Q&A with leading prediction markets entrepreneur John Delaney
The founder/CEO of Tradesports.com on the basics and what's next


Maybe the prediction markets industry seems new to most of the business world, but John Delaney's already been at it a while. In 1999 he founded the Trade Exchange Network, parent to websites such as tradesports.com and intrade.com and others, which together make Delaney's business the market leader in public, real-money prediction markets (the kind that individuals can participate in, trading in predicted outcomes ranging from sports results to geopolitical crises). Overall, Trade Exchange has some 70,000 registered members worldwide, and processes approximately half a million orders a day, with members trading some 150,000 different contracts over the course of a year.

Delaney, a 30-something, wire-thin Irishman who climbs mountains, races bikes, and manages to come across as both buoyantly charming and ambitiously obsessed, is what more than one industry observer has described as a "true believer." He'll tell anyone who'll listen that his main aim is to grow the prediction markets industry overall--bring the idea to the attention of more people both in business and out. With visibility, he believes, will come converts. And as the pie grows, Trade Exchange will do fine with its piece.

He took some time to speak with us about:

--The basics of prediction markets and how they work.
--Which questions they're good at answering (and which ones they're not).
--What gets in the way of managers/companies using prediction markets.
--Whether or not prediction markets pre-empt management authority when it comes to decision-making.
--And what drove him to launch Trade Exchange in the first place.

FutureMonitor (FM): Description, please--what is a prediction market?

John Delaney:  A prediction market is a very simple mechanism that we can all use to gain a little bit of insight or forecast about the future of certain events.  It’s not a black art, and it’s not magic.  It’s just a tool or technique.

To use it you would formulate a question in a way that would give you a yes or no answer.  For example, Will the revenue growth of your company increase by greater than 100% this year?  You list that as a contract or a stock.  Then, you have participants who will trade that issue and by so trading it, they are aggregating a huge amount of information, and you will have a very precise and concise prediction of the probability of achieving your 100% growth.


FM:   In that example, how would people trade? And who would you want to be trading?
Delaney: The theory and the empirical evidence would suggest that the wider the user group that you have, the greater the predictive power.  If you have people just from your own company, then certain biases appear.  If you have your peers, your competitors, or your clients, and it’s a wider audience, then intuitively, you will get better information and more predictive information.  Then, the empirical backs up that information as well.


FM: How does the actual trading work?
Delaney: Let’s go back to the sample prediction relating to 100% or greater year-to-year growth for a company.  Let’s say, for example, that when that prediction or contract as we like to call them on TradeSports was initially put on the prediction market the first person to predict was the financial controller, though he may not always know best.  So your financial guy comes in and says, “Well, I think growth is probably going to be less than 100%, but I’m not completely sure of that.  I think the probability is about 30% that we'll hit that number.”  He has already put up one price.  He would go to a very simple user interface and enter that.

Now, let's say that one of your competitors is about to lose a big client and you are the obvious person for that client to go to, so the competitor says, “Golly, I think these guys will hit it.  I’m not completely sure either, but I think they’re about 30% sure of hitting [that 100% annual growth target].”  So when you have two people who will reconcile on a price, then you have your first probability through the market dynamic where equilibrium is derived.  You have someone willing to buy an issue, you have someone who is willing to sell an issue, and they have agreed on a price.  The first point or first probability derived by the prediction market is, “My company has a 30% chance of achieving 100% year-on-year growth.”


FM: Describe the exact mechanics of the trade again. What is our hypothetical financial officer actually going to do to create that opportunity for the competitor to buy
Delaney: Let’s not overcomplicate or intellectualize it.  It’s very simple.  He goes onto a web-based user interface…


FM: …an exchange like yours or others.
Delaney: ...And he reads what that contract is.  It’s 100% year-on-year growth.  If you think that proposition is true--meaning yes, the company will achieve that growth--you buy, and if you think it’s no, you sell.  As the market's first participant on this contract you see there's no price yet and what you are saying to yourself is, “I believe this will happen, and I’m 30% sure.”  Eventually you type that in.  You don’t type it in longhand.  There is a simple way to do it.  That’s what he does.  He has initially put up a marker or a "bid" in the market.  “I will buy this contract," which is like saying, "I am predicting with 30% certainty that this will happen.”

Then, you have somebody else coming on board and they say, “I think that is fair value,” or, “I think it’s a little bit more or a little bit less.”  It’s just like when the price of  Cisco or any other security moves on a day; it’s just the way the market operates to find the equilibrium.  You have somebody willing to buy a contract on an issue and you have somebody willing to sell an issue.  The price is derived as a result of that.  Explaining how it works is like trying to explain how to drive a car--it can seem a little bit tedious. It’s far easier to just go to the user interface [www.tradesports.com or www.newsfutures.com (where you can trade without real money)] and actually see how we do it.

