SimpleFunctions

Alternative · Forecasting platform

Metaculus vs
SimpleFunctions.

Metaculus aggregates community superforecaster consensus on long-horizon questions — AI timelines, biosecurity, nuclear risk, geopolitics — through a public REST API and an open-source Python toolkit. SimpleFunctions operates a different layer: a causal-tree thesis system that decomposes any claim into tradeable sub-claims mapped to live exchange contracts, an autonomous Portfolio Autopilot with a 7-gate risk cascade, computed indicators across 48K+ real-money contracts on Kalshi and Polymarket, and a 56-tool MCP server. The two platforms surface probability estimates on overlapping topics; where they diverge is execution surface, market microstructure access, and agent tooling.

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 calibrated real-money probabilities alongside a full agent toolchain. SF ships a causal-tree thesis system with auto-evaluation cycles, a Portfolio Autopilot with a 1M-context LLM and 7-gate risk cascade, computed indicators across 48K+ live Kalshi and Polymarket contracts, and a 56-tool MCP server that installs in Claude Code or Cursor in one line. Live Brier scores at /api/calibration let you verify SF's own accuracy by venue, category, and price bucket.

Choose Metaculus if

Metaculus is the right choice when you need community superforecaster consensus on long-horizon questions — AI development timelines, biosecurity scenarios, nuclear risk — that have not yet reached a prediction market, or when financial stakes would distort the signal. Their AI Forecasting Bot Tournament also provides a structured environment for benchmarking AI forecasting models against calibrated human baselines.

Metaculus aggregates community forecasts on play-money long-horizon questions. SimpleFunctions covers real-money Kalshi and Polymarket with a full agent layer above.

At a glance

Three things that
actually differ.

01

Everything Metaculus gives you — probability estimates on world events, a public REST API, and research-grade forecasting on AI timelines and geopolitics — SimpleFunctions also gives you, on real-money Kalshi and Polymarket feeds covering the same topic categories.

02

On top of that, SF ships a causal-tree thesis system with auto-evaluation cycles, an autonomous Portfolio Autopilot (1M-context LLM, 7-gate risk cascade), computed indicators across 48K+ contracts, and 56 MCP tools that no community forecasting platform exposes.

03

SF publishes live Brier scores for itself at /api/calibration — Kalshi 0.20, Polymarket 0.12 on T-24h price, past 90 days — broken down by venue, category, and price bucket. You can re-verify them with a single curl call.

Side by side

10 dimensions · verified 2026-04
Data source

SimpleFunctionsReal-money Kalshi and Polymarket feed, 48K+ active contracts, normalised prices updated live.

MetaculusCommunity superforecaster consensus; questions exist independently of any exchange and carry no financial stakes.

Orderbook depth

SimpleFunctionsFull bid/ask ladder at /api/public/market/{ticker}?depth=true — spread, slippage estimate, and depth visible.

MetaculusNo orderbook; community probability is an aggregate of individual forecaster submissions, not an exchange-listed contract.

Topic coverage

SimpleFunctionsAny question with a Kalshi or Polymarket listing, updated live as prices move.

MetaculusLong-horizon questions on AI timelines, biosecurity, nuclear risk, and geopolitics — including topics not yet on any exchange.

Computed indicators

SimpleFunctionsIY, CRI, LAS, EE, τ-days, and regime labels pre-computed across 48K+ contracts and surfaced at /screen.

MetaculusResolution history and community calibration curves available; no market microstructure indicators.

Calibration data

SimpleFunctionsLive Brier scores at /api/calibration by venue, category, and price bucket — machine-readable and curl-verifiable.

MetaculusForecaster-level calibration tracked over resolved questions; aggregate venue-level breakdown not published in verifiable public sources.

Causal thesis system

SimpleFunctionsPOST /api/thesis/create decomposes any sentence into a causal sub-claim tree, scans Kalshi and Polymarket for tradeable edges, and runs an auto-evaluation heartbeat.

MetaculusQuestions are authored and submitted by community members; no automated causal decomposition or market-scanning pipeline.

Autonomous trading

SimpleFunctionsPortfolio Autopilot — 1M-context LLM, 13 data sources, 7-gate risk cascade before any real-money order touches an exchange.

