification, liquidity availability, and additional risk parameters. Each gate must clear independently before a trade is submitted.
What is the SF MCP server and how do I connect it?+
SF ships a 56-tool MCP server that exposes its full API surface — market data, thesis management, orderbook depth, cross-venue pairs, trade ideas, calibration data, and more — to any MCP client. Connect it with one command: claude mcp add simplefunctions --url https://simplefunctions.dev/api/mcp/mcp. It works with Claude Code, Cursor, and any other client that implements the MCP spec. No authentication is required for read tools.
How does SF measure and publish its own calibration?+
SF computes Brier scores against its own price predictions by venue, market category, and price bucket, looking back 90 days. The results are published live at /api/calibration — currently Kalshi 0.20 and Polymarket 0.12 on T-24h price. The endpoint is public, requires no authentication, and is re-verifiable with a single curl call. Most prediction market data products do not publish comparable self-audit data.
Can I access SimpleFunctions data without signing up?+
Read access to SF's REST API, MCP server, and CLI requires no authentication. This includes market prices, orderbook depth, cross-venue pairs, computed indicators, calibration data, trade ideas, and the world snapshot. Authenticated access is required only for stateful operations like thesis creation, signal injection, and portfolio autopilot — free up to 15M tokens, then pay-per-token.
Start for free.
Public endpoints are free for normal usage and rate-limited for reliability. Authenticated endpoints are free up to 15M tokens, then pay per token. No credit card to start.