Prediction Markets
A Prediction Market is an emerging forecasting tool that has
seen predominant use at major tech companies, such as Intel, HP and
Microsoft. It works like an internal
stock exchange where employees trade stocks representing future outcomes – the
price of each stock represents the likelihood of that future event becoming
true (it is a little more complicated than that). Many times prediction markets have proven
more accurate than standard forecasting methods. The two most common prediction market topics
are related to sales forecasts and project completion dates, but there are many
other possibilities such as bug counts and resource allocation. So a prediction market can help you answer a
question such as, “how many units will we sell of Product X in the coming
quarter?”, and “when will the new software version be available?”
The idea behind a prediction markets is that employees have
detailed knowledge about specific parts of their company’s business, the kind
of stuff they learn from their daily activity. This knowledge is difficult to access via “traditional” means, such as meetings
and surveys. Some of the shortcomings
that prediction markets overcome are that they are dynamic reflecting the most
up date information, intrinsically scalable in terms of participants, they force
participants to quantify their opinion, i.e. “put their money where their mouth
is”, and operate according to market efficiency theory. Thus, a prediction market allows employees to
express their opinion on what next quarter’s sales will be or whether the
project they are working on will finish on time. For management it means access to information
in all corners of the company and potentially more accurate forecasts. And better forecasts allow for better
decisions about the future.
While the basic idea behind prediction markets is powerful,
they are many issues related to the design of a market that are not fully
understood, such as how many participants do you need, how do you induce
participants to trade, etc. Then, there
is the issue of having average employees appear smarter than the boss (for some
managers this is an issue since they are supposed to know everything).
The bottom line is that prediction markets can provide
superior forecast and should be explored by leading tech companies.
Additional info:
http://en.wikipedia.org/wiki/Prediction_market
- a great overview.
http://www.chrisfmasse.com
- exhaustive resource.