Belated post... I've moved!

As you can see from the posts below, I haven't blogged here in quite some time.

Please visit Mercury's Blog at http://blog.mercury-rac.com to see my current writing on prediction markets. Thanks!

On prediction markets

"Chris Masse is a mean-spirited snob who obsesses with bloggers' prediction markets, which is a concept he's totally incapable of explaining clearly..."

Well, I don't believe that statement but he DID dare someone to write it, and it's hard to resist a snarky dare. Now to the main point of this post:

Over at Chris F. Masse's blog, he's been discussing the role of certain start-up prediction markets, specifically CrowdIQ and Inkling. For the most part, he's been fairly negative towards them as he doesn't see much useful purpose in how they open their systems for most anyone to create a market. Specifically, he writes with dripping sarcasm: "Inkling Markets and Crowd IQ have persuaded themselves that user-created prediction markets are going to be a worldwide phenomenal success, to which the public adoption of the telephone in the last century will compare palely."

In some ways, I agree with him. User-created markets tend to be pretty poor. There are simply too many people acting as market makers that just don't understand how prediction markets should work. Perhaps more critically they are asking questions poorly. (I believe there was one market on Inkling for who would win a particular World Cup match, but there was no trade option for a draw!) Some markets makers also don't understand how to properly pay out a contract, which has led to some problems according to some user forums.

However, I think he's missing a bit of a bigger picture on why these companies exist. Right now they are still developing their software. By throwing their doors open wide and letting a variety of people create markets and trade, they get a crash-course in what works and what doesn't in their software, and more importantly what users value and what they don't value. To me, this is the reason why they are marketing their "user-defined" prediction markets so heavily. They're doing exactly what Guy Kawasaki recommends innovators do: "Ship, then test," "Don't worry, be crappy," "Churn, baby, churn," and "Don't let the bozos grind you down."

To imply (as Chris Masse has) that this define their entire business model is incorrect. To me, the user-defined markets are the "loss-leader" that get people interested in prediction markets, and serve as a testbed for their software. Clearly where they think they're going to make money is from corporate prediction markets. (CrowdIQ mentions this purpose, and Inkling is marketing its use on their front page.) While I'm not privvy to any of their business plans or strategies, this would clearly make the most sense.

I would love it to get even more companies out there developing prediction markets. The more, the merrier! Let everyone fight it out in the marketplace to see what users want, what they're willing to pay for, and what exactly is the value of a prediction market to the people that are willing to pay for it. The field is in its infancy; there is a lot left to be developed.

In response to his question about blogger-created prediction markets, that probably is the way that small-scale prediction markets will get sufficient volume to succeed. Bloggers have a specific focus (ie, politics, technology, etc.) and if they feed their readers to a prediction market, that market would have a much better chance of succeeding because of the user's interest level in that focus area. Specifically, blogs like Slashdot or Guy Kawasaki's blog could drive some interesting markets in their field. (Though I would suspect that the diversity of some blogs could limit their success; for example, a Democratic or Republican blog that creates a prediction market for an election could potentially be significantly biased.)

Finally, for full disclosure's sake, I am doing my graduate school project with Inkling. Please see my previous posts below for details, but my research will hopefully investigate how small prediction markets can get and still provide valid results, as well as investigate the types of trading behaviours that make a prediction market successful.

Cheers,
Jed

Research project in motion

I've discussed this briefly on the Prediction Markets Google Group, but I am currently conducting some primary research in small-scale prediction markets. Specifically, I'm looking to determine how accurate the results of prediction markets can be, and what behaviours cause these results.

To be specific, I am running prediction markets in a very niche sport: rowing. Why? Well, I'm on a tight timeframe and this fits in well with my interests and fits my timeframe available to complete my paper perfectly. I expect that some people may find fault in my methods since the markets are sports related, but I think I've minimised any distinction here. First, you'll be hard-pressed to find a bookmaker that publishes odds, unlike football, horse-racing, etc., so the same kind of widely published betting information doesn't exist. Second, the competitions are races so there aren't issues with teams fouling or otherwise interfering with each other. Third, there is near-zero media coverage outside of someone who participates in the sport.

