Datasets:

SimpleFunctions
/

world-state-daily

license: mit
task_categories:
  - question-answering
  - text-generation
language:
  - en
tags:
  - prediction-markets
  - world-model
  - real-time-data
  - ai-agents
  - geopolitics
  - economics
  - calibrated-probabilities
  - kalshi
  - polymarket
size_categories:
  - n<1K
pretty_name: World State Daily Snapshots

World State Daily Snapshots

Daily snapshots of the world's state as measured by prediction markets. Each snapshot contains calibrated probabilities for geopolitics, economy, energy, elections, crypto, and tech — derived from 9,706 contracts on Kalshi (CFTC-regulated) and Polymarket.

Why This Dataset Exists

LLMs have a knowledge cutoff. This dataset provides ground truth for what the world looked like on any given day — not from news (narratives) or surveys (opinions), but from prediction markets where participants risk real money. Get it wrong, lose money.

Use cases:

Agent evaluation: Test whether your agent can accurately answer "what's the recession probability?" on a given date
Training data: Fine-tune models with calibrated world state data
Research: Analyze how prediction market probabilities evolve over time
Backtesting: Validate agent decisions against historical world state

Schema

Each daily JSON file contains:

{
  "date": "2026-04-02",
  "timestamp": "2026-04-02T20:56:00Z",
  "index": {
    "uncertainty": 22,
    "geopolitical": 62,
    "momentum": 0.06
  },
  "traditional": [
    {"symbol": "SPY", "price": 655.13, "changePct": 0.81}
  ],
  "topics": [
    {
      "name": "Geopolitical",
      "movers": [
        {"title": "Iran invasion probability", "price": 53, "delta": 5, "volume": 225000, "venue": "kalshi"}
      ]
    }
  ],
  "topEdges": [
    {"title": "Market X", "edge": 15, "direction": "yes", "price": 35}
  ],
  "divergences": [
    {"description": "Stocks and gold both up — unusual risk-on + haven bid"}
  ],
  "markdown": "# World State — ...",
  "markdownTokenEstimate": 667
}

Fields

Field	Description
`index.uncertainty`	Market uncertainty (0-100), derived from orderbook spreads
`index.geopolitical`	Geopolitical risk (0-100), from geo-related market velocity
`index.momentum`	Directional market bias (-1 to +1)
`traditional`	SPY, VIX, TLT, GLD, USO prices and daily change
`topics`	6 categories with anchor contracts and significant movers
`topEdges`	Largest mispricings detected by thesis models
`divergences`	Cross-market anomalies (e.g., stocks and gold both rising)
`markdown`	Raw markdown output (~800 tokens, ready for LLM injection)

Data Source

SimpleFunctions — aggregates 9,706 prediction market contracts from Kalshi (CFTC-regulated) and Polymarket. Updated every 15 minutes. The daily snapshot captures the final state of each day.

Live API

For real-time data (not just daily snapshots):

pip install simplefunctions-ai

from simplefunctions import world, delta
print(world())           # current state (~800 tokens)
print(delta(since="1h")) # what changed (~30-50 tokens)

Citation

@dataset{simplefunctions_world_state_2026,
  title={World State Daily Snapshots: Calibrated Probabilities from Prediction Markets},
  author={SimpleFunctions},
  year={2026},
  url={https://huggingface.co/datasets/SimpleFunctions/world-state-daily},
  note={Daily snapshots from 9,706 prediction markets on Kalshi and Polymarket}
}

License

MIT