The important thing to understand is that what is being traded is the probability of something happening. When you start trading, the mechanics quickly become intuitive.


FM: When you describe prediction markets to people who haven’t seen them before, what are the things that cause the most confusion for them?  What is it that you most often have to try to explain?
Delaney: Well that is exactly the right word: “explain."  Think of the car-driving analogy.  It’s far easier to show people how this works than to explain it in abstract terms.  Other questions come up, of course.  There is the issue of credibility: Do these actually work?  To what benefit?


FM: Does that tend to be the first question people ask?
Delaney: No.  It’s more like, “We are intrigued by this, and we believe that teams work very well.  If you put five people together, the whole of their collective effort is a better result than the sum of their individual efforts would have been."  Implicitly people understand that.  When you explain that prediction markets are an extension of that, people get it--they are intrigued. But then there is the question, “How does it actually work.”  You go through the mechanics, and the empirical evidence and user interface of our and other platforms, and then they say, “Alright, I understand that, but how is it going to extract those little jewels of wisdom.”  It does take a little bit of time for people to understand that part.  Then, once they understand that, they say, “All right, we are interested.”

Then, when they see the potential of these markets, they say, “Golly.  Are these going to replace some of our decision-making authority?”  That is one of the issues that needs to be addressed very forcefully.  You don’t replace decision-making.  Prediction markets supplement the tools that we are all using, have been using, and will continue to use.  The prediction market should not be seen as a trap, just an additional tool.


FM: Can you tell us how you, personally, found your way to prediction markets and to creating a company that provides them? Seeing your path might help a novice understand the logic of prediction markets in another way.
Delaney: I worked in the investment industry for about a dozen years in Dublin in various different roles.  I remember the impeachment of Clinton, and I remember watching the Dow.  I was thinking “There are so many things that people are interested in trading that have linkages and knock-on effects."  It became very apparent to me that if there was a marketplace where people could trade all of these issues--issues that they were often very passionate about, and that were very important to them--then they would.


FM: Had you heard of prediction markets or decision markets before you were seeing those linkages between the markets we’re all familiar with and political events like Clinton’s impeachment?
Delaney: No, not the concept of prediction markets in that exact sense--I hadn’t heard of the term. But a stock exchange is a prediction market. When the price of Cisco or any other company is derived, you are essentially predicting the implicit future cash flow rather than applying your discount factor. There are prediction markets of a sort all around us, but the new kind of markets that we're talking about here are dealing with new concepts and new kinds of contracts.


FM: So you had your epiphany and realized what markets might be able to predict in a more direct way.  What did you do next?
Delaney: The prediction market is a literal marketplace.  You need the infrastructure to support that market.  You need a software platform, technology, and web presence in a community.  We started off by creating the technology and building a community.


FM: You went straight from your epiphany to launching a company.
Delaney: Yes, with a team of people and the support of in 1999.


FM: At the time, how much was known about prediction markets? What was the state of the art, in terms of their science and application?
Delaney: There wasn’t nearly as much academic research as there is now, though there was a reasonable amount.  People like Robin Hanson, and Justin Wolfers for example, would have been involved in these things from before we were, but we didn’t approach it from an academic perspective.
That wasn’t our background or grounding.  We came at it from the financial market perspective.


FM: Was anybody else at that time trying to actually build markets rather than just do research and think about the theory of it?  Were there other companies?
Delaney: The Iowa Electronic Markets exchange, for one.  That has been around since the early 90’s, I think.  It’s run by the University of Iowa, and is essentially focused on politics.  It has had fantastic results by comparison to opinion polls; the accuracy of the predictions on political issues coming from the Iowa market have been far better than the opinion polls relating to the same issue.


FM: When your company began building its markets for public trading, what were the toughest problems to solve?
Delaney: We knew we wanted to be a very global business and that the technology would have to support a very large user base and a very large number of propositions or contracts.  The number of contracts, the number of users, the activity, and the volume in our marketplace would be far higher than any other prediction market to date--so there were significant challenges as we created our technology. Then there was the whole legal and regulatory grayness.  I guess what we were doing, in a lot of cases, may be ahead of laws and regulations in any of the 192 countries that we were accessible from.  There was a lot of due diligence that we had to do there, and that continues on an ongoing basis when the laws change and cases change.  Those were some of the significant challenges.

Once we launched, the key challenge--and this is a challenge for anybody in this space be it at public or private--is participation. There is a correlation between the predictive power of a market and participation so you need to get people into those markets.  It’s called liquidity; you need liquidity in your market.


