PROPHET vs
SimpleFunctions.

PROPHET is an expert-curated prediction newsletter covering politics, finance, and geopolitics. Its audience is readers who want synthesized, data-driven analysis from domain experts delivered in a readable format — researchers, executives, or analysts who prefer editorial judgement in prose form over raw market data or programmatic access. It does not require any technical setup and is not designed for API consumers, agent builders, or systematic traders.

POST /api/thesis/create accepts any plain-language sentence. SF decomposes it into a causal tree of independently testable sub-claims, maps each claim to live Kalshi and Polymarket contracts, and begins a continuous evaluation heartbeat: news scanning, price refresh, milestone checks, LLM confidence scoring, and probability propagation back up the tree. You can inject new signals at any point via /api/thesis/{id}/signal. Public theses are forkable, so other users can branch and extend your reasoning. No competing product in the prediction market space has an equivalent system.

Based on publicly observable information, PROPHET operates as a newsletter product with no documented REST or HTTP API. There is no published CLI or MCP server. If your workflow requires consuming prediction signals programmatically — in an agent, a trading bot, a research pipeline, or a Claude tool — SimpleFunctions exposes all of this via REST, a 60-command CLI, and a 56-tool MCP server, with no authentication required for reads.

SF's /api/calibration endpoint returns Brier scores computed from SF's own past predictions, segmented by venue (Kalshi, Polymarket), event category, and price bucket. Current figures: Kalshi 0.20, Polymarket 0.12 on T-24h price over the past 90 days. Brier score is a proper scoring rule — lower is better, 0.25 is random. Publishing this publicly means any user can re-verify SF's accuracy with a single curl call, rather than relying on marketing claims. Most prediction data providers do not publish structured accuracy audits.

Portfolio Autopilot is SF's autonomous trading agent. It runs a 1M-context LLM across 13 data sources and gates every trade through a 7-step risk cascade — including a kill switch, position limits, drawdown gate, and regime classification — before any execution on Kalshi or Polymarket. This layer does not exist in newsletter products. PROPHET provides analysis for human readers to act on manually; SF's Autopilot closes the loop from research signal to executed position without requiring human intervention at each step.

It depends on what the research workflow looks like. If the goal is reading synthesized expert analysis in prose — understanding context, narrative, and editorial interpretation of geopolitical events — PROPHET is the right surface. If the goal is querying live market-implied probabilities, decomposing hypotheses into testable sub-claims, backtesting against calibration benchmarks, or feeding signals into an automated agent, SF is the right surface. The two products occupy different layers and are not direct substitutes.

SF is designed primarily for developers, researchers, and agent builders who want programmatic access. The /screen page exposes computed indicators in a browser-readable format, and the MCP server makes tools available inside Claude Code or Cursor without writing application code. However, SF's core value — the thesis system, autopilot, cross-venue arbitrage pairs, and calibration data — is accessed via API, CLI, or MCP. Users who want prediction content without any technical interaction are better served by a newsletter product like PROPHET.

SF indexes 48K+ active contracts from Kalshi and Polymarket, which include extensive coverage of US and international politics, geopolitical events, financial markets, and economic indicators. The coverage is market-driven — contracts exist where liquidity and trading interest exist. SF does not produce editorial analysis or expert commentary; it surfaces live probability estimates, calibrated signals, and computed market indicators. The thematic overlap with PROPHET is substantial, but the form of delivery and the intended workflow are different.