Cashing in on Online Prediction Markets

garzpacho writes "BusinessWeek takes a look at the use of prediction markets to forecast business success. These markets have been taking the form of games online--the Hollywood Stock Exchange, for example, allows players to bet on the success of movies. Hollywood is currently one of the largest consumers of prediction market data, in part because movies' broad appeal leads to a large number of players, but also because the markets have been surprisingly accurate--92% in picking Oscar winners over the last three years. Because of this success, other industries are taking a look; pharmaceutical and tech storage businesses are currently working to set up their own markets."

Comments

92% Oscar winner accuracy?

[enitime]

"...the markets have been surprisingly accurate--92% in picking Oscar winners over the last three years."

Only 92%? I'm sticking with The DaColbert Code.

Well, duh (re: Hollywood)

[truthsearch]

The Oscars are a popularity contest, so of course a prediction market will be accurate. Plus everyone knows if you throw in a mentally challanged main character or uglify a hot actress you're 10 times more likely to win the Oscar.

Re:Well, duh (re: Hollywood)

[TubeSteak]

The Oscars are a popularity contest, so of course a prediction market will be accurate.

I find myself agreeing with that sentiment.

From TFA: "The Hollywood Stock Exchange gives traders just enough information to be able to play the market."

You don't have to worry about risk to predict the success of a movie.

To do predictions for something like pharmaceuticals, you need vastly more technical information... something that pharma companys are loathe to part with (results of unreleased clinical trials for example).

Otherwise, you're just guessing at the risks involved.

Google

[Anonymous Coward]

When Google went IPO, I wanted to join in but had no idea what price it was going to emerge at and didn't want to overbid. Then I found out an exotic derivative was being marketed, betting on Google's IPO price. I watched the price of the derivative over the next few weeks; it stayed fairly stable, and I used its value to choose my price. The end result? I placed several bids at different values, and the lowest value I chose was $4/share over the final IPO price, meaning my entire order was executed (though neither I nor anyone else got the full boat).

Apply this to CS

[neonprimetime]

Maybe we should do this for Source Forge projects? Then we know which ones we shouldn't waste or time on.
Or, we could do this with Linux Distros so we know which ones we should just abandon before wasting all our efforts.

DistroWatch.com

[licamell]

Distrowatch does this... shows the popularity of different distros. It sometimes influences me to try knew distros that manage to make it to the top of the list even if I have never heard of them.

More serious example

[hcdejong]

StrategyPage runs a prediction market on geopolitical events, with a similarly high success percentage.

Re:More serious example

[TopShelf]

MIT's Technology Review magazine used to run a great set of market contests, called Innovation Futures. They had great prizes, of which I won a few (Sony DVD burner, HP Tablet PC, oodles of gift certificates, etc.).

That was important because if you want people to really make these markets work, there has to be some incentive for them to do so.

Terrorism Futures Market

[rothlmar]

If you'll recall, a similar terrorism futures market was planned by DARPA; it fell victim to political pressure, got deep-sixed, and the proponent, John Poindexter, resigned in a cloud. But the truth is that the idea works well. Here's a summary of the controversy: http://www.wired.com/news/politics/0,1283,59818,00 .html.

Re:Surprising?

[andrewman327]

Applying this system to business (especially pharmaceuticals) is a horrible idea. Oscars are selected in part by popularity. Of course the most popular movies will have the most votes. Predicting if Levoxyl can outsell Synthroid in the Pacific Northwest in Q1 2007 is another proposition entirely. I remember that the DoD tried something like this with terrorism but cancelled it because of public outrage.

No real money involved?

[ArcticCelt]

I understand that there is no real money involved. Do someone know a "good or recommended" web site with similar concept where you can cash on your knowledge and instinct?

I just did a search and found two but I am not sure if they are any good:

http://www.intrade.com/ where you can in many political, entertainement and world events

http://www.tradesports.com/ Mostly sports but also a bit of world events

Re:No real money involved?

[Red Flayer]

Do someone know a "good or recommended" web site with similar concept where you can cash on your knowledge and instinct?

Yes, try this one. They don't explain it too well, but you're wagering on whether given companies will meet, surpass, or fail to meet others' expectations of how they'll perform. Sometimes you're even wagering on just how they'll perform, but normally you're wagering against how others think they'll perform -- that is, can you guess better than everyone else?

