Metaforecast

📋Page Status

Page Type:ResponseStyle Guide →Intervention/response page

Quality:35 (Draft)⚠️

Importance:15 (Peripheral)

Last edited:2026-01-29 (3 days ago)

Words:1.6k

Structure:

📊 13📈 0🔗 4📚 10•15%Score: 13/15

LLM Summary:Metaforecast is a forecast aggregation platform combining 2,100+ questions from 10+ sources (Metaculus, Manifold, Polymarket, etc.) with daily updates via automated scraping. Created by QURI, it provides unified search and GraphQL API access but lacks historical data and is limited to passive aggregation rather than enabling new forecasting capabilities.

Issues (1):

QualityRated 35 but structure suggests 87 (underrated by 52 points)

Quick Assessment

Dimension	Assessment	Evidence
Coverage	Comprehensive	2,100+ active questions from 10+ platforms
Data Freshness	Daily	Automated scraping pipeline
Accessibility	High	Free web interface, GraphQL API
Integration	Active	Twitter bot, Discord bot (Fletcher), GlobalGuessing
Open Source	Fully	GitHub repository, part of Squiggle monorepo
Maintenance	Active	Ongoing updates, platform additions

Project Details

Attribute	Details
Name	Metaforecast
Organization	QURI (Quantified Uncertainty Research Institute)
Creator	Nuño Sempere (with help from Ozzie Gooen)
Launched	2021
Website	metaforecast.org
GitHub	github.com/quantified-uncertainty/metaforecast
License	Open source (part of Squiggle monorepo)
API	GraphQL endpoint for programmatic access

Overview

Metaforecast aggregates forecasts from multiple prediction platforms into a single searchable interface, solving the platform fragmentation problem that reduces forecasting’s utility as a public good. While individual platforms like Metaculus, Manifold, and Polymarket each host valuable forecasts, researchers and decision-makers must check multiple sites to find relevant predictions. Metaforecast eliminates this friction by combining data from 10+ platforms with unified search.

The project was initiated by Nuño Sempere, with help from Ozzie Gooen, at QURI. Data is fetched daily via automated scraping, showing immediate forecasts without historical time-series data. This design is optimized for quick lookups (“What’s the current probability of X?”) rather than trend analysis.

Design Philosophy

Metaforecast treats forecasting as a public good that becomes more valuable with aggregation. The platform is deliberately simple: search for a question, see current probabilities from multiple sources, compare platforms side-by-side. No user accounts, no social features, no complexity—just forecasts.

Platforms Aggregated

Metaforecast currently indexes approximately 2,100 active forecasting questions across these platforms:

Platform	Type	Questions	Notes
Metaculus	Reputation-based	≈1,200 (55% of total)	Largest source; research-focused, community prediction aggregation
Manifold	Play money market	Varies	Research-focused, no profit motive, uses Mana currency
Polymarket	Real money (crypto)	Varies	Rose to prominence in 2024 US election, real financial stakes
Good Judgment Open	Superforecaster platform	≈100-200	Connected to Good Judgment Project, expert forecasters
PredictIt	Political market	≈50-100	US political focus, regulated real-money trading
Kalshi	CFTC-regulated market	≈100-200	Regulatory compliance focus, legal real-money markets
Insight Prediction	Prediction market	Varies	Additional market source
INFER	Policy forecasting	Varies	Government-adjacent forecasting

Additionally, Metaforecast provides access to 17,000+ public models from Guesstimate, Ozzie Gooen’s earlier Monte Carlo spreadsheet tool.

Platform Characteristics

Platform	Incentive Structure	Typical User Base
Metaculus	Reputation points, tournament prizes	AI researchers, academics, rationalists
Manifold	Play money (Mana)	EA/rationalist community, hobbyists
Polymarket	Real money (cryptocurrency)	Traders, serious forecasters, arbitrageurs
Good Judgment	Reputation, selective admission	Professional superforecasters
Kalshi	Real money (USD), CFTC-regulated	Finance professionals, serious traders

Technical Architecture

Data Pipeline

Component	Description
Scraping	Daily automated fetching from each platform’s API or web interface
Storage	Centralized database with normalized question format
Search	Elasticsearch-powered full-text search
API	GraphQL endpoint for programmatic access
UI	Simple web interface with platform comparison

Data Freshness

Metaforecast fetches data daily, ensuring forecasts are current. However, the system is optimized for point-in-time forecasts rather than historical trend analysis:

Shows: Current probability estimates from each platform
Does not show: Historical time-series or forecast evolution over time

This design choice reduces storage requirements and simplifies the interface at the cost of trend analysis capabilities.

Features

Unified Search

Search across all platforms simultaneously:

Query: "AGI by 2030"
Results: Shows all matching questions from Metaculus, Manifold, Polymarket, etc.

