Opinions are cheap. Dated, quantified predictions are expensive — and the cost is what makes them informative. Prediction Arena turns your vague priors ('AI hiring will slow in Q3') into stakeable forecasts ('60% chance unemployment in software engineering roles exceeds 4.5% by Aug 1').
Every prediction you log is time-stamped, probability-weighted, and resolved against reality when the event date lands. Over months, your reveals your true — not the one you claim in conversation.
This is the mode that produces the best long-form evidence of your judgment. Superforecasters weren't born with a gift; they accumulated thousands of these receipts and learned to read them.
This trains
Forecasting
Long-horizon across self-authored predictions, with explicit probability statements and hard resolution dates.
How a session feels
4 steps, 3–5 minutes. Repeat until the feedback starts shaping your instincts.
- 1
Write a prediction
Must have a resolution date and a specific, checkable outcome. 'Will Xi meet Modi before June 30?' counts. 'Things will get weirder' doesn't.
- 2
Assign a probability
Anywhere from 5% to 95%. The extreme tails cost you badly when you're wrong — be deliberate.
- 3
Wait
Resolution happens automatically on the due date. Nothing to do but hold your rating and avoid unconscious revisions.
- 4
Read your curve
After 20+ resolutions, your personal curve emerges. This is the mode's real output.
Who it's for
- Traders, VCs, intelligence analysts — the job is distribution thinking
- Anyone preparing for a public-stakes forecast (elections, market moves, product outcomes)
- Writers and pundits who want to back claims with track records
What the research says
Try a challenge — no sign-up
The demo pulls from the public challenge bank. Your confidence rating and result are the same mechanics you'll see in the real mode.
Start with Prediction Arena
Your first session generates a score baseline in under 10 minutes.