The Betfair Prof: "The Forecasters are meeting in Nice, and some are talking politics!"

The Betfair Prof / Leighton Vaughan Williams / 23 June 2008 / Leave a Comment

The Betfair Prof, Leighton Vaughan Williams, tells us what insights prediction markets can offer about the electability of candidates...

This week I am in Nice, France, attending the 28th Annual Symposium on Forecasting, presented by the International Institute of Forecasters. The line-up of speakers is as stratospheric as usual, including Nobel Prize winner and founder of modern portfolio selection theory, Harry Markowitz, and the theme of the conference is typically topical - 'Information Communication Technology: Forecasting in a Digital World'.

My particular contribution to the conference is as chair of a session on forecasting methods and as presenter of a paper entitled, 'What insights can prediction markets offer about the electability of a US presidential candidate? A case study'.

The abstract of the presentation reads as follows: 'The 2008 US Presidential election cycle has attracted unprecedented volumes of trading on a variety of markets, including those available during the 'primary' season about the likely Democratic and Republican party nominees. Much of this trading is conducted on person-to-person betting exchanges, the records of which provide a useful source of information for testing a variety of hypotheses about the value of such markets as a forecasting tool. In this paper data is gathered on bets placed during one phase of the election 'primary' season and is employed to provide insights into the electability of candidates in the general election'.

Core to this study is what I like to call the 'Texas surprise', which reflects the rollercoaster evening of Tuesday, March 4, when Barack Obama shortened up to prohibitive odds to win Texas on the basis of a few early returns, odds which gradually unravelled through the course of the evening as Hillary Clinton went on to snatch victory in the 'Lone Star State.' While this was happening, my primary concern has been to consider the knock-on effect in two immediately contingent markets, i.e. the odds on offer about the winner of the Democratic nomination and about the winner of the Presidency itself. This is very much still a work in progress, but the basic conclusions are perhaps unsurprising - as the odds swung away from Senator Obama to win Texas they swung to a somewhat parallel degree away from the Illinois Senator to win the nomination and also to win the Presidency, though never at any time anything like sufficiently to seriously dent his advantage over Senator Clinton.

More interesting, I think, is the impact on Mr. Obama's 'electability index' as the results gradually emerged, i.e. the conditional probability of him winning the Presidency IF he won the nomination. As his showing in