Prediction markets

Description

Participants trade shares in mutually exclusive outcomes, or estimate others' behaviour independently; aggregated prices or estimates are read as crowd probabilities for product questions.

Instead of asking 'would you buy this?', a prediction market asks people to bet on what others will do. In the market form described for product research, each participant receives a stake (for example $10) to buy shares in answer options whose prices sum to 1 and move with demand; the market closes when prices stabilise, around 150 participants, and payouts equal shares times final price, so participants are rewarded for accuracy rather than for expressing their own preference. A simpler form aggregates independent estimates of a quantity, such as how many people will buy in six months. Crowds outperform individuals on simple questions with definite answers when members are diverse and independent; authority and social influence spoil the aggregate. Cash incentives and trickle-in participation are what make the prices informative.

Attributes

Loss type
Siloed
Effort
Weeks

Steps

  1. Write a market question that is measurable, specific and concise, with mutually exclusive options.
  2. Recruit a diverse panel that can enter over time, and fund a cash stake per participant.
  3. Give each participant a fixed budget and require them to spend most of it (for example at least $7 of $10) within a time limit (for example 10 minutes).
  4. Let prices update with demand, keeping option prices summing to 1, and keep participants independent of each other's reasoning.
  5. Close when prices stabilise and pay out shares times final price.
  6. Read final prices as probabilities and compare them with survey results and, later, with actual outcomes.

Sources

  • Quant UX Con 2022 Proceedings (Chapman et al., 2022)
  • Universal Principles of Design (Lidwell et al., 2023)
  • Wolfers, J., & Zitzewitz, E. (2004). Prediction markets. Journal of Economic Perspectives, 18

Also known as

  • Betting (prediction) markets for product research
  • Prediction market
  • Information market
  • Prediction markets for estimating demand
  • Crowd intelligence estimation
  • Wisdom of crowds

Categories

  • Analytics & Behaviour
  • Behavioral Science & Economics