When Others Bet You'll Finish
Markets forecasting Manifold users’ own futures were too optimistic on average, including when those users did not trade.
What should a crowd’s forecast add to your decision about a deadline? You could use the answer to reconsider the scope, ask for help or move the date. A specific objection may be more useful for that choice than reassurance alone.
On Manifold, people trade shares in answers such as YES and NO; the price is displayed as a probability. Manifold uses play money called mana. Its market price reflects trading, not a simple count of people who agree. Someone can buy an outcome they consider underpriced without thinking it is certain—or even the likelier outcome.
An October 29 preprint examined 6,150 resolved Manifold markets about users’ own futures. YES occurred in 45.0%, against mean final prices of 49.4%; external-event markets’ overstatement was 0.9 percentage points.
A public forecast need not supply the independent perspective a person hopes to get by asking others. The useful question is how that number becomes informed about the particular person and the work still ahead of them.
A forecast is not a verdict
In April 2022, Bethany Soule asked whether she would deploy usable Stripe-related code by Sunday. A commenter called Austin acknowledged having “No real context,” but reported buying YES because of his experience finding Stripe easy to deploy. The market eventually resolved NO.
One NO result does not establish that a YES purchase was unreasonable. A sound forecast can assign a substantial chance to an outcome that does not occur. Calibration is tested across many forecasts: do outcomes happen about as often as their stated chances imply?
The research supplies that broader comparison. It also reports that the overstatement emerged after opening, including where the person being forecast never traded.
Austin’s comment makes a different distinction visible. Experience deploying a technology can inform a judgment about the task. It does not, by itself, establish how far this particular developer has progressed, what else needs doing or how much time remains available. That is a limit of the reason offered, not a claim that his reasoning caused the study’s pattern. The historical market is an illustration; its membership in the research sample is unverified.
Who can change the answer?
Over the markets’ lives, subjects’ orders supplied 38% of net price movement toward eventual outcomes, versus 9% for external-market creators. Moves away from the outcome are subtracted.
The observational study cannot separate progress knowledge, control over behavior and outcome reporting. Those possibilities matter: learning that the work is nearly done differs from deciding to do it, and both differ from deciding how an ambiguous result should be settled.
A personal market therefore has an unusual relationship to the work it describes. The person can continue working, tell others what has changed, take a position in the market and help determine the eventual answer. The displayed probability does not put them outside that process.
Accountability is another use
Nor is prediction the only purpose. Manifold presents personal-goal markets as a form of accountability, including betting on yourself while allowing others to bet against you. That is a proposed use of the platform, not evidence that it improves follow-through. A person could value the public commitment while still needing an honest estimate of whether the deadline is realistic.
Wanting support does not oblige you to mistake support for a schedule. Conversely, wanting an accurate estimate does not make a public commitment pointless. The commitment and the estimate can be worth something for different reasons.
For a developer, that leaves a concrete contribution to make: explain what is already working and what remains unfinished. The aim is to make the remaining work discussable, not to guarantee a better forecast. Instead of treating a favorable price as the end of the conversation, the person who set the deadline can bring the unfinished part of the task back into it.
Sources
- Zihan Chen, Di Zhu and Lei Nico Zheng, Crowds Believe in You: Online Prediction Markets Overestimate How Often People Achieve Their Own Goals, October 29, 2026 preprint, version 1; §§3–5 and Appendices A–C.
- Manifold, FAQ, platform mechanics and stated uses; inspected November 1, 2026.
- Bethany Soule, historical deployment market and comment, April 2022; dates and outcome checked against the public market record, with comment chronology checked through the documented public API.