Prediction markets leverage the wisdom of the crowd by aggregating information that is dispersed among individuals. They are innovative trading platforms where individuals can place bets on the outcomes of future events.
How do they work?
- Each prediction market asks a question about an unknown future event that is relevant for the firm, with a specific terminal date. The answer to the question can either be a “Yes” or a “No.”
- For example, “Will our firm’s revenue surpass €10 million in the next quarter?”
- Participants are given an amount of money that they can use to buy or sell “Yes” shares, which pay €1 if “Yes” is the outcome, or “No” shares, which pay €1 if “No” is the outcome. Thus, participants become traders.
- At the start of the prediction market, both the price of the “Yes” and the “No” shares are set at €0.50. If enough “Yes” shares are purchased by traders, their price will increase; if enough “Yes” shares are sold by traders, their price will decrease. The same holds true for the “No” shares. The price of a share will always range between €0.01 and €0.99. The sum of the “Yes” and the “No” prices is always €1.
- On the terminal date, if the “Yes” outcome materializes, for each “Yes” share a trader owns, they will earn €1. If the “No” outcome materializes, for each “No” share they own, they will earn €1. In addition, any unspent money they have from their monetary endowment will be added to their total earnings.
- The price of the “Yes” shares is viewed as the market’s predictive probability that the “Yes” outcome will materialize, and similarly for “No”.
- Each prediction market requires at least 30 active traders.
- Traders will also be required to complete two short surveys (around 20 minutes each), before and after the market.
The following video showcases our platform.
What does it cost for a firm to run prediction markets?
There is no cost to firms for running prediction markets on our platform. We will also provide the monetary endowments to employees, subject to a cap on their number. If they wish so, firms can contribute to the monetary endowment, for example to increase the number of traders or prediction markets. The monetary endowment is relatively low, for example on average €5 per trader per market, plus another €5 for completing the surveys. Employees can participate in multiple prediction markets at the same time. They trade on our platform using an artificial currency, which is converted to euros when the market ends.
Do firms need to surrender any of their sensitive data?
No, data privacy is one of our utmost priorities. Each trader will be issued with a username and a password. We do not need to know their personal or work email, their name, or any other personal data. We will record and retain each trader’s buy and sell orders, but we do not need to know the prediction market question. The identity of the participating firms will never be made public without their consent. Our study has passed a rigorous and academic peer-reviewed process, and our research team is experienced in conducting surveys and in data management. The data will be treated with the utmost confidentiality. Once the data is collected, we will remove any information that leads back to the person that participated; that is, everything will be anonymised. We will then aggregate the data and conduct analysis with the aggregate data.
What are the aims of this project?
We have two aims. First, we want to use the anonymised data to produce research papers that analyse the effectiveness of prediction markets. Second, we want to inform firms about the usefulness of prediction markets and educate them on how they can employ them to improve their decision making. To that end, we also plan to organise workshops, where both academics and businesspeople will participate. Currently, only big firms, such as Google and Ford, use prediction markets internally.
Are prediction markets accurate?
As the prices of the “Yes” and “No” shares move up or down, they reveal to everyone the private information of the traders, prompting them to update their beliefs and trade again. This feedback loop can make prediction markets more accurate than polls and professional forecasters. Their effectiveness stems from an old economics principle, that prices aggregate information.
What are their advantages?
- They provide timely information and enable a manager to act quickly.
- They are private, as data on trades and prices are never made public.
- The design of the market is tailored to the firm’s needs. For example, the firm chooses who participates.
- Predictions can circumvent the hierarchical structure of a firm as information from all participating employees is aggregated, independently of their position.
Examples of questions
- Will our firm’s sales surpass $10 million in the next quarter?
- For a given product or service, will feature A be more popular than feature B with our customers?
- Will our firm’s app exceed 1 million users by a certain date?
- Will the total market for a product grow by more than 10% year on year?
Who uses prediction markets?
Prediction markets can be public, where anyone can participate, or private. Ford uses private markets to predict the future demand for their cars. A salesman in New York can predict accurately the demand in his own area, but not the demand in California. A prediction market can aggregate this information among different salesmen, though prices. A prediction market can work well if traders have differential information, because they are in different departments or areas.
Google uses prediction markets for specific quarterly goals of high importance, such as the number of users in a service, a third-party quality rating, or the on-time completion of key products. The aim of Google is twofold. First, to aggregate information for management about the success of a particular project. Second, to communicate the management’s interest in the success of the project.
Who are we?
The research team consists of two Professors of Economics and two postdoctoral researchers. We have secured two research grants, from the UK and France, totalling around £1 million. You can find more information about the team here.
Spyros Galanis, Professor of Economics, Durham University Business School
Christos Ioannou, Professor of Economics, Université Paris 1 Panthéon-Sorbonne
Sergei Mikhalishchev, Postdoctoral Research, Durham University Business School
Dominik Schmidt, Postdoctoral Research, Université Paris 1 Panthéon-Sorbonne