The Betfair Prof / Leighton Vaughan Williams / 19 August 2008 /

What do opinion polls tell us about the race between Barack Obama and John McCain this November? The Betfair Prof, Leighton Vaughan Williams, tells all...

There is a part of conventional wisdom that dictates that the betting markets are simply driven by the opinion polls, and that you may as well look to the polls as the markets if you want to predict the outcome of an election.

Of course, polls do have an important influence on markets - it would be strange if they did not - but a recent examination of movements in the Betfair markets and of statistical models based on the polling data casts serious doubt on the closeness of fit of that relationship.

Take as a starting point the sophisticated statistical modelling of the polling data available at a site called www.fivethirtyeight.com. As of this morning the site with the slogan, 'Electoral Projections Done Right', gives Barack Obama a 56.9% of winning the election against John McCain. In terms of electoral votes, this translates according to the methodology used into an advantage of 283.5 to 254.5 for Obama, and a one point lead in the popular vote.

Compare the electoral vote projection with the polling averages available at www.realclearpolitics.com and we have a similar story. This site gives Obama a slightly smaller edge (273 to 263 electoral votes). Simultaneously, the Betfair markets allow you to back the Democrat at [1.52] and lay him at [1.53]. Take the mid-point between the back and lay price as a rough-and-ready guide to the implied probabilities in the odds and we have Senator Obama with almost a 66% chance of winning the election.

But it's not just the difference between the implied probabilities in the odds and polls as of today that is so striking. It is how this has changed over the last few days. Look back just a week or so, for example, and www.fivethirtyeight.com gave Obama well over a 60% chance of winning the election and www.realclerarpolitics.com awarded the Senator from Illinois comfortably over 300 electoral votes based on their polling averages. At that time, however, Obama was much easier to back in the markets than he is today, trading at between [1.58] and [1.59].

In other words, as the polls and sophisticated statistical models based on the polls have indicated that the race for the White House is tightening, the traders on Betfair have seen fit to shorten up the Democrat quite a few points.

Whatever the reason for this, what it does tell us is that the markets are not simply shadowing the opinion polls. Perhaps they are taking into account factors that the headline poll numbers are not. For example, they may be reflecting how much money the campaigns have in the bank or that state of their get-out-the vote operations. Some commentators argue indeed that the polls are under-estimating Obama's support because some Hillary Clinton supporters are telling pollsters they would support McCain in the hope that this will persuade the delegates at the Democratic convention even now to switch to the former First Lady. Most of these people, so the story goes, will revert to the official nominee by the day of the election whatever happens. Or perhaps the markets are simply slow to respond to new information.

Whatever the explanation, it's certainly getting very interesting!

Professor Leighton Vaughan Williams is the Director of the Political Forecasting Unit and Betting Research Unit of Nottingham Business School, Nottingham Trent University, and Editor of the Journal of Prediction Markets

'.$sign_up['title'].'

'; } } ?>

:

:

:

?

: