Metaculus Guide: How to Use the Best Free Prediction Market Platform (2026)
April 12, 2026 · PolyMath Team · 13 min read
If you want to sharpen your forecasting skills without risking real money, Metaculusis the best platform you've never heard of.
It hosts thousands of carefully curated questions about science, politics, AI, economics, and world events. It tracks your accuracy over time with a rigorous scoring system. It shows you how your predictions compare to experts. And it's completely free.
This guide covers everything you need to know: how Metaculus works, how it compares to Polymarket and Kalshi, how to make your first prediction, and how to use it to become a sharper, more calibrated thinker — with a real edge when you trade on real-money platforms.
📬 Get weekly market analysis & new calculator alerts
Free newsletter — no spam, unsubscribe anytime.
What Is Metaculus?
Metaculus is a community forecasting platformwhere users make probabilistic predictions on real-world questions. Unlike Polymarket or Kalshi, there's no money involved — instead, forecasters earn points based on accuracy, and the platform aggregates community predictions into a "Metaculus prediction" that consistently outperforms individual experts.
Founded in 2015 by physicists Anthony Aguirre and Greg Laughlin, Metaculus has grown into one of the most respected forecasting communities in the world. It has been cited in academic research, used by policy organizations, and recognized as a leading example of wisdom-of-crowds prediction at scale.
Who uses it: researchers studying group epistemics, serious forecasters building a track record, analysts wanting calibrated probability estimates, and anyone who wants to think more rigorously about uncertain events.
Metaculus vs Polymarket vs Kalshi — Key Differences
These three platforms serve different purposes but overlap in subject matter. Here's a direct comparison:
| Feature | Metaculus | Polymarket | Kalshi |
|---|---|---|---|
| Real money | No (points only) | Yes (USDC) | Yes (USD) |
| US availability | Yes (worldwide) | Restricted | Yes (most states) |
| Regulation | N/A | Offshore | CFTC-regulated |
| Market types | Science, AI, politics, econ | Politics, crypto, sports | Politics, econ, weather |
| Accuracy tracking | Yes (Brier score) | No | No |
| Liquidity mechanism | Community aggregation | Automated market maker | Order book |
| Mobile app | Yes | Web only | Yes (iOS + Android) |
| Best for | Skill-building, track record | High-volume trading | US-legal trading |
The biggest difference: Metaculus rewards accuracy over capital. On Polymarket, someone with $50,000 can move prices regardless of their forecasting skill. On Metaculus, your only currency is being right — repeatedly, over time.
How Metaculus Works
Questions and Resolutions
Every Metaculus question is a well-defined, resolvable statement. Examples:
💬 "Will GPT-5 be released before January 1, 2026?"
💬 "Will US CPI inflation fall below 2.5% in Q2 2026?"
💬 "Will any country deploy autonomous AI weapons in combat before 2027?"
Every question has a resolution date, resolution criteria (exactly what conditions trigger a YES), and a resolution source (the authoritative reference used to judge the outcome). Good Metaculus questions eliminate ambiguity by design.
The Brier Score — How Accuracy Is Measured
Metaculus uses the Brier score to measure forecasting accuracy. Named after meteorologist Glenn Brier, it calculates the mean squared error between your probability and the actual outcome:
Where outcome = 1 if YES, 0 if NO. Lower scores are better.
| Brier Score | Meaning |
|---|---|
| 0.00 | Perfect — said 100% and it happened (or 0% and it didn't) |
| 0.10–0.18 | Excellent — well-calibrated forecaster range |
| 0.25 | Chance level — equivalent to always guessing 50% |
| Above 0.25 | Worse than chance — overconfident or poorly calibrated |
The key insight: the Brier score punishes overconfidence. If you say 95% and the event doesn't happen, that's a 0.9025 hit. This forces honest probability reporting — which is exactly the skill that transfers to profitable trading on real-money platforms.
Getting Started on Metaculus
1. Create a Free Account
Go to metaculus.com and sign up — free, no credit card. Your profile will track your Brier score, calibration curve, number of predictions, and ranking vs the community.
2. Browse Open Questions
The homepage shows active questions across categories. The most active in 2026:
- AI & Technology — model releases, capability benchmarks, safety milestones
- Politics & Government — elections, policy decisions, geopolitical events
- Science & Health — medical research, climate, physics breakthroughs
- Economics & Finance — inflation, Fed decisions, GDP, market events
Start with questions in areas where you already have domain knowledge. Your edge starts from the inside out.
3. Make Your First Prediction
Pick a question, read the resolution criteria carefully, then work through these steps:
The Forecasting Checklist
- What does the base rate say? How often do events like this happen historically?
- What is the current community prediction? This is your anchor.
- Do I have any information advantage? Am I more informed than the average community member?
- Am I more or less confident than the current odds suggest?
- What would change my mind? Set a reminder to update if it does.
4. Update as Evidence Changes
The best Metaculus forecasters aren't those who set-and-forget — they continuously update as new information arrives. Set calendar reminders on key dates. Update after news events. Track your reasoning in the comments.
Metaculus's Track Record
Metaculus has built an impressive track record of predictive accuracy compared to professional forecasters and expert panels:
- Predicted COVID-19 becoming a WHO pandemic in early February 2020 — weeks before official declaration
- Accurately forecasted multiple Fed rate decisions in 2022–2023 ahead of consensus estimates
- Community predictions on AI milestones (GPT-4, LLM benchmarks) have been remarkably well-calibrated
- Multiple academic studies find the Metaculus aggregate outperforms individual domain experts
Why it works: careful question wording eliminates ambiguity, Brier scoring incentivizes honest probability reporting (not overconfident hot takes), and the aggregation algorithm upweights historically accurate forecasters. The result is a wisdom-of-crowds effect that's greater than the sum of its parts.
Advanced Metaculus Features
Forecasting Tournaments
Metaculus runs tournaments where top performers win prizes (cash or recognition). These are excellent for motivated practice under competitive pressure. Domain-specific tournaments (e.g., AI progress forecasting) let you demonstrate expertise in your area.
Notebooks and Research
Metaculus Notebooks let forecasters publish long-form analysis alongside their predictions. Think of it as prediction markets meets Substack — you can write a detailed essay on why you expect a specific outcome, with a precise probability attached. This is invaluable for building a credible, citable forecasting track record.
Conditional and Series Questions
Many questions are part of series— linked questions on the same topic over time (e.g., yearly AI capability benchmarks). Conditional questions ("if X happens, will Y?") let you reason about dependent events — a more sophisticated form of forecasting than simple binary questions.
Using Metaculus Alongside PolyMath Tools
Metaculus is free and excellent for building skill — but PolyMath tools help you translate that calibration into a real edge on real-money platforms.
The Metaculus → PolyMath Workflow
Step 1: Practice calibration on Metaculus
Track whether your 60% calls resolve YES ~60% of the time. If you're systematically overconfident, correct that before risking real money.
Step 2: Find the edge with EV Calculator
Once you have a calibrated probability estimate, use the Expected Value Calculator to check if the Polymarket/Kalshi price offers positive EV.
Step 3: Size with Kelly Criterion
After finding positive EV, use the Kelly Criterion Calculator to determine the optimal bet size based on your edge and bankroll.
Step 4: Cross-reference Metaculus vs market prices
When Metaculus says 70% but Polymarket is at 55%, investigate the gap. Sometimes the community has an edge; sometimes markets have faster information.
Common Beginner Mistakes on Metaculus
Ignoring base rates
The single biggest forecasting error. Before any analysis, ask: how often do events like this happen? Elections are lost by incumbents ~40% of the time. Drug trials succeed in Phase 3 ~50% of the time. Start with the base rate and adjust from there.
Setting and forgetting
Predictions should update as new information arrives. A 30% call in January may be 70% by October if conditions change. Set reminders and revisit your active predictions regularly.
Anchoring on the community prediction
The community prediction is a reference point, not an answer. If you have genuine information or reasoning that differs from consensus, express it — that's what moves the aggregate toward truth.
Using extreme probabilities
Saying 99% or 1% is almost always overconfident. There's almost always some probability the resolution criteria don't work out as expected, even if the underlying event itself is near-certain.
Not writing comments
Comments force you to articulate your reasoning. If you can't explain your forecast clearly, that's often a sign of fuzzy thinking. Write it out — you'll spot the flaws yourself.
Which Platform Should You Use?
Use Metaculus if you:
- Want to build forecasting skills without financial risk
- Are interested in science, AI, or long-horizon questions (years, not days)
- Want your accuracy tracked over hundreds of predictions
- Are building a credible public forecasting track record
Use Polymarket if you:
- Want real financial upside from your predictions
- Focus on near-term events (days to months)
- Are comfortable with crypto (USDC wallet required)
- Are not a US resident
Use Kalshi if you:
- Are a US resident who wants fully legal, CFTC-regulated trading
- Prefer fiat USD over crypto
- Want FDIC-insured cash holdings
The ideal approach: use all three. Metaculus for calibration and practice. Polymarket for volume and price discovery. Kalshi for US-legal regulated exposure. Each serves a different layer of your forecasting stack.
The Bottom Line
Metaculus is the most rigorous free tool available for developing prediction market skills. If you want to be a better forecaster — and translate that into an edge on real-money platforms — Metaculus is where you build the foundation.
Start making predictions. Track your Brier score. Update your beliefs as evidence changes. And when your calibration is sharp, bring that edge to where it earns real returns.
Put Your Calibration to Work
Use PolyMath's free calculators to find +EV opportunities and size positions mathematically on Polymarket and Kalshi.
📬 Get weekly market analysis & new calculator alerts
Free newsletter — no spam, unsubscribe anytime.
Frequently Asked Questions
What is Metaculus?
Metaculus is a free community forecasting platform where users make probabilistic predictions on real-world questions about science, politics, AI, economics, and world events. Unlike Polymarket or Kalshi, there is no real money involved — instead forecasters earn points based on accuracy, and the platform aggregates predictions into a crowd forecast that consistently outperforms individual experts.
How does Metaculus measure accuracy?
Metaculus uses the Brier score to measure forecasting accuracy — the mean squared error between your probability estimate and what actually happened. Lower Brier scores are better. A perfect score is 0 (you said 100% and it happened). A score of 0.25 is equivalent to always guessing 50%. The Brier score penalizes overconfidence and rewards well-calibrated, honest probability estimates.
Is Metaculus free to use?
Yes, Metaculus is completely free. There is no real money involved — forecasters earn points based on accuracy. The platform is available worldwide with no geographic restrictions, unlike real-money platforms like Polymarket (restricted for US users) or Kalshi (US-only).
How is Metaculus different from Polymarket?
The key difference is money: Polymarket uses real USDC cryptocurrency, while Metaculus uses a points-only system with no financial stakes. Metaculus excels at long-horizon questions (months to years) on science, AI, and policy. Polymarket focuses on near-term events with real financial incentives. Metaculus tracks your accuracy via Brier scores; Polymarket does not. Metaculus is available worldwide; Polymarket is restricted for US residents.
What is the Metaculus community prediction?
The Metaculus community prediction is an aggregated forecast from all users who have submitted predictions on a question. Metaculus also generates a separate 'Metaculus prediction' that uses a proprietary algorithm to upweight historically accurate forecasters. Studies have shown this aggregated prediction consistently outperforms individual expert estimates across many domains.