Category: Gnosis (GNO)

The Power of Prediction Markets

Prediction markets effectively aggregate a variety of information and beliefs.

As Hayek describes, markets induce an intrinsically dynamic and selective process which generates continuous change in global data [1]. Markets make us not only acquire, but also convey knowledge and beliefs. Their inherent dissemination of information through market prices best makes use of the dispersed knowledge which all the different individuals possess.

The ability of markets to aggregate this dispersed information has inspired us to create an open prediction market platform on Ethereum.

What is a Prediction Market?

Sometimes called information markets, idea futures, event derivatives, decision markets, or virtual stock markets, prediction markets are exchange-traded markets where individuals trade the outcome of a future event in the form of event contracts. These event contracts specify the different possible outcomes of a future event, the payment structure based on those outcomes, and the contract’s expiration date.

Where will Amazon locate its next headquarters? (Image via Austin Monitor)

For example, a prediction market could ask Where will Amazon locate its next headquarters? Market participants can choose between different outcomes such as Austin, Phoenix, or Chicago, and would be paid out $10 if their predicted outcome occurs or $0 if it does not occur. The price of the prediction market reflects the market’s expected probability of the outcome occurring. Thus, for a contract paying $10 if a specific outcome occurs, and if the current market price of Austin is $9, then the market believes that the probability for Austin becoming the city for Amazon’s next headquarters is 90%. Watching the market prices change over time hence shows us how the probability estimates of the market participants on a certain outcome also change.

The market’s expected probability of an outcome occurring is the aggregated probability estimate of each market participant. Let’s say you participate in the prediction market about Amazon’s headquarters by investing $10. For your investment of $10, you will receive 10 shares for Austin, Phoenix, and Chicago, respectively. You can then go ahead and sell them with other market participants at the current market share price.

If you believe that Austin will be Amazon’s chosen city, you would sell all your Phoenix and Chicago shares. The price you can ask for the Phoenix and Chicago shares depends on the market’s estimate on Chicago or Phoenix being selected by Amazon. Assuming the market believes that Chicago will win with a 50% probability, and Phoenix with a 10% probability, you can sell the Chicago shares for $5 and the Phoenix shares for $1.

That means that you’ve only invested $10 (initially put in) – $5 (received from selling the Chicago shares ) – $1 (received from selling the Phoenix shares) = $4 to purchase the 10 Austin shares. If Amazon then ends up choosing Austin, your Austin share becomes worth $10 while the Chicago and Phoenix shares become worth $0. Therefore, you’ve made a profit of $10 – $4 (invested to purchase the Austin shares) = $6. If you wouldn’t have traded both the Chicago and Phoenix shares, you would have just gotten your money back (since both Chicago and Phoenix shares would have become worthless and the Austin share worth $10).

A preview of the Prediction Market Page on Gnosis

What could a Market predict?

A prediction market could be an efficient tool to predict whether a particular legislation is likely to be passed; to forecast inflation or deflation in house sales in a certain area at a specific time (this would help contextualize quotations from said experts on the future of the housing market). Traffic patterns, the number of restaurants that will open up in a certain neighborhood, or future test scores in school districts that are implementing reforms are some other examples of relevant information that could be captured in a prediction market.

Will the UK’s GDP increase by more than 5% by the end of 2018? (Image via LinkedIn)

Prediction markets could also be extremely useful capturing conditional estimates, such as the chance of a country’s GDP increasing given that it leaves the European Union, or more importantly, estimates that depend on a decision to be made, such as the chance of a health care system reform given that we elect a particular person for president or the stock price of a company given that the current CEO is fired. Such estimates can serve as a foundation for our decision making in any type of organization.

Unlike financial markets like stock or commodities futures that are created for traders to hedge risks (farmers use futures markets in their crops to hedge against lower crop prices; airlines use futures markets for oil to hedge against the risk of higher fuel prices), prediction markets primarily seek to aggregate information on particular topics of interest. The main informational value of a prediction market lies in its price which does not simply represent an average assessment by the market participants, but also reflects the confidence level that different participants have in their predictions. Since no one is forced to participate in the market, those who do tend to be those who have reliable information or at least those who can acquire this information at low cost.

What makes Prediction Markets such a Powerful Tool?

