Metaculus: Platform Profile and Performance Data

Metaculus is a reputation-based forecasting platform where a community of over 20,000 forecasters submit probability estimates on structured questions. Founded in 2015 and based in Berkeley, California, it operates without real-money trading, distinguishing it from exchange-based platforms like Polymarket and Kalshi.

What Is Metaculus?

Metaculus is a community forecasting platform, not a financial exchange. Users do not buy or sell contracts. Instead, they submit probability distributions on open questions — anything from “Will the WHO declare a new pandemic by 2027?” to “What will US GDP growth be in Q3 2026?” The platform aggregates these individual forecasts into a community median prediction.

This design reflects a fundamentally different approach to prediction aggregation. On Polymarket or Kalshi, prices emerge from traders risking capital. On Metaculus, probabilities emerge from forecasters risking reputation. Both methods produce useful probability signals, but the incentive structures and participation barriers differ substantially.

Metaculus was founded in 2015 by Anthony Aguirre and Greg Laughlin, both physicists at UC Santa Cruz. The platform grew out of academic interest in collective intelligence and probabilistic reasoning. It has since expanded beyond academia into policy forecasting, pandemic preparedness, and AI safety research.

The platform hosts two distinct products. The public forecasting platform is free and open to anyone. Metaculus Pro is a commercial offering where organizations commission private forecasting tournaments staffed by vetted, high-accuracy forecasters.

How Does Metaculus Work?

A Metaculus question has a defined resolution criteria, an open date, a close date, and a resolution date. Forecasters submit their probability estimates at any point while the question is open. They can update their forecasts as new information becomes available.

The platform supports two question types:

• Binary questions — Will event X happen? Forecasters submit a probability between 1% and 99%.
• Continuous questions — What will the value of X be? Forecasters submit a probability distribution over a defined range, specifying their median, 25th percentile, and 75th percentile estimates.

Continuous questions are a major differentiator. Platforms like Polymarket and Kalshi trade binary contracts (yes/no at a fixed price). Metaculus captures richer information by asking forecasters to express full distributions over numeric outcomes. This is particularly valuable for economic indicators, scientific measurements, and timeline estimates.

The community prediction is the median of all individual forecasts, weighted by the platform’s internal model. Metaculus also publishes a “Metaculus prediction” on some questions, which applies additional algorithmic weighting based on forecaster track records.

Scoring uses a log-based system. Forecasters earn points proportional to the log score of their prediction relative to the community baseline. Early forecasts receive a time multiplier, rewarding those who commit to a position before the consensus forms. The scoring is strictly proper — the mathematically optimal strategy is to report your true beliefs.

What Is the Metaculus Fee Structure?

Metaculus charges no fees because there is no money to move. This is the simplest fee structure of any platform we track.

The absence of financial friction means Metaculus attracts a different user base than exchange platforms. Participants are motivated by intellectual challenge, reputation, and the desire to contribute to better collective forecasting — not profit. This selection effect has implications for accuracy, which we discuss below.

For a detailed fee breakdown across all major platforms, see our platform comparison guide.

What Markets Does Metaculus Cover?

Metaculus refers to its listings as “questions” rather than “markets,” reflecting the non-financial framing. The platform covers a broad range of categories:

• Geopolitics and policy — Elections, international conflicts, diplomatic agreements, regulatory decisions. Metaculus hosted extensive question sets on the Russia-Ukraine conflict and US-China relations.
• Science and technology — AI capability milestones, climate benchmarks, space exploration timelines, biomedical research outcomes.
• Economics — GDP growth, inflation readings, unemployment, Fed policy. These overlap with Kalshi’s strongest categories but are expressed as probability distributions rather than binary contracts.
• Pandemic and biosecurity — Metaculus gained significant recognition during COVID-19 for its pandemic forecasting questions, many of which outperformed institutional projections.
• AI and existential risk — A substantial portion of Metaculus questions focus on long-range AI development timelines and global catastrophic risks, reflecting the platform’s roots in the effective altruism and rationalist communities.