FM: How did you go about trying to promote that?
Delaney: That’s a trade secret!  But OK, the first thing is that you list issues or contracts that are relevant and interesting to people.  Then, you create a community of people that will trade those.  There is very little point in listing a contract that no one is interested in.  It just won’t trade.  The other important thing is to have somebody make the initial market.  Earlier we used an example of year-on-year growth for a company, for instance.  Perhaps the first thing is to have a number of your financial guys come up with the initial market.  "I think it will happen with 30% or 35% probability." You need somebody to dip their toe in first.  Those would be some of the key challenges once you launch your market.


FM: Have there been any big revisions to your model since starting the company?
Delaney: No. When I go back and look at our business plan from day one, what we documented is what we are executing now and what we will continue to execute.  It was to list contracts and issues that people were passionate about and that people would trade.  Our business is a simple business; we list contracts and we charge fees.


FM: In his book "The Wisdom of Crowds," James Surowiecki wrote with some surprise about how little prediction markets were being used as forecasting or decision-making tools inside companies. He noted that managerial decision-making is all about looking at strategic alternatives, valuing their probabilities of success, and then acting on those probabilities--yet while there is nothing more tailor-made to provide the kind of information a company needs to make those decisions than prediction markets, they don’t use them.  The book was published in 2004, of course, and things have changed over the last couple of years, but I’m really curious to know why you think companies haven’t made better use of prediction markets.  God knows they do surveys.  They do focus groups.  Their decisions are constantly driven by market research. Yet they haven’t used these.
Delaney: Well, to say that they haven’t used them is probably a bit of a reach. Companies ARE using them--more and more.  Why hasn’t the traction and the acceleration been higher than it has been to date? Let's go back to some of the challenges:  for one thing, there are costs to running prediction markets and people are very conscious of costs--rightfully so.

People are also very conscious, and I think this may even be a greater issue than cost, of the potential pressure that they see prediction markets placing on them as middle managers and even senior executives.  For example, let’s say a dilemma would be that your prediction market gives you some information that you, as the person responsible for the business area, aren't happy with.  Do you really want that?  You may be fearful that a market may replace you to some extent, undermine your authority to some extent, or give you information that you might really rather not have.


FM: But that’s no different from when you get unwanted results from survey research.
Delaney: Yes, but I suppose arguably those kinds of research results haven’t shown themselves to be as accurate as prediction markets.  Opinion polls, and that kind of thing, have been going on for a long, long time.  They are engrained in businesses right now.  Hundreds of millions of dollars are spent every year on them, and we now know that there is a better way than just opinion polls.  I’m not saying prediction markets will replace opinion polls, but I am saying that these are better in a lot of instances.  They certainly supplement in many more.


FM: So cost is one obstacle to use of prediction markets--although there are plenty of costs attached to survey research, too, but people understand those costs and have already budgeted for them.  Another obstacle is that they potentially pose a threat to the decision-making autonomy of the manager when results come back from a prediction market that are counter to the way that the manager wanted to go.  Are there other obstacles?
Delaney: Yes, there is the concept itself.  How in blazes do these actually work?  Is this Nostradamus coming into the boardroom?  What kind of black art is this?  Yet it really isn’t that at all.  It’s a very simple, intuitive concept that is very analogous to how teams operate.  You put a group of people together with differing opinions and skill sets.  By adding that group together you will, in the main, get a better result than the five separate individual efforts would have yielded.  It’s very similar to that, but there is an educational process that people need to go through.

FM: Please elaborate on something you mentioned earlier: literally, what is it that prediction markets aggregate?  What are they collecting or what are they tapping into that makes them more accurate as predictive tools than some polls?  What are they extracting from the participants that helps a manager get better information?
Delaney: You asked and answered the question.  Prediction markets are just extracting the information and using a mechanism that has shown to be better than an opinion poll.  Let me give you an example why.  Imagine a typical opinion poll, perhaps where there is no financial or other valuable reward for the person being asked a question. Say you're stopped on the street and asked, “Who do you think will win the next election?"  You might say who you would like to win it, or you might say the first thing that comes to your mind because you are rushing home to your dinner.  Let’s say the same question is posed to you and you have to back that up with some of your own personal capital.  Let’s say the same question is put to you and there is a reward if you get the question right.  Then, instead of saying who you might like to win, or rushing home to get your dinner, you will pause for a second, probably, and say, “Hmm.  That’s who I’d like to win, but putting that aside for a second, who do I think will win?”  There is a difference there.  Opinion polls are very successful, but prediction markets are different.


