Live — 0 events tracked

Structured AI forecasts
on real-world events.

Four frontier models forecast macro, policy, earnings, and geopolitical events independently. Every prediction has fixed resolution criteria and gets verified against real outcomes.

0
Active Events
4
AI Models
0
Resolve 30d
0
High Spread
0%
Avg Consensus

Active Forecasts

UPDATED 2× DAILY

Each event is independently forecasted by GPT-4o, Claude, Gemini, and Grok. Resolution criteria fixed at creation.

TypeEventResolvesAIModels
Loading forecasts...
METHODOLOGY

Auditable pipeline, no black boxes

Every forecast follows the same structured process from event creation to verified outcome.

01
Detect
AI scans upcoming macro releases, policy decisions, earnings dates, and geopolitical events. Questions are structured with binary resolution criteria.
02
Forecast
GPT-4o, Claude, Gemini, and Grok independently predict each event. No model sees another's output. Consensus and spread are computed.
03
Track
Every prediction is timestamped and stored with model reasoning, confidence, and belief trajectories as resolution dates approach.
04
Verify
Outcomes verified against named sources — BLS, Fed, SEC filings, election authorities. Every prediction gets a correct/incorrect label.
WHY THIS DATA

Not a prediction market. A structured dataset.

Schema-consistent
Every event follows an identical structure: question, resolution criteria, authoritative source, date, asset linkage. Machine-readable from day one.
Multi-model consensus
Four competing frontier models forecast independently. Model disagreement is a first-class signal, not noise to be averaged away.
Longitudinal tracking
Belief trajectories captured over time create a unique dataset of how AI reasoning evolves. This data cannot be recreated retroactively.
APICOMING SOON

Programmatic access to every forecast

Model reasoning, belief trajectories, resolution outcomes. All structured, all versioned.

GET /api/v1/forecasts

{
  "events": [{
    "id": "CRENE-CPI-20260311",
    "question": "Will US CPI YoY for Feb 2026 exceed 2.4%?",
    "category": "macro_data",
    "resolution_date": "2026-03-11",
    "consensus": 60, "spread": 10,
    "models": {
      "gpt4o":  { "probability": 65, "reasoning": "..." },
      "claude": { "probability": 60, "reasoning": "..." },
      "gemini": { "probability": 55, "reasoning": "..." },
      "grok":   { "probability": 60, "reasoning": "..." }
    },
    "belief_trajectory": [
      { "date": "2026-02-26", "consensus": 60 },
      { "date": "2026-02-27", "consensus": 62 }
    ]
  }]
}
USE CASES

Built for people who need correct answers

Quantitative Researchers
Structured AI belief data as an orthogonal signal. Model disagreement as a feature. Full history for backtesting and calibration analysis.
Macro Analysts
Multi-model consensus on every major economic event. Pre-computed views across CPI, NFP, FOMC, GDP, earnings, and central bank decisions.
Data Teams
Clean, schema-consistent API feed. No scraping, no ambiguous resolution. Every field typed, every outcome verified against authoritative sources.

Structured intelligence
on what happens next.

Explore AI forecasts across macro, earnings, policy, and geopolitical events. Contact us for institutional API access.

Crene — AI Forecasting Benchmark