What does it mean to resolve a prediction market?

The Ultimate Guide to Crowdsourced Forecasting

« Back to All Topics

We often say that a good prediction market question will eventually have a "knowable" answer, which is to say that, one day, you need to know whether the answer to the question was right or wrong. Fairly frequently people want to ask questions like "Which of these ideas is the best?" Unfortunately, this type of question would be a poor fit for a prediction market (though it may be a good fit for other, non-prediction market approaches that we can support), since the answer(s) will be opinions with no right or wrong answer. A good heuristic for good prediction market questions is whether or not you could make a bet with a friend about the answer to the question. If you can bet on it, it's likely a good prediction market question.

When the day comes that we know the correct answer to a question, we can now "resolve" the question. Resolving a question is also known as judging the question. Imagine we ask the question, "Which team will win the Super Bowl in 2018?" and each of the NFL teams is a stock in the market. Participants trade in the market all season long and we end up with a Bears-Bengals Super Bowl. For all teams besides the Bears and Bengals, we can resolve them as incorrect, since we definitively know that they will not win the Super Bowl. For each of those teams, the "correctness" was known the moment that they were eliminated from the playoffs (or from playoff contention). 

Say the game ends February 4 at 9:28pm. We would say that the "correctness" of the Bears and Bengals was known February 4, 9:28pm. Let's assume the Bears win. We would then resolve the Bears as correct and the Bengals as incorrect. In a prediction market platform like Cultivate Forecasts, the platform will automatically resolve all other remaining answers when you mark an answer as correct (since if one answer is correct, then de facto all the other remaining ones are incorrect).

Once an answer is resolved as correct, then the platform can pay out winnings, similar to a sports book. In a prediction market, users who placed trades in the market will hold shares of the stocks (for more information about prediction markets, trades, and stocks, check out our prediction markets intro). Shares of the correct answer are worth $100/share, while shares of any incorrect answer are worth $0/share.

Returning to our Super Bowl example, let's say you bought 10 shares of the Bears early in the season for $5/share. You held onto those shares all season long, and now they've won the Super Bowl. You are then paid $100/share for each of your ten shares, meaning your payout is $1000. Since you bought the 10 shares at $5/share, you only spent $50 to acquire those shares, netting you a $950 profit. 

Market resolution is typically performed by an administrator of the platform, rather than any participants. Some platforms, including Cultivate Forecasts, support participant-suggested resolutions (which must be approved by an administrator).  Once we've designated the correct answer and paid all participants their winnings, we consider the market "resolved."

Go back to The Ultimate Guide to Crowdsourced Forecasting and Prediction Markets