Platform Comparison

See how different sources estimate the same event:

Question	Metaculus	Manifold	Polymarket
AGI by 2030	45%	60%	—
Nuclear war by 2050	9%	12%	—

This comparison helps identify consensus vs. divergence in forecasting communities.

GraphQL API

Programmatic access for integrations:

query {
  questions(limit: 10, query: "AI capabilities") {
    title
    platform
    probability
    url
  }
}

The API enables:

Custom analysis and visualization
Integration with research workflows
Automated monitoring of specific questions

Open Source

The GitHub repository is part of QURI’s Squiggle monorepo, allowing community contributions and transparency in the scraping methodology.

Integrations

Metaforecast has been integrated with several external services:

Service	Description	Status
Twitter Bot	Posts notable forecasts	Active
Fletcher	Discord bot for forecast lookups	Active
GlobalGuessing	Forecasting community platform	Active
Elicit	AI research assistant	Previously integrated

These integrations make forecast data accessible in contexts where users already work.

Use Cases

Research

Researchers use Metaforecast to:

Find consensus forecasts on emerging technologies
Compare expert vs. market-based predictions
Track how different communities view the same question
Validate assumptions in models with community forecasts

Decision-Making

Organizations reference Metaforecast for:

Strategic planning timeline assumptions
Risk assessment probability inputs
Validation of internal forecasts against external views

Forecasting Meta-Analysis

Researchers studying forecasting accuracy use Metaforecast to:

Compare platform performance
Identify systematic biases
Analyze calibration across question types

Notable Use: 2024 US Election

During the 2024 US presidential election, Polymarket (one of Metaforecast’s sources) demonstrated the value of prediction market aggregation. Polymarket strongly favored Trump’s victory even as traditional polls showed a closer race. The market’s prediction proved correct, with reports of approximately $10 million in bets from individual traders helping shape the odds.

Metaforecast allowed researchers to compare Polymarket’s market-based odds against Metaculus’s reputation-based aggregation and other sources, providing multiple perspectives on the same event.

Comparison with Individual Platforms

Aspect	Metaforecast	Individual Platform
Coverage	10+ platforms, 2,100+ questions	Single platform, typically 100-1,000 questions
Search	Cross-platform unified search	Platform-specific search
Historical Data	None (current forecasts only)	Often available
Community	Read-only aggregation	Active participation, social features
Incentives	None (passive aggregation)	Reputation points, money, social status
Effort Required	Single search	Must check multiple sites

Tradeoffs

Metaforecast gains:

Convenience: Single search across all platforms
Comparison: See multiple sources side-by-side
Completeness: Less likely to miss relevant forecasts

Individual platforms gain:

Historical data: Track forecast evolution
Community context: Discussion, reasoning transparency
Participation: Ability to contribute forecasts

Strengths and Limitations

Strengths

Strength	Evidence
Solves fragmentation	Aggregates 10+ platforms into single interface
Daily updates	Fresh data from automated scraping
Free and open	No paywall, open-source code
API access	Enables custom integrations and analysis
Simple interface	No learning curve, immediate utility
Broad coverage	2,100+ active questions plus 17,000+ Guesstimate models

Limitations

Limitation	Impact
No historical data	Cannot track forecast evolution or analyze updating patterns
Scraping-dependent	Breaks if source platforms change their structure
Heterogeneous questions	Different platforms use different operationalizations for similar events
No participation	Passive aggregation; users cannot contribute forecasts
Limited metadata	Reasoning, comments, and community context not aggregated
Maintenance burden	Each new platform requires custom scraper implementation

Relationship to QURI Ecosystem

Metaforecast complements QURI’s other tools:

Tool	Purpose	Relationship to Metaforecast
Squiggle	Probabilistic modeling	Metaforecast provides probability inputs for Squiggle models
Squiggle Hub	Model sharing platform	Can link to relevant Metaforecast questions as external validation
SquiggleAI	LLM model generation	Could use Metaforecast data to ground AI-generated probability estimates
Guesstimate	Monte Carlo spreadsheets	Metaforecast hosts 17,000+ Guesstimate models

Funding

As a QURI project, Metaforecast is funded through QURI’s grants:

Source	Amount	Period
Survival and Flourishing Fund	$150K+ to QURI	2019-2022
Future Fund	$100K to QURI	2022
Long-Term Future Fund	Ongoing to QURI	2023-present

Metaforecast represents a small portion of QURI’s overall budget but provides a valuable public good for the forecasting community.

Future Directions

Potential enhancements based on community feedback:

Enhancement	Benefit	Challenge
Historical data	Trend analysis, forecast tracking	Storage costs, backfill complexity
Reasoning aggregation	Expose why forecasters believe what they do	Scraping complexity, heterogeneous formats
Custom question creation	User-defined aggregations	Moderation, quality control
Alert system	Notify when specific forecasts update	User accounts, notification infrastructure
Calibration tracking	Show platform accuracy records	Requires resolution data, complex analysis