MetaculusAI Forecasting Bot Tournament benchmarks AI forecasters against human superforecasters in a play-money environment; no real-money execution.

MCP server

SimpleFunctions56 tools via claude mcp add simplefunctions --url https://simplefunctions.dev/api/mcp/mcp; works with Claude Code, Cursor, and any MCP client.

MetaculusNo MCP server documented publicly.

Open-source tooling

SimpleFunctionsMIT CLI with 60+ commands, installable via npm i -g @spfunctions/cli.

MetaculusOpen-source forecasting-tools Python repository on GitHub.

Pricing

SimpleFunctionsPublic REST, MCP, and CLI reads require no auth; pay-per-token only on thesis creation and intent execution, free up to 15M tokens.

MetaculusCommunity participation is free; API tier details are not listed in publicly verifiable sources we can confirm.

Methodology

Verified 2026-04 from public sources only — Metaculus's documentation, public website, and publicly observable behaviour. We never claim non-public information about Metaculus'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 live real-money prediction market data with computed risk indicators.

SimpleFunctions · best fit

SF exposes the full Kalshi and Polymarket feed with pre-computed indicators — implied yield, cliff risk index, liquidity availability score — plus cross-venue arbitrage pairs, an 800-token agent world snapshot at /api/agent/world, and 56 MCP tools installable in one line. Portfolio Autopilot can autonomously manage positions behind a 7-gate risk cascade.

Metaculus

Metaculus has no connection to Kalshi or Polymarket and aggregates community forecasts, not exchange-traded prices. It is not designed for agents that need to query live market microstructure or execute real-money trades.

Scenario 02

Researching long-horizon AI timeline or biosecurity scenarios not yet listed on any prediction exchange.

SimpleFunctions

SF covers questions that have a corresponding Kalshi or Polymarket contract, updated live. For AI timeline questions without a market listing, SF does not have coverage.

Metaculus · best fit

Metaculus specializes in exactly this: long-horizon questions on AI development, biosecurity, nuclear risk, and geopolitics, with a community of trained superforecasters submitting calibrated estimates. This is Metaculus's core product and the scenario where it outperforms any exchange-data platform.

Scenario 03

Decomposing a geopolitical thesis into tradeable causal sub-claims and tracking them automatically.

SimpleFunctions · best fit

SF's thesis system (POST /api/thesis/create) decomposes any sentence into a causal tree, maps each leaf to live Kalshi and Polymarket contracts, and runs an evaluation heartbeat that scans news, refreshes prices, and updates confidence scores. Public theses are forkable; external signals can be injected via /api/thesis/{id}/signal.

Metaculus

Metaculus allows users to author questions and track community forecasts, but the process is manual and crowd-driven. There is no automated causal decomposition or market-scanning pipeline.

Scenario 04

Benchmarking an AI forecasting model's accuracy against human superforecaster baselines.

SimpleFunctions

SF's live Brier scores at /api/calibration provide a baseline for how well real-money market prices predict outcomes, by venue and category, over the past 90 days. You can compare a model's Brier score against SF's published numbers on the same resolution events.

Metaculus · best fit

Metaculus's AI Forecasting Bot Tournament is purpose-built for this use case: a structured competition environment with human superforecaster baselines, diverse question sets, and a leaderboard. For head-to-head benchmarking of forecasting models against human crowds, Metaculus's tournament infrastructure is the more direct fit.

Migrate

From https://www.metaculus.com/api2/questions/ to SF.

Same shape, no auth, same venues. Python example.