  • Prediction markets efficiently aggregate a variety of information and beliefs. Individuals have different sets of information and different
    beliefs—our societies are full of what Hayek calls dispersed knowledge [2]. Due to this dispersed knowledge, individuals arrive at different expectations of the probabilities of future outcomes. Prediction markets excel at identifying this dispersed knowledge; they’re a mechanism for weighing the estimates of market participants based on the information that the individuals possess.
  • Prediction markets create financial incentives for truthful revelations.
    By forcing market participants to bear the financial consequences of their prediction, individuals who continually lose money by making bad predictions will stop participating in the market. Those who make good predictions, however, will be rewarded and therefore have an incentive to continue participating in the market. With economic incentives, individuals are more likely to be truthful about what they believe, and show how strongly they believe it.
  • Prediction markets provide incentives for gathering relevant information. Rather than setting up a polling panel with a pre-defined group of experts who are thought to possess useful information based on criteria used by the person who makes the selection, prediction markets create an incentive for individuals with relevant information to come forward and trade that information, and thereby revealing it to the market. Individuals without the relevant information are incentivized to acquire it in order to participate in the market and gain profits. It’s a process similar to the selection of unbiased predictors, but rather than getting better predictions from existing information, it brings new information to the market.
  • Prediction markets incorporate new information quickly.
    Prediction markets are happening in real-time and are constantly updating. Hence, new information is rapidly incorporated in the market — a characteristic that polls, expert surveys, and other methods of aggregating information have a hard time to replicate. For many forecasts, such as the probability of outbreaks of a deadly disease or virus, the speed at which new information is reflected in the market price is key.
  • Prediction markets are difficult to manipulate.
    When multiple prediction markets exist and price divergences based on market manipulation occur, there’s an arbitrage opportunity for traders. However, Theory suggests that traders, when they can’t exactly identify the manipulators, compensate for the price divergence by an opposite trade, and then compensate for the price variation by trading more, and by trying harder to acquire the relevant information. Thus, current prices in prediction markets in which considerable trading occurs provide pretty accurate estimates of the probability that a certain event will happen or not.

A Track Record of Successful Forecasting

Prediction markets are speculative markets with the primary purpose of aggregating information rather than hedging risks. Regardless of the main purpose, speculative market prices generally turn out to be quite accurate estimates of future prices due to their ability to aggregate vast amounts of available information. Orange juice commodity futures markets, for example, make better weather predictions than the government, and horse races are better predicted by betting markets than by professional handicappers.

Despite the availability of scientific polling, poll aggregators, and a wide range of forecast and expert opinions, prediction markets have a track record of successfully forecasting events — from political outcomes like elections, wars, or revolutions; natural outcomes like earthquakes or the weather; company-wide performance indicators such as predicted sales or revenue; to economic variables like the GDP or the unemployment rate.

Empirical evidence confirms that prediction markets are more accurate and have half the forecast error compared to traditional polls. For example, the Iowa Electronic Market forecasts were more accurate 451 out of 596 times when compared to major opinion polls on U.S. presidential elections. Reducing forecast error by 5 percent on average, prediction markets also outperformed a survey of experts in forecasting payrolls, unemployment claims, retail sales, business confidence, and other macroeconomic indicators.

Iowa Electronic Market on the US election ’08: Obama vs. McCain

Companies have made use of the information-revelation benefits of prediction markets, too: Hewlett-Packard’s prediction markets on sales of printers were more accurate 6 out of 8 times, even though the official forecasts were made after the markets closed, i.e. when the market prices were widely known by official forecasters. Prediction markets outperformed HP’s traditional forecasts in the price of computer memory three and six months ahead as well, being 70 percent more accurate. Best Buy used prediction markets for several purposes such as the demand for digital set-top boxes, store opening dates, or whether new services will be launched on time.

Prediction markets even triumphed over historically based forecasts in health: The number of dengue fever outbreaks was accurately forecasted by these markets.

Thinking about the endless use cases, their inherent ability of inducing people to acquire and reveal information, and their successful track record when used by both public and private institutions, we find the idea of prediction markets immensely powerful. Decentralized prediction markets even more so.

[1] Competition is essentially a process of the formation of opinion: by spreading information, it creates that unity and coherence of the economic system which we presuppose when we think of it as one market. It creates the views people have about what is best and cheapest, and it is because of it that people know at least as much about possibilities and opportunities as they in fact do. It is thus a process which involved a continuous change in the data and whose significance must therefore be completely missed by any theory which treat these data as constant.
In Hayek, F.A., 1946. The Meaning of Competition. Reprinted in Hayek, F. A., 1948. Individualism and Economic Order.

[2] “The peculiar character of the problem of a rational economic order is determined precisely by the fact that the knowledge of the circumstances of which we must make use never exists in concentrated or integrated form but solely as the dispersed bits of incomplete and frequently contradictory knowledge which all the separate individuals possess.”
In Hayek, F.A., 1946. The Use of Knowledge in Society. Reprinted in Hayek, F. A., 1948. Individualism and Economic Order.

The Power of Prediction Markets was originally published in Gnosis on Medium, where people are continuing the conversation by highlighting and responding to this story.

<div class="infobox"><span class="appendinfo">This article was originally published on: <a href="" target="_blank">The Gnosis Blog</a> on </span></div>