Polls and Prediction Markets: A Flashback to January, 2008

These are the markets which accurately predicted the winner of every single state in the 2004 US Presidential election and the winner of every single contested state in the 2006 US Senate elections. These are the markets which said that Barack Obama would win the Iowa caucuses comfortably, as would Mike Huckabee. These are the self-same markets which forecast a handsome victory for John McCain in New Hampshire at the same time as the polls and pundits were declaring the race too close to call.

And yet the markets showed not a clue that Hillary Clinton would overcome the momentum of the charismatic Senator from Illinois, but instead declared the race for Obama with the same confidence as the media and the exit polls. Indeed, it took something of an avalanche of real results showing the former First Lady handily ahead before the baton of market favouritism changed hands.

So what happened?

The conventional wisdom is that the models used by the pollsters under-estimated the turnout of female voters by a significant factor. In the event, women, who make up 57 per cent of the New Hampshire electorate, went for Hillary by a margin of 12 clear points, in contrast to Iowa where she lost the female vote by five.

It was the tears that did it, came the familiar cry. Not so, in my opinion. The defining moment for me came in the final debate when the New York Senator was asked a question about a likeability problem. Her response – “Well, that hurts my feelings!” was funny, warm and engaging, only to be interrupted by a curt “You’re likeable enough” from Mr. Obama. In his defence, the apparently dismissive tone in which the words were delivered was probably unintentional. But the damage was done.

It was what those familiar with Hillary history call the “Rick Lazio moment”, when her Republican Senate opponent in the 2000 New York campaign marched across the stage at her during a debate and demanded she sign a pledge card he brandished in her face. Instantly Lazio turned off a good proportion of New York’s women voters, and a not insignificant number of the men.

It all goes to show that the markets are usually a much more accurate predictor of election outcomes than are the polls, but there are times when those trading the markets are a little too dependent on charting and interpreting the numbers. Sometimes voters are motivated by factors which cannot be reduced to raw numbers. Those who are wise to this when it occurs stand to make a lot of money. As time would tell!