Metaculus
import requests

resp = requests.get(
    "https://www.metaculus.com/api2/questions/",
    params={
        "search": "artificial intelligence",
        "status": "open",
        "forecast_type": "binary",
        "limit": 20
    },
    headers={"Accept": "application/json"}
)
for q in resp.json()["results"]:
    print(q["title"], q["community_prediction"]["full"]["q2"])
SimpleFunctions
import requests

resp = requests.get(
    "https://simplefunctions.dev/api/public/scan",
    params={"q": "artificial intelligence", "limit": 20}
)
for m in resp.json()["markets"]:
    indicators = m.get("indicators", {})
    print(
        m["title"],
        m["price"],
        indicators.get("cri"),
        indicators.get("las")
    )

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 is SimpleFunctions' thesis system, and how does it compare to Metaculus community questions?+

SF's thesis system (POST /api/thesis/create) takes any declarative sentence and decomposes it into a causal tree of testable sub-claims. Each leaf is matched to real-money contracts on Kalshi and Polymarket, and an evaluation heartbeat continuously refreshes prices, scans news, and updates confidence scores. You can inject external signals via /api/thesis/{id}/signal. Metaculus works differently: community members author questions, forecasters submit probability estimates, and the platform aggregates them. Metaculus excels at surfacing human consensus on topics without a market; SF's thesis system is designed for automated, agent-driven causal reasoning over live financial data.

Does SimpleFunctions have autonomous trading where Metaculus has none?+

Yes. SF's Portfolio Autopilot is a 1M-context LLM with 13 data sources and a 7-gate risk cascade — kill switch, position limits, drawdown gate, regime check, and others — that must all clear before any order reaches an exchange. Metaculus has an AI Forecasting Bot Tournament, which benchmarks AI forecasters against human superforecasters in a play-money environment; there is no real-money execution. SF's Autopilot is designed for live Kalshi and Polymarket positions, not model benchmarking.

Can SimpleFunctions cover AI timelines, biosecurity, and geopolitics the same way Metaculus does?+

Partially. Both platforms surface probability estimates on these topics, but the source differs. SF queries live Kalshi and Polymarket contracts — real-money markets on AI-related elections, policy events, and macro outcomes. Metaculus aggregates community forecasters on a broader set of questions, including multi-year AI timelines and biosecurity scenarios that may not yet have corresponding exchange-listed contracts. If the question has a market on Kalshi or Polymarket, SF covers it with added indicators; if it does not, Metaculus is more likely to have community coverage.

Does Metaculus have an MCP server for AI agent integration?+

Metaculus publishes a public OpenAPI spec at metaculus.com/api/ and an open-source Python forecasting-tools repository, but no MCP server is documented publicly. SF ships 56 tools via a single MCP endpoint — install with claude mcp add simplefunctions --url https://simplefunctions.dev/api/mcp/mcp — giving Claude Code, Cursor, or any MCP-compatible client direct access to markets, thesis creation, computed indicators, cross-venue pairs, and portfolio data. No additional SDK or auth is required for read operations.

How does calibration compare between SimpleFunctions and Metaculus?+

SF publishes its own Brier scores at /api/calibration, broken down by venue (Kalshi 0.20, Polymarket 0.12 on T-24h price, past 90 days), category, and price bucket. This is machine-readable and re-verifiable with a single curl call. Metaculus tracks calibration at the forecaster level over resolved questions, which is valuable for evaluating individual and aggregate human forecasters. The two platforms are measuring different things: SF audits its market-price accuracy on short-horizon real-money questions; Metaculus audits human forecaster accuracy on a broader, longer-horizon question set.

Which platform is better for building an AI forecasting agent?+

SF is purpose-built for agents: a 56-tool MCP server, a /.well-known/ai-world-state spec, llms.txt, a structured world snapshot at /api/agent/world, and a thesis system that automates the research-to-trade pipeline. Metaculus offers an AI Forecasting Bot Tournament and a public REST API, making it a reasonable testbed for evaluating a forecasting model against human superforecaster baselines. If your agent needs to act on prediction markets — place orders, manage positions, track PnL — SF is the purpose-built surface. If you want to benchmark forecast accuracy against human crowds, Metaculus provides a structured arena for that.

Does SimpleFunctions provide orderbook data that Metaculus lacks?+

Yes. SF exposes full bid/ask ladders at /api/public/market/{ticker}?depth=true for any Kalshi or Polymarket contract. Metaculus is a community platform with no market microstructure — individual forecasters submit point probabilities and the platform aggregates them into a consensus. There is no spread, no depth, and no liquidity score in Metaculus data. SF also pre-computes a Liquidity Availability Score (LAS), event overround (EE), and cliff risk index (CRI) across all 48K+ active contracts, surfaced at /screen.

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.