[extropy-chat] US election prediction markets failure

[BillK]

Before the latest US election, pundits were noting that the Irish
prediction market Tradesports.com was surprisingly accurate at
predicting US election results.

<http://www.vnunet.com/vnunet/news/2168001/luck-irish-predicts-elections>

After the election, The Register has a very cutting article detailing
how the prediction markets failed in this case.

<http://www.theregister.co.uk/2006/11/11/prediction_markets_not_so_wise/>

Quotes:
In Silicon Valley this year, "collective intelligence" is the
mandatory piece of psycho-babble necessary to open a Venture
Capitalist's cheque book. Surowiecki's faith in prediction markets
appears unshakeable. Writing in Slate three years ago, in an attempt
to save Admiral Poindexter's "Terror Casino" - punters were invited to
bet on the probability of state leaders being assassinated, for
example - Mystic Jim begged for understanding:

"Even when traders are not necessarily experts, their collective
judgment is often remarkably accurate because markets are efficient at
uncovering and aggregating diverse pieces of information. And it
doesn't seem to matter much what markets are being used to predict."

"Whether the outcome depends on irrational actors (box-office
results), animal behavior (horse races), a blend of irrational and
rational motives (elections), or a seemingly random interaction
between weather and soil (orange-juice crops), market predictions
often outperform those of even the best-informed expert. Given that,
it's reasonable to think a prediction market might add something to
our understanding of the future of the Middle East."

A heart-warming fable, then, for a population robbed of their
pensions, and beset by uncertainty after the dot.com bubble.
Suroweicki failed to mention however that experts are regularly
outperformed by chimps, or dartboards - but no one talks about "The
Wisdom of Chimps".

This week however the people spoke - and the markets failed.
-----------------


The problem with predicting trends, whether by computer software or by
markets, is that a trend is only a trend until it isn't.  Drastic
changes are notoriously difficult to predict.


BillK