Here are some of the issues I expect to address:
* How accurate are the groups predictions?
* How accurate are predictions relative to the number of traders in the market?
* What types of behaviours do traders exhibit?
* How do price manipulations affect small markets?
* How do behaviours influence the success of a market? How does this relate to the number of traders in a market?
* (potentially) How accurate are the groups predictions relative to expert predictions?

By the middle of July, I expect to have completed in excess of 60 individual markets, each with anywhere from 2 to 15 stocks. This should be sufficient data to make some interesting observations. Most stocks trade in the $0-$5 range, though a few get up to the $60-80 range.

I am doing this research with Inkling Markets. In my anecdotal working with amateur traders, I've found that most people are not fluent in trading techniques, and the user interface they have developed is one of the best I've seen. That, and the instant liquidty makes much more intuitive sense for novice traders.

This is the reason it has been quite some time since I've posted here; the project has taken up much more time than I anticipated! (And for that matter, continues to take up more time than I anticipated.) If anything, the only problem I have is that my marketing of these markets was perhaps a little too good, and I'm getting more people than I ever thought I would get participating. This result is still far better than the reverse, however.

I will publish preliminary results here as I compile them.

Cheers.

Resources

Just wanted to let you know that I've been slowly updating the left frame on this page. Essentially, I've added some of the more interesting prediction markets and blogs. Chris F. Masse still has the biggest list of every market/exchange and blog on the topic, but I'm not as interested in creating an encyclopedia as him!

Next up is creating an academic paper "warehouse." It'll be a page that links to all the important and interesting papers that have been written on prediction markets. I only had enough patience today to get through about a quarter of my list. With my recent stress fracture I'll hopefully be able to take care of these soon.

Cheers.

Election Markets

As a politics junkie, I've been keeping up to date with Tradesport's markets for each of the US Senate seats up for grabs this election cycle.  Here's some (sort-of) interesting things I've found:

* Out of 33 Senate seats up for election, only 9 are actually competitive.  (I define competitive as less than a 50-point spread between GOP/Democrats, ie., a 75-25 race isn't competitive but a 68-32 race is.)

* Of those nine seats, six are held by the Republicans and 3 by the Democrats.

* Based on the current market values, PA will be lost by the GOP, but TN will be held.  These two are at the extremes of the competitive spread I mentioned.

* The remaining seven races are tight (comparatively).  Current trends have each party keeping control of the seats they already control, so little would change.  However, MT and OH are particularly tight and easily a toss-up.

* Essentially, the markets currently say that Democrats will gain ground on the Republicans toward control of the Senate (anywhere from 54-46 to 51-49), but would need a minor miracle to actually take control.  (A slightly less minor miracle would cause a 50-50 split and likely result in split committees but important votes going to Republicans with the VP Cheney tie-breaker.)

** BIG DISCLAIMER: I'm somewhat violating my own personal rule here about expectations and probabilities.  I'm calling the results as if the election were today.  The markets currently take into account that election day is several months off, and that's why so many races are nearly 50-50.  Not only that, but a 60% probability that one party wins a seat also means a 40% probability that the other party takes the seat.  Just like the markets thought Brokeback Mountain would win for best picture, they weren't right.  These markets are calibrated, not perfect.

Finally, you can play the home game at inklingmarkets.com... I understand they're adjusting their market-maker to reduce the extreme volatility that some people have seen in their trading.

Issues with betas...

I want to write a little more on Inkling and CrowdIQ now that I've
had a chance to play with them for a little while.

