Prediction Intelligence

Structured forecasts.
Tracked over time.

Four AI models forecast 10,000+ prediction markets every 4 hours. Each prediction is taxonomized, tracked through belief shifts, and scored against real outcomes.

0%
High Confidenceaccuracy
0%High
0%Medium
0%Low
0Resolved
10,000+MARKETS
—RESOLVED
4AI MODELS
—BRIER
3MODES
10,000+MARKETS
—RESOLVED
4AI MODELS
—BRIER
3MODES
By the numbers

Intelligence at scale

10K+
Active markets
Polymarket + Kalshi
—
Resolved outcomes
Brier scored
—
High-confidence acc.
Divergence signal
3
Prediction modes
Informed · Blind · Augmented
Platform

Four layers of prediction intelligence

Every prediction goes through a pipeline: forecast, classify, track, and resolve. The result is structured data that gets smarter over time.

Performance

Accuracy by domain

Every forecast resolved against verified outcomes. Category-level accuracy with full Brier scoring.

Loading calibration data…
AI Alpha vs Market
Data Structure

Every prediction is classified

Structured taxonomy on every forecast — domain, entities, time horizon, and resolution deadline. Combined with belief trajectory tracking, you can see exactly how intelligence evolved over time.

// taxonomy + lifecycle
{
  "domain":     "fed_rates",
  "entities":   ["Fed", "Powell"],
  "time_horizon": "months",
  "deadline":   "2026-03-19",
  "specificity":  5,
  "trajectory": [
    { t: "Jan 15", trigger: "initial" },
    { t: "Feb 12", trigger: "prob_shift",
      changes: "speculative → likely" },
    { t: "Mar 19", trigger: "resolution",
      outcome: "correct", ai_error: 8 }
  ]
}
domainbtc_price, fed_rates, ceasefire
Specific forecast sub-domain
time_horizondays / weeks / months / years
Resolution timeframe
specificity1–5
How precise the forecast question is
entitiesSEC, BlackRock, Putin
Companies, people, orgs referenced
resolution_deadline2026-06-01
Expected outcome date
trajectory[{t, trigger, changes}]
Belief evolution over time
Use cases

Built for teams that need structured forecasting data

Quantitative Research

Structured prediction data with full taxonomy, belief trajectories, and resolution outcomes. Filter by domain, time horizon, and confidence — designed for systematic analysis.

Lifecycle APITaxonomy filteringCSV / JSON export

AI Labs & Eval Teams

Benchmark model forecasting ability with per-model Brier scores, calibration curves, and blind vs informed comparison. The largest public LLM forecasting dataset.

Per-model BrierBlind vs informedCalibration data

Intelligence & Strategy

Track how AI consensus shifts on geopolitical events, policy outcomes, and market-moving questions. Confidence signals flag when models strongly disagree with markets.

Divergence signalsConfidence tiersEntity tracking

Data Science & ML

Clean, structured forecasting data with standardized schema. Full provenance, reproducible methodology, and API access for integration into pipelines.

REST APIStructured schema10K+ markets
API

Prediction intelligence via API

Forecasts, taxonomy, lifecycle data, and evaluation metrics. Public endpoints for exploration, authenticated for bulk export. Free to start.

REST APIJSON + CSVFree tierLifecycle dataTaxonomy
API reference 
$ curl api-get.crene.com/api/predictions/lifecycle/
  kalshi_KXFEDRATE-26MAR19/

{
  "question": "Fed cuts rates by March?",
  "current": {
    "ai_consensus": 45,
    "market":       38,
    "confidence":  "HIGH"
  },
  "taxonomy": {
    "domain": "fed_rates",
    "entities": ["Fed", "Powell"]
  },
  "trajectory": [3 entries],
  "modes": {
    "informed": 45, "blind": 52, "augmented": 41
  }
}

Prediction intelligence
for serious teams.

Explore the data, or get in touch about research collaboration and enterprise access.

Crene — AI Forecasting Benchmark