Darjeelink - Alexis Perrier

Is BluBet another variation of prediction markets ?

BluBet is a web site where one can bet on issues ranging from political, people, science, to personal ones. ( will I get drunk this saturday ?). It’s a fun and easy to use web site. It’s easy to create your own bet, to bet, to leave comments, …
A pity many bets are not really up to it, a bit too mumbo jumbo for my taste.
I also regret the constant push for an access to my gmail account or some other personal data (gender, zipcode, …). Makes you wonder what do they want to do with the data they gather this way.

In their press release, BluBet states that :

“BluBet is fueled by the best ideas of user-generated content and predictive markets. Instead of building a complicated prediction market that is hard to understand for the average user, BluBet took a simpler approach. By layering an entertaining gambling game on top of a prediction market,BluBet has built a service that is fun to use and easy to understand for the casual participant.”

I agree with the simplicity and fun parts but the prediction market is less obvious.

BluBet’s bets boils down to a mix between polling and straightforward pari-mutuel. It’s quite similar as betting on horse races, sports games, … with play money.

What are prediction markets exactly ?

A prediction market is a market, in the trading meaning of the word, where people can buy and sell shares of a contract. Prediction markets, also called information markets, are used to aggregate information from people in order to forecast the probability of some event outcomes.
In many cases, prediction markets forecasts outperforms experts or opinion polls.
Do read the wikipedia entry on Prediction markets, if you want to get up to speed on the subject.

Most of the time, such contracts will consist of an uncertain future event that has a finite number of possible outcomes at a given date.
For instance :
- Will Barack Obama be the democrat candidate in 2008 ? (here and here)
- Will the foot and mouth recent break out in the UK be resolved before the 1st of September ? (in fact ,it already has)
The outcome of the event must rely on an independent third party data that acts as a reference upon which all the traders can agree to.

At closing time, the trader is rewarded depending on his / her portfolio according to different predefined payoffs schemes.

It as been shown by Justin Wolfers and Eric Zitzewitz in their article “Interpreting prediction market prices as probabilities” (pdf) that the price of the share of each event possible outcome can be interpreted as an estimation of the probability of the outcome actually happening.
The raison d’être of the prediction market is to give an estimation of the probability of the event outcomes.

Another type of contract, consists in items such as films, actors or Internet related technologies.
One can buy shares of the next blockbuster or celebrity or some techno related subject : Open Office, microformats, Skype, …
The value of the contracts shares is interpreted as an indicator of the item’s future success in the box office or as a widespread technology.

During the life of the contract, it’s always possible to see the evolution of the share’s value. And share values change over time as a function of the trading.

Prediction markets have been used in many different fields : to predict elections, video game success, product market shares and as decision tools in policy decisions, company strategies, risk management, …

Designing prediction markets is a precise and complex task that must try to answer to several questions such as :

• How to calibrate for small markets or for small probabilities ? (How many shares ? At what price ? on opening day)
• How to attract sufficient traders for the market to start ?
• How to get them to trade during the life of the contract ?
• How to write items of interest that can be broken down into events that can be verified ?
• How to limit market manipulation ?
• How to incite traders to share information ?

The interpretation of the market can also be tricky. When for instance an event with a low probability estimation happens nonetheless. Does it mean that the prediction market was wrong ?

To simplify thing, what makes a prediction market efficient as a forecasting tools is the whole process of the traders information flow being translated into a serie of share exchanges over time.

In prediction markets, results can be interpreted as forecasts because of the whole supply and demand process that is intrinsic to markets. It has been shown by J. Wolfers and E. Zitewitz (same paper) that market values can be interpreted as probabilities because the prediction market tends to minimize the gap between supply and demand.
When there is no supply and demand of a given contract, there is no prediction attached to the result.
Things work in a prediction market because traders believe that a given event will result in a given outcome. When there is no outcome, there cannot be any prediction.
Belief is not sufficient to create a prediction market. It works for information aggregation but not for prediction.

And that’s the problem with BluBets saying that their game is built upon prediction markets.
Either the bet statement is not related to an event ( Who was the best James Bond?)
or the event cannot be verified : Will BluBet ever make money?
or there is no publicly available info on which to base one’s bet ; Will Kiran buy a Scooter by the end of August?
And in any case, it’s not possible to know the current betting ratio nor it’s evolution since the beginning of the bet.

Bets resolution on BluBets are based on the ratio between the number of bets for one choice and bets bets for the other choice(s). The choice that gets the most bets wins.

In BluBet, there is no prediction as there are no markets.