First of all, CrowdIQ is now officially beta. The various markets
are populated with contracts, and other contracts are in auction. I
have a number of comments on this site:

* As a trader you have quite a bit of control over what you're
doing. There are options available for "Market Buy" and "Market
Sell" as well as Limit Buys and Sells and finally Market Short-
Sells. This is certainly not an algorithm-based market (such as
Inkling)
* A number of the markets have contests associated with them, so
traders have incentives. (Amazon.com gift certificates and the
like). Hopefully that will help "thicken" the market.
* The user interface on CrowdIQ still feels a bit unwieldy. Perhaps
part of that stems from the fact that the main groupings of markets
are at the bottom of the page and sometimes I need to scroll down to
see them. I'd also like to see a "New Markets" and/or "Most Active
Markets" listing.
* I'm also not a huge fan of the different currency for every market,
though I understand the reasoning. A person's performance in order
to win incentives is based on that particular market alone.
* I think it is brilliant how CrowdIQ has made it so easy to create a
new market and devise contracts within that market. I created the
BirdFlu market myself. There is a lot of control for a user to
specify the number of contracts available, shares available, how long
a contract is in "auction" status before you can start trading on
contracts, etc. I confess that I created the BirdFlu market and
contracts with little thought to all of that (I was interested in the
user interface at the time since the site wasn't even active) and am
interested to see how it all turns out.

Overall, I really love the flexibility of the CrowdIQ site, and hope
they can work through some of the usability issues.

As for the Inkling market, I have to admit that the simple elegance
of their user interface is still fantastic. I'm a little put off by
the trading algorithm, however. I think that when purchasing shares
the site needs to present more data than what is currently there, so
that you don't accidentally drive the price of a contract well above
what you meant to do. For experienced traders this interface will
very likely be frustrating, since we're used to bid/ask prices and
standard terms like that. In the Inkling world, you have to tweak
your purchases and short sells so as to not unbalance the market.
The way the user interface works, though, makes it fairly intuitive
for a person new to trading.

I do still have a problem with the way one particular market works,
however. (The US House of Reps contract, but I've discussed that
before.)

Both sites are still very much beta, and I'm sure their founders/
owners both appreciate and probably get a little frustrated with some
of the extra attention. At the same time, I think it's great that
two new prediction markets opened up within the last month. It gives
us all new sources of examples and hopefully demonstrates the
potential in the marketplace for these markets.

Cheers for now.

Another new site

Thanks to a post on the PM Google Group by Bo Cowgill, I heard about yet another new prediction market website.

This one, called CrowdIQ is particularly easy to join. It looks to be extremely beta right now, but gives some idea what it's about. In fact, according to the CrowdIQ Blog, they aren't actually launching the site until Monday, March 13th.

Right now there are a variety of markets listed, but no active contracts. There are a few main differences between this site and other active PM's.

* The goal of CrowdIQ is to encourage the public to come up with their own prediction markets, which would be essentially hosted on their site. Their example was that if you wanted to establish a market that would predict if the Georgetown basketball team would make it to the NCAA tournament, you could just sign up and create it. (This is obviously similar to the Foresight Exchange.)

* The markets that people create can be either public or private. If public, anyone can join and trade. If private, you must be approved prior to trading.

* Each market has its own currency. This is an interesting development. As an example, I managed to sign up with three markets. In the first (Apple Rumors), I was granted A$5000. The next I received IQ$2500, and in the final market (Enron Jail) I received K1000. I don't want to assume too much since none of these markets have opened yet, but this seems like an odd decision.

It might make more sense if the markets were extremely general (such as one for politics, one for entertainment, etc.) so that a successful trader in one field wouldn't necessarily be able to leverage their success and overwhelm other markets. But with such specific markets, it appears to me that it will simply be an averaging mechanism. If everyone trading in a market starts with the exact same amount of money and can only use it in that market, they would use all of it to bet their preference. Provided every trader uses that same strategy, all the market serves to do is average the traders opinions.


So those are my initial thoughts. I look forward to their markets opening and playing with it a bit. The user interface at least looks nice, but without making trades it's difficult to tell how it really feels. That will be an important test for broad user acceptance.

Cheers for now.