Gambling with Prediction Markets

As part of its internship program, Navigator asks its interns to write a blog post about the intersection of communications and an area of personal interest. This week’s post is from Connor Whitworth.

Do you gamble? With gambling, you make a bet on a potential outcome. If you’re correct, you are rewarded for predicting that outcome. You get some money. You feel like you won. The more certain you are of a particular outcome, the more money you’re willing to bet. People bet on all kinds of things — on horse races, on sports games, on cards.

Prediction markets differ from traditional market research. Rather than asking questions like, ‘what do you prefer,’ or ‘which would you rather,’ prediction markets ask, ‘what do you believe is going to happen,’ or ‘which product do you believe has a greater likelihood to succeed.’ In this scheme participants in the market, play with real money and invest in their opinions.

Essentially, prediction markets are bets to see who is ‘the most right’. Participants form a market by agreeing to wager their predictions against each other for a predetermined amount of time. When that time passes the ‘market’ closes and their scores are tallied. Once closed, predictions end with a numeric value that relates to the other predictions or derivatives. If you invest early, aka predict early, and others make the same prediction, you see an increase in profit. Conversely, those who predict incorrectly see the value of their investment decline.

Outside of finance, prediction market principles can be applied to analyze public expectations.We’ve already seen the wisdom of crowds harnessed to predict political outcomes. In 2013, David Rothschild, an economist from the Wharton School of Business, published an academic study with some interesting findings. Over the last 60 years, poll questions asking people which candidate they believe has a better chance of winning consistently act as a better guide to the outcome of the presidential race than polls that simply ask people for whom they plan to vote (for interest’s sake, Bernie is currently going off at 16/1 on popular European betting site PaddyPower.)

In politics, predictive markets are powerful tools but they’re not new. In fact, prediction markets are among the oldest exchanges in America. Over the last decade we’ve seen a resurgence of prediction markets in Canada after they fell out of favour in the late 1990s. The Sauder School of Business at the University of British Columbia operates a political prediction market covering a wide range of events. It has four objectives:

1. Accurately predict the outcomes of political events;
2. Learn about trader behaviour in a controlled environment;
3. Engage traders to follow the political process more intensely; and
4. Teach participants about financial markets and trading strategies.

Its predictions are often more accurate than public opinion polls. In the October 2015 Canadian federal election, the Sauder Political Prediction market was the most accurate body using models to predict the Canadian popular vote, it operated with a margin of error equaling only 2.3%. The next closest body, a scientific polling firm, had a margin of error of 4%.

And now, individuals are capitalizing on collective intelligence to better large corporations. An increasing number of companies are turning to predictive markets to gauge the future success of products, concepts, promotions and campaigns.

The California Institute of Technology found that Hewlett-Packard’s masterful use of prediction markets produced forecasts that were closer to the actual outcome for 6 out of 8 events in its model. HP’s data scientists were able to distill large business problems into a series of smaller, easier to conceptualize questions. Running simultaneous prediction models of those smaller questions allowed them to understand how the larger issue was most likely to unfold. Hewlett Packard is only one corporation using prediction markets today. Companies such as Motorola, Intel, Best Buy, Microsoft, Google, and Pfizer all use internal prediction markets to determine which new products will be among their best sellers, as well as other initiatives.

So, can scientific-polling be manipulated to include the accuracy of prediction markets? Perhaps Dr. Rothschild is on to something. Prediction markets could be used to guide companies, rather than making decisions on gut-feelings or educated guesses. In the United States, prediction market sites have had legal difficulties with The Commodity Futures Trading Commission. Some consider prediction markets a form of gambling since they allow people to earn money by predicting the outcome of future events. That said, whether or not people decide to bet on a prediction has no impact on that prediction’s accuracy. Prediction markets can still be a valuable prognostication tool, regardless of how they are regulated.

Whether Bernie Sanders will be the Democratic Nominee for President of United States in 2016 is a political question that is sure to interest many corporations. More importantly, it is the type of question that can be easily applied to a predictive market. Assuming the prediction is Sanders will be the nominee, the derivative would be redeemable for a certain amount if Bernie wins a number of Democratic primaries but worthless if he is not chosen as the eventual nominee. Up until the convention on July 25th, this derivative or bet can be traded on a prediction market and, as such, will command a fluctuating market price. In this case, Sanders’ performance in the primaries leading up to the convention affects the derivative or how much you would win for betting that, yes, he will be the eventual nominee at a given point in time. If Sanders is doing well in the primaries, betting on the prediction that he will be the nominee is safer and therefore pays less, while a poor primary performance would make that prediction riskier with a higher derivative and larger payout.

American politics actually has a rich history of prediction markets. By the 1870s political predictions had evolved into a semi-formal financial market. The Economist suggests that trading volume began to approach that of actual shares: in 1916 $10 million ($218 million in today’s prices) was wagered on the photo-finish presidential race between Wilson and Hughes. While the markets were wrong that year, incorrectly predicting a win for Hughes; they were correct in accurately forecasting the next 11 contests. In an era without reliable polling, the newspapers diligently reported on presidential betting odds, publishing the lines five-days-a-week in the months before an election.

Crackdowns on unauthorized gambling drove political bookmakers into the ground. However, the final nail in the coffin for political prediction markets was reputable, scientific public-opinion polling. Other forms of gambling were also a factor: the Economist highlights that when betting on horses became legal it allowed ‘punters to slake their thirst for action dozens of times a day rather than once every four years, without any risk that a bookie would fail to pay out.’ By the 1940s what was once an eight-figure market place had nearly disappeared.

Today the internet makes it very easy for political junkies to bet on campaigns. Most of these are traditional betting markets where the wager is on the eventual outcome with relatively stable odds. Prediction markets where wagers are placed on a given opinion of what an eventual outcome will be have much less stable odds and probably do exist somewhere online. With corporations beginning to apply prediction markets to situations outside of politics, the potential for prediction markets is endless.

So what happens if prediction markets become a go-to source for forecasting data? With improved predictive capabilities, companies and organizations could fine-tune their marketing and targeting even more. Being able to accurately predict the next trend, the best-selling product, or the next political leader, would have big implications for a number of industries and sectors. Being able to wager on those predictions opens up a whole other can of worms in terms for ethical, legal, and profit lines. That said, if the data surrounding predictive markets continues to be more accurate than traditional polling, they will no doubt experience a resurgence in popularity.

Since crystal balls went out of style, pollsters, marketers, weatherman, and shamans have been looking for the newest way to accurately peek into the future. Maybe today it’s prediction markets? There likely will not be a return to 1870s style ‘what if’ gambling. Prediction markets could become the next big thing. Wanna bet on it?

Photo: John Hoey, Thisonesforruthie