Re:No real money involved?

[xcaverx]

I've tried Intrade. Tradesports is a division of the same company. You might want to run a few practice transactions on Intrade without putting up any money -- just so you can learn some of the pitfalls:

One is the idea (which I shared for a while) that the majority on any contract position is almost always right -- read the news stories, after all, they all suggest the prediction markets are damn near infallible. But the majority position may be all wrong for most of the life of the contract and then readjust in the space of a few trades, leaving you holding some worthless contracts if you had bought with the market. For example, the Republican gubernatorial hopeful in Virginia was the favorite on Intrade for months, then the market reversed itself very rapidly. Likewise Bush's female Supreme Court nominee was strongly favored to win approval until Arlen Specter voiced his misgivings and her contracts fell from the sky. I guess Intrade racked up wins on both predictions -- but in both cases the markets were wrong until it became dead obvious to everyone that Kaine would win Virginia and Harriet Miers would have to drop out.

The market is not liquid -- usually only a few contracts trade every day. Look at the most popular contracts -- such as will Hillary become the Demo nominee -- just a few thousand contracts outstanding. If you want out of a contract immediately, you may have to drop your asking price markedly below the last traded price.

New information is factored into the pricing with breathtaking speed -- you are going to outsmart this market only by accurately predicting the outcome well in advance of expiration. You can't count on being able to keep pace with the change of pricing, unless you hover over a computer screen and clear all your contracts before you go to bed.

Old field

[tansey]

This is actually a fairly old field of statistics. The story goes that a mathematician at a state fair saw there was a contest to see who could guess the weight of a pile of recently slaughtered beef. Everyone was allowed to guess at it: farmers, house wives, butchers, etc. In the end, the experts (i.e. the butchers) didn't win, and their answer was off by about 10%. However, the total of all the answers, averaged, was off by only .2 lbs.

This sparked the idea that the knowledge of the whole group can lead to better answers than the knowledge of a select group of experts. It's also been shown to be true with things like artificial intelligence and mathematical proof programs.

So it seems like hollywood and the like have just finally realized that the entire group of people can do better at predicting movie success than just some panel of marketing experts.

Re:Old field

[renfrow]

This idea, that while nobody knows the answer, but, everybody knows the answer, was one of the plot devices in the book 'The Shockwave Rider' by John Brunner in 1975. One of my top ten favorite scifi books.

Perception and reality

[EmbeddedJanitor]

which diseases people most wanted a cure for (something you could likely figure out from how many had those diseases to begin with)

Perception and hype are not the same as reality. For instance, TB kills more people than AIDS, yet which is perceived to be more important to treat? There's a lot of discussion and hype about breast cancer, yet prostate cancer is more serious.

Perception drives our irrational thinking.

Re:Surprising?

[eikonoklastes]

Applying this system to business (especially pharmaceuticals) is a horrible idea

Not necessarily. Allowing Joe Q. Public to partake in the market would be a bad idea. Allowing the designers, testers, etc. to (anonymously?) participate is a great idea. Check out The Wisdom of Crowds for a great read.

I remember that the DoD tried something like this with terrorism but cancelled it because of public outrage.

Cancelled before it was ever used. It probably would have worked, too.

Re:Surprising?

[timeOday]

Applying this system to business (especially pharmaceuticals) is a horrible idea.

Uh, that's what the stock market is. I'm not saying it's accurate, but it is how resources (money) are allocated in our global economy. "Betting on futures" for a movie? It's just a little more direct than buying stock in Sony.

The article didn't mention the flu market...

[exp(pi*sqrt(163))]

...discussed here where you can buy stock in your favorite flu virus. Apparently it performs well in predicting flu outbreaks.

The "Delphi Effect"

[farrellj]

This is a well known phenonena, and has been used for a long time now. If you ever read the book "The Shockwave Rider" by John Brunner, and shame on you if you haven't, you will recognize the "Delphi Boards". They were basically a legalized betting system that allowed you to wager on the possiblity of something happening...for example, a cure for cancer. I won't go more into the story, but you can check it out for yourself!

It is technically known as the Delphi Effect, and you can read more about it here:

http://en.wikipedia.org/wiki/Delphi_effect

ttyl
          Farrell