How does one make Prediction Market work?

[Tibaut houzanme]

Background:

As Aurora WDC is currently running a prediction market test with the following bold question "Can the Wisdom of Crowds Predict Competitive Telecom Victory?" (see newsletter exerpt below) one question came to my mind and I'd like to share the rationale with the experts on this group.

 

Ambivalent statement:

As it is getting established that crowds have wisdom  and have been proven to predict outcomes, it can also be admitted that they cannot always do so. And by putting the veracity of the wisdom of crowd theory to rigorous test, it feels intellectually uncomfortable to imagine a crowd of Mexicans predicting with accuracy the outcome of the presidential elections in the Republic of  Benin (West Africa), or a crowd of high school students predicting with exactitude the outcome of the war in Irak. And examples/cases can go either directions, which leads to my question.

 

Question:

As many experts support and believe in PM and as James Surowiecki mentionned,  "[...] when our imperfect judgements are aggregated in the right way, our collective intelligence is often excellent" nowquestion is How then to aggregate "the right way"? How to mix qualified enough Crowd Members to effectively gess an outcome?

 

Does anyone have a good take on this?

 

Thanks for your contributions in this discussion.

 

Ulrich Tibaut Houzanme,

With Functional Interests in:

| Knowledge Management
| Competitive Intelligence
| Digital Information Mgmt;
| Electronics, Telecoms,
| Retail & Services Industries
|---------------------------------------
| Contact:
| Voice: (+1) 317-332-3296
| Email: tibaut_...@yahoo.com
| ---------------------------------------

 

 

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Arik Johnson's ReconG2 Weekly from Aurora WDC
     Need-to-Know News for 16,000 Subscribers Since 1995
           Monday 2 April 2007 | www.ReconG2.com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

"The gambling known as business looks with austere disfavor upon the
business known as gambling." - Ambrose Bierce

Can the Wisdom of Crowds Predict Competitive Telecom Victory?

We'll find out, as Aurora runs a Prediction Market around decisions by
the world's largest telecommunications customer (the U.S. Federal
Government) to qualify contract awards worth as much as $48 billion over the
next 10 years for agency bidding by AT&T, Verizon and Qwest.

If you have anything to do with the telecommunications industry, you
already know that, last week, the General Services Administration
announced these three firms as the winners of its "Networx Universal" contract
covering everything from VPN and VoIP to frame-relay and ATM services.
But that's just the first part of the world's biggest telecom deal.
Read the full situation analysis online for background.

>>> http://recon.c.topica.com/maaf87habxzL6b7uFKbbaehom5/

Coming up in May, the GSA is expected to award the second half of the
contracts, called "Networx Enterprise", to teams of vendors led by one
of five competing companies, and this time the list will include our
original three winning prime contractors above plus Sprint-Nextel (the
only team spurned by the Feds in last week's qualifying round) and a fresh
team led by Level 3 Communications as additional bidders.

Just to keep things interesting - and in no small measure because Jim
"Wisdom of Crowds" Surowiecki himself is keynoting the SCIP Conference
in New York a month from now - we decided it'd be fun to setup a
"prediction market" (one of the techniques described in his book) to forecast
the big winner (or winners - plural - if they split the awards again)
of the second part of the deal. And, we're going to give away a brand
new Video iPod from Apple to the most accurate predictor.

If you already think you know which of these companies will win and
can't wait to setup your trading account to get in on the action, click
the link below and login with our friends at Inkling to get started -
they've got good "how-to" help online for PM newbies if you're not sure
what you're doing and they've also got 5,000 shiny new "Inkles" (the
fictional unit of currency used to keep score) waiting for you when you do.

Join us online and see if you can predict whether Sprint can make a
comeback, new kid Level 3 will emerge victorious or some other combination
seems most plausible. You can spread your risk by hedging your bets and
maybe win a new iPod in the process, but only if you're better than all
the rest... and, worst case scenario, you lose all your Inkles and walk
away having learned a new set of techniques to anticipate the future of
your company's business decisions. And, while you're there, you might
want to weigh in on who will emerge as the real "Number 1" in tonight's
NCAA Basketball Championship.

>>> http://recon.c.topica.com/maaf87habxzL7b7uFKbbaehom5/

Even if you're not compelled by my guarantee that you'll come away
having learned more about prediction markets, at least you'll know what
Surowiecki's talking about at SCIP when he tells you how the answers to
the competitive questions your company cares about most - deciding which
new products to develop, what markets to invest in, whether sales
volumes will change, who to watch among the new breed of competitors - can
be anticipated by aggregating the information of a community of traders
trying to out-pick one another. Whatever the question, as the Inkling
motto goes, "none of us is as smart as all of us"...

There's more background on all of this at the link to prediction market
related items below - online now at www.ReconG2.com.

>>> http://recon.c.topica.com/maaf87habxzL8b7uFKbbaehom5/

[H. Hofkirchner]

Hello Ulrich,

The answer to your question about "qualified mix" and "right
aggregation" depends on your market design characteristics, which in
your case are already chosen, as you decided for a play-money / prize
approach. With these choices, you have established the value of the
information (its value to you?) and set a benchmark for the "quality"
of experts which will be attracted (and before play-money ethusiasts
in this group start shouting ... yes: assuming no bragging value in
such a sensitive affair as contract awards).

Without knowing details I'd guess that the prize is very small
compared to the prospective contract size, so maybe you could have
gone for much more valuable information (bid prices?) to be discovered
with a different question and a real-money approach. Obviously, the
lower the economic value of the question asked, the less you should
expect to attract valuable, qualified opinions ... the aggregation by
the market mechanism then becomes a case of garbage in, garbage out.

But let us know how the game went ... I'd warn against using it as a
"general" test of the validity of the prediction market concept,
though.

Best regards,
Hubertus