FM: We could argue that both opinion polls and prediction markets are about extracting local information, or private information from many, many different sources.  In some cases extracting from us even information that we don’t really know we possess and couldn’t articulate.  That unarticulated information contributes to our decision about whom we like to win or who we think will win.  But you’d argue that opinion polls involve a different bias than prediction markets do. Is that why prediction markets are likely to be more accurate?
Delaney: Let’s simplify it.  It’s human nature.  “If I get this right, this is what I get.”  I’m not aware of many opinion polls that phone you back or stop you in the street, and then six months later send you a check or a new iPod.


FM: So you are simply saying, “I’m going to do a better job if I get a reward.”
Delaney: Reward does motivate a lot of people.


FM: Back to managers and THEIR motivation for a minute.... You and I heard Surowiecki tell an audience that he thinks the time may come when you can in fact run a company by making decisions determined utterly by the outcomes of prediction markets.  If you are smart enough to create the right contracts--an art and science in itself, admittedly--then the crowd through the market can tell you, the manager, which is the better way to go.
Delaney: Would you invest in a company where the board is a piece of technology


FM: No. But I've been wrong before.
Delaney: Well, I don't think that will ever happen.  I think that prediction markets will supplement management thinking very significantly and a lot more than they have to date, because they will be more widespread, but I think that, at some stage, people do have to make decisions.


FM: You’ve been building Trade Exchange Network for five or six years.  Have you ever used prediction markets to help drive the strategic direction of your own business?
Delaney: Yes, we have in an informal way. To forecast our sales growth, for example.


FM: How have the results from that market changed, if at all, the way you’ve managed the business?
Delaney: A couple of the things that using that kind of market has done for us is that, first of all, it’s given voice to people that may not necessarily have a voice.  That is one great use of prediction markets that hasn’t really been spoken about a lot.  You have people, be it in large organizations or even small organizations like ours, which may not necessarily feel that they have input into some of the major issues effecting the organization.  Prediction markets give them a voice.  People who may not necessarily want to speak at the board meeting, management meeting, or company meeting, now have a voice channel.

The other thing is that when you have three or four people who think that you are going to exceed 165% year-on-year growth, this year for example, and then the prediction market comes in saying that achieving that growth is a 90% certainty, it focuses the mind.  So there are various benefits in addition to the discrete point of information.


FM: Have there been ways that you’ve thought about using markets in your own management that have proven too difficult or not as useful as you might have wished
Delaney: Yes, there are limitations to prediction markets, and I suppose that when you think about the limitations, you should perhaps think about the issues about formulating the question.  When you have wicked problems--problems that can not be very easily defined within a yes/no future--using prediction markets can be impossible. Even a standard-type organizational question, such as, "What is the right benefit package to implement for the organization?" can be difficult to form as a market contract.  So there are things that you would love to have more information on, but there are limitations to prediction markets.


FM: What are the kinds of things, either on a micro level inside a company or on a macro level, that prediction markets are especially good at doing?
Delaney: Addressing things where you can, first of all, frame the question or the subject essentially in a yes or no; predicting, for example, the winning contestant in a contest.  Prediction markets have shown to be quite good at that.  There are various empirical studies that show that prediction markets have been very good at predicting weather, for example.  Obviously, they are very good in a financial context in predicting the value of a stock or currency.  In a whole host of areas, they have worked very well in the past.


FM: OK, to recap the essentials of a prediction market that's going to work well, then: We know that we need to have a question that can be framed in yes or no terms.  And we need to have adequate participation--a certain liquidity, as you would say, so that we have enough information flowing into the market.  What kind of people do we need participating to make a market most effective?
Delaney: Broader rather than narrower, and this is a little bit counter-intuitive in that you would say, “Well, obviously the best people to involve in this market are the people who are most knowledgeable about it.”  At first blush, that may seem counter-intuitive, but then when you think about it a little bit more, maybe it’s not.  Sometimes, if you have a group of experts, whether you like it or not, or they like it or not, the whole group consensus dynamic is there to some degree.  When you have people who aren’t necessarily expert, or aren’t very informed in that area, they bring a freshness to the questions.


FM: Last question.  It’s about the prediction markets industry in general.  How soon do you think prediction markets are going to be accepted by the general public as the equivalent of opinion polls? And how quickly do you expect them to be accepted by businesses, privately, and widely applied as a management tool?
Delaney: We ran a little experiment on one of our sites where we have a number of industry-specific contracts.  One of the contracts was, “Will we see major year on year growth [in the prediction markets industry]?”  meaning growth of at least 100%. The market is saying that the industry prediction has an 81% chance or thereabouts hitting that growth number in 2006.


FM: Is that undifferentiated between public and private prediction markets?
Delaney: We are just talking about the industry.  Myself, I’m a buyer of that contract.  I think the probability of major year on year growth is higher than that.  I suppose I have some inside information, because I see where our volumes are going, but then other people obviously know other things externally and they are selling that market over what was advertised.