Our dataset tracks question coverage across platforms. Metaculus has the deepest coverage on long-range and scientific questions — categories where real-money markets have limited liquidity because resolution dates are years away. Exchange platforms concentrate on near-term events where capital can be deployed and returned quickly.

Metaculus Journals, a built-in long-form analysis feature, add another dimension. Forecasters publish detailed rationales alongside their predictions, creating an analytical record that does not exist on exchange platforms. This makes Metaculus particularly useful as a research tool, not just a probability source.

Is Metaculus Accurate?

Metaculus has one of the strongest public calibration records of any forecasting platform. The platform publishes its own calibration data, and independent analyses have generally confirmed the results.

Calibration measures whether a platform’s stated probabilities match observed frequencies. If a platform assigns 80% probability to a class of events, roughly 80 out of 100 of those events should occur. Metaculus community predictions show tight calibration across the probability spectrum, with slight overconfidence in the 85-95% range — a common pattern across forecasting systems.

Our analysis of cross-platform coverage shows that on questions where both Metaculus and a real-money platform offered equivalent predictions, the forecasts generally converge on similar probabilities for high-profile events. Divergences tend to appear on niche questions where exchange liquidity is thin and Metaculus has deeper forecaster engagement.

Several factors contribute to Metaculus’s accuracy:

• No financial barrier to entry means a larger and more diverse forecaster pool. Exchange platforms are limited to users willing to risk capital.
• Strictly proper scoring incentivizes honest reporting. On exchange platforms, strategic behavior (hedging, liquidity provision) can distort prices away from true probabilities.
• Continuous updating allows forecasters to revise estimates as events develop. Exchange prices also update continuously, but Metaculus makes the revision history transparent at the individual level.
• Long time horizons attract forecasters comfortable with multi-year questions. Exchange platforms rarely list contracts resolving more than 12 months out due to capital lockup costs.

The primary limitation is participation volume. A Metaculus question with 30 forecasters produces a less robust signal than a Polymarket market with $500,000 in volume. Forecaster count is the closest analog to liquidity, and many Metaculus questions have fewer than 50 participants.

Cross-platform accuracy data is available on the Odds Reference dashboard, updated as questions and markets resolve.

How Does Metaculus Compare to Other Prediction Markets?

Metaculus occupies a distinct niche in the prediction market ecosystem. The table below summarizes the key structural differences.

The fundamental tradeoff is between financial incentive and accessibility. Exchange platforms attract capital-motivated participants who have skin in the game. Metaculus attracts intellectually motivated forecasters who face no financial barrier. Both approaches produce useful signals, and our data suggests they are largely complementary rather than competing.

For users interested in understanding how prediction markets work, Metaculus serves as an excellent entry point. The zero-cost structure means new forecasters can build a track record and develop calibration skills before engaging with real-money platforms.

One area where the comparison gets interesting is question overlap. On events covered by both Metaculus and a real-money exchange, we track how the probability estimates diverge and converge over time. In our dataset, the two signals tend to agree within a few percentage points on high-profile events but can diverge significantly on lower-profile questions where exchange liquidity is thin and Metaculus forecaster engagement is deeper.

Key Takeaways

• Metaculus is a reputation-based forecasting platform, not a financial exchange. No real money is traded. Forecasters submit probability estimates and earn points based on accuracy.
• Continuous probability questions set Metaculus apart. While exchange platforms trade binary yes/no contracts, Metaculus captures full probability distributions over numeric outcomes, providing richer information on economic indicators and timeline questions.
• Calibration data is strong. Across thousands of resolved questions, Metaculus community predictions show tight alignment between stated probabilities and observed outcomes.
• The platform excels on long-range and scientific questions where real-money markets lack liquidity due to extended resolution timelines.
• Track live prediction data across Metaculus and other platforms on the Odds Reference dashboard.