Are Prediction Markets for Real?

Are Prediction Markets for Real?

The impending US presidential election has many asking are prediction markets for real?

To explain, a prediction market is a platform that takes bets on the potential outcome of events. The idea behind prediction markets is that bettors are more honest than poll respondents and experts.

How Prediction Markets supposedly Work

For example, a person who dislikes Donald J. Trump (R-New York) but thinks Trump will win next year’s election may not admit it. However, that person could bet money on Trump’s reelection.

Thus prediction market operators hope the possibility of monetary reward will overcome emotions that distort judgments. Additionally, prediction markets operate on the premise that large groups of people could have access to information experts or pollsters lack.

For instance, a person who lives in Ohio or New Mexico could have a more accurate impression of political opinions in Middle America than a journalist in New York. A prediction market can theoretically harvest that data and transmit it.

Do Prediction Markets Work?

Data from the 2016 American presidential election show prediction markets are less accurate than other popular methods of data-based forecasting. However, the same data shows all the major data-based forecasting methods flopped in 2016.

Notably, prediction markets forecast Hillary R. Clinton (D-New York) receiving 54.6% of the popular vote, and Trump receiving, 45.4% of the popular vote, Professor Allan J. Licthman calculates. In reality, Clinton won 48.2% of the popular vote to Trump’s 46.1%, CNN estimates.

Licthman’s analysis indicates all the data-based forecasters failed in 2016. Poll aggregators, for instance, gave Clinton 52.1% and Trump 48%. Meanwhile, citizen forecasters predicted Clinton would receive 52.2% to 47.8%.

Interestingly, the experts were just as wrong as ordinary people.  Licthman, a distinguished professor of history at American University, estimates experts forecast Clinton at 52.2% and Trump at 47.8%. Finally, index models gave Clinton 53.5% and Trump 46.6%.

In the final analysis, prediction markets were worse than the other methods. The picture predictors painted was just as cloudy as the other popular crystal balls.

Can Prediction Markets Work?

I think Licthman’s data shows emotional judgments are just as likely to cloud prediction markets as other forecasts. Specifically, subjects that generate strong emotions can skew opinions.

A major flaw in prediction markets is the rational market delusion. To clarify, the rational market delusion is the belief that all or most of the players in market think and act rationally. Additionally, rational market models operate under the illusion that human beings always act rationally in economic matters.

The $495 billion Statista predicts the global gambling industry will rake in, in 2019 demonstrates how irrational people are with money. Yet the prediction market operators claim their marketplaces can make rational decisions about human behavior.

Consequently, I think present-day prediction markets cannot work. My guess is that prediction markets will need better data aggregation and analysis technology to work.

Algorithms that could screen out human emotions or sentiments could make a prediction market effective. On the other hand, artificial intelligence or machine learning could soon provide that technology.

Can Prediction Markets Make Money?

Oddly, the global gambling industry’s revenues prove prediction markets could still make money.

Specifically, Statista estimates global gambling’s profits could grow from $450 billion in 2016 to $495 billion in 2019. Not surprisingly, many cryptocurrency speculators are prediction market true believers.

One of the most popular prediction market tokens Gnosis (GNO) had a Market Capitalization of $26.50 million, a Coin Price of $24, a Circulating Supply of 1.105 billion GNO, and a 24-Hour Market Volume of $294,568 on CoinMarketCap on 3 June 2019. Gnosis was CoinMarketCap’s 179th most popular cryptocurrency on June 3, 2019.

Meanwhile, CoinSwitch estimates Gnosis (GNO) had a Coin Price of $23.907, a Market Capitalization of $26.408 million, and a 24-Hour Market Volume of $292,648 million on the same day.

CoinMarketCap estimates Gnosis’s competitor Agur (REP) was even more popular. CoinMarketCap gave Augur a Market Cap of $209.396 million, a Coin Price of $19.04, a 24-Hour Market Volume of $22.677 million, and a Circulating Supply of 11 million REP on 3 June 2019. Augur was CoinMarketCap’s 43rd most popular cryptocurrency on that day.

Meanwhile, CoinSwitch estimates Augur had a Coin Price of $18.92, a Market Cap of $208.192 million, and a 24-Hour Market Volume of $22.648 million on the same day. Hence, there is a strong demand for prediction markets even though their utility is questionable.

Can Prediction Markets Predict the 2020 Presidential Election?

The 2020 US Presidential Election could make or break prediction markets because of the interest being shown in them.

Interestingly, some recent prediction market forecasts look accurate to me. For instance, PredictWise’s aggregate of prediction markets forecasts Democrats will win 52% of the popular vote in the 2020 presidential contest. PredictWise estimates Republicans will receive 49% of the popular vote.

Incredibly, this prediction could have little value even if it is accurate. In America the unelected Electoral College not the voters, elects the President.

Moreover, the Electoral College and the popular vote are diverging. In 2016, Clinton won a clear majority of the popular vote 48.2% to 46.1%. Yet Trump won a decisive majority in the Electoral College 306 to 232, 270toWin calculates.

Could Artificial Intelligence Forecast the Presidential Election?

However, I think somebody could build an artificial intelligence (AI) that can accurately mimic the Electoral College vote. Plus, it could be possible to use AI to build a prediction market that mimics the Electoral College.

An AI could analyze the Electoral College members. Then the AI could select a pool of people that matches the Electoral College members’ characteristics. A tabulation of those people’s votes could give you an accurate forecast of the 2020 presidential election.


If it exists, such an AI; or prediction market, could give one party or candidate an unfair advantage in a Presidential Election. Consequently, such prediction markets and AI will raise serious ethical and political questions.

Even though they are far from accurate Prediction Markets are here to stay. I predict interest in prediction markets will remain intense for the next 19 months and keep Gnosis and Augur’s coin prices high.

Originally published at on June 3, 2019.

Close Menu