FORECASTS WITH DEFINED QUESTIONS
Prediction Vote API
Make the forecast question precise.
A forecast needs a question that can eventually be resolved. A preference poll asks something different. A prediction vote API concept should preserve that distinction, along with the time horizon, resolution source, and meaning of the reported probability. This page helps readers plan those fields and evaluate their interpretation; it does not provide predictions, claim forecasting accuracy, or describe a trading service.
Define the event before collecting opinions
Write the event in terms that another reviewer can apply consistently. Include the subject, qualifying condition, deadline, and evidence that would resolve the question. Explain what happens if the event changes, the source disappears, or a required condition cannot be assessed. The Metaculus FAQ provides a reference for how a forecasting platform describes question types and resolution. Your own proposed question still needs explicit wording. Avoid a heading such as “Will it happen?” when the page leaves the event and relevant horizon unstated.
Label probability, participation, and confidence separately
A forecast probability is not the percentage of people who prefer an outcome. Record the aggregation method and the relevant observation time without converting them into a promise. If a display also includes participant counts or a range of estimates, label those measures separately. Explain whether a number reflects one forecaster, an aggregate, or another specified method. In an illustrative interface, a seventy-percent estimate should remain an estimate; it should never become a “seventy percent complete” progress indicator or an observed election result.
Preserve revisions and unresolved outcomes
Store changes to a forecast as dated observations so a reader can inspect what was estimated before the outcome was known. Keep the final resolution separate from the prediction history. Define unresolved, ambiguous, and canceled conditions rather than forcing every question into a binary result. For evaluation, choose a documented method and a complete set of eligible questions before examining performance. Selecting only favorable examples after the fact would make an accuracy claim difficult to assess and unhelpful for choosing an integration.
COMMON QUESTIONS
A clearer
starting point.
Can poll support be used as an outcome probability?
Not automatically. A preference share describes responses under a poll's participation and sampling conditions. An outcome probability describes a forecast about a defined event. Connecting the two would require a stated model and evidence for its interpretation, rather than relabeling the poll's largest percentage as a prediction.
What should a forecast page say when resolution is delayed?
Explain that the question remains unresolved, identify the relevant resolution condition, and preserve the last forecast with its original timestamp. Avoid backdating a new estimate or presenting a missing outcome as failure or success. A clearly labeled unresolved state is more useful than an invented certainty.
CONTINUE EXPLORING
Connect the ideas.
All 18 topic guides01 / 18
Vote API
The phrase vote API can describe a public results feed, an informal poll, a forecasting interface, or a governance tool. Those systems.
EXPLORE THE GUIDE
04 / 18
Presidential Vote API
Presidential elections do not share one universal data model. A page must identify the country, election event, geographic scope, and process represented.
EXPLORE THE GUIDE
18 / 18
Political Vote API
A political information interface should help readers understand who a record describes, which office or event it relates to, and where the.
EXPLORE THE GUIDE