A Few Words on Prediction Markets

A recent conversation with a friend made me made realize how easy it is to be misinterpreted, especially on more complex issues.

Let me clarify my views on the so-called prediction markets.

In the past, I have very strongly dissed some people’s concept of these markets and their ability to “predict” the future. The primary reason for my view is the unfortunate tendency for some people to place way to much faith in the crowd to know the unknowable.

I have yet to see a reasonable explanation as to why IEM missed Dean’s collapse in the primaries; a commentor notes that the American Idol contract on Tradesports had Bice 90% favored in the final minutes — but lost.

That said, I rely on many forms of market action to provide color to my expectations. Trend Following is one obvious area. The Fed futures are another. All of technical analysis requires interpreting price and volume changes. I would be lying if I said I didn’t think the Bond market was saying something.

So when I dis prediction markets — especially about their lack of liquidity and small trading pool — its a slight. It reflects my general disdain for the efficient market hypothesis, amongst other flawed theoretical concepts. When it comes to explaining market behavior, I’m more of a Chaos-theory man myself.

The difference between the Bond market and say, the Iowa Electronic Markets, NewsFutures and the Hollywood Stock Exchange or Trade Sports is in the size, scale, and liquidity. I am agnostic about these smaller markets, but I do not discount the possibility that any mechanism can potentially provide some insight into future market behaviors.

On a related note, I am reading (quite skeptically) the Wisdom of Crowds — something quite related to the concept of prediction markets. Its a book that I find slow going because of the continual fisking it requires. While its a pleasure to read Surowiecki‘s prose, on most of the market related items I’ve gone through there are readily identified logical flaws and unsupported assumptions. (More on this in the future).

As to the Prediction markets value, I remain an interested skeptic.
>

If you want  to learn more about Prediction Markets, than the meta page you want to see is Chris Masse’s prediction market vortal. There’s a ton of data and links there to get you started . . .

>

What's been said:

Discussions found on the web:
  1. JRivera commented on May 27

    For those of you that still doubt the predictive power of a large pool informed bets, look no further than this simple and entertaining read. If you like the article you will LOVE the book.

  2. Barry Ritholtz commented on May 27

    My point isn’t that Prediction Markets are utterly worthless — Just that they are far from perfect, and often far less prescient than many people believe.

    Incidentally, the missing Submarine example mentioned above (contextmag link) is in The Wisdom of Crowds book

  3. M Knesevitch commented on May 27

    Prediction markets aren’t perfect markets. After-all if we had perfect markets there would be no capital markets – there would be no reason to take risk.

    In defense of Tradesports, remember the exchange correctly predicted the voting outcome of every state plus DC AND 33 out of 34 senate races in Election 2004 as well as who would be the next Pope in arugable the most secretive selection process known to man.

  4. Barry Ritholtz commented on May 27

    Thats exactly the sort of comment that is so frustrating when it comes to these markets.

    Define your terms: How do you measure when they “predicted” the results — a year before? week? a day?
    No one has yet to disprove these predictions aren’t merely tracking public polls. Indeed, they ARE a form of public polling.

    Here’s a prediction, via May 27 2005: “The US stock market pulls back to support over the next 2 weeks, and then has a big rally until November/December — its stronger than many expect, with Nasdaq breaking 2500 and the Dow going to 11,800. But thats the peak, and the Dow drops 20- 30% and Nadaq 40-50% over the next 2 years.”

    Thats two years worth of forecasts, very specific, multiple time frames IN ADVANCE.

    I don’t place a lot of stock in predictions the day before an election.

    AND, let me hasten to point out that the prediction markets had Bush losing as of 5pm on election day —
    (If you want me to move you into the ugliest category “mental midgets to ignore for all eternity,” then just trot out the bullshit that it was Soros).

    So I ask the question here and now — when where these oh-so accurate predictions made?

  5. Financial Rounds commented on May 27

    Prediction Markets -Another View (via The Big Pic

    As I’ve written previously, I’m a big fan of prediction markets (both as a classroom tool and as a means of extracting information from dispersed groups). However, Barry Ritholz at The Big Bicture has a less favorable view:

  6. Chris. F. Masse commented on May 27

    Hello Barry Riholtz,

    You’re asking the good question and I wish one of the prediction markets scholars will come here and respond to your question.

    I would make a micro-point.

    The week preceding the 2004 US presidential election, all the MSM commentators were singing the Joe Trippi-inspired song:
    – “The polls giving Bush slightly ahead on average were to be discounted because cellular phone (and other ADSL or cable Skype-type) users were discounted —thus diminishing the number of young, urban, Democratic voters polled by traditional pollsters.”

    The bottom line of the liberal-centrist-antiwar crowd was:
    – “Forget the poll. Look at the early voting poll lined-ups in Florida. This is proof that the young generation comes by the barrel. On Tuesday night, we will have a new president.”

    Barry, do you remember the feeling at the time????

    And what were saying IEM, TradeSports/InTrade and BetFair?? They were all saying Bush is ahead —a 52%–57% range from memory. Which was indeed a slim margin (and IEM’s Tom Rietz may have had a point if it weren’t for his poor choice of vocabulary and gross exaggeration.)

    So, IEM, Trade Exchange Network and BetFair were positive on Bush while a competing institution, the MSM commentariat, were negative on him.

    (And I won’t put up the infamous NewsWeek poll.)

    That’s all the point of the prediction markets. They JUST aggregate information from those in the crowd willing to put up their money where their mouth is. They tell it candidly and brutally, while the liberal-centrist columnists would tell it slant. They should be judged, and that’s the message from scholars like Robin Hanson, against their competitors (the polls, the mathematical models, and the media commentariat).

    A last parting shot, regarding Donald Luskin. He broke the news on the aberrations that occurred on the Bush futures markets. He proposed an interpretation for these strange facts —namely someone inspired by the Soros philosophy wanting to break the Bush momentum and create artificially a counter-buzz.

    It is indeed only a hypothesis (or a theory, depending on one’s vocabulary). Could be Donald Luskin was right. Could be he was damn wrong. Now, where is the competing, PUBLISHED hypothesis?

    L’art est difficile mais la critique est aisée.
    Art is difficult but critique is easy.

    With respect,

    Chris. F. Masse

    Disclosure: I’m a French libertarian, and an anti-Iraq war.

  7. Mike Linksvayer commented on May 27

    Regarding “So I ask the question here and now — when where these oh-so accurate predictions made?” see Historical Presidential Betting Markets.

    The point is not that markets make perfect predictions or that EMH is required (I also think that markets are inherently chaotic), rather that markets are the best available means of aggregating information, and that includes information about the predictions.

    You’re right that many prediction markets are too thin or badly structured. Buyer beware, but I don’t think those are fundamental problems with prediction markets.

  8. Jorge Naya commented on May 30

    A useful adjunct to “Wisdom of Crowds” might be “Extraordinarily Popular Delusions and the Madness of Crowds.

  9. alex commented on Jun 1

    The difference between the Bond market and say, the Iowa Electronic Markets, NewsFutures and the Hollywood Stock Exchange or Trade Sports is in the size, scale, and liquidity.

    No. The difference rests in the character of the underlying market.

    Speculative “event” futures unlike bond futures aren’t used to hedge vital borrowing and lending, ie to hedge bilateral [non-speculative] industrial or commercial risk. Oil futures hedge oil production and consumption. Bond futures hedge forward lending and borrowing risk. The knowledge contained by such instruments reveals price signals, helping businesses manage the unknowable and deploy capital efficiently. It’s not meant to be fortune telling.

Read this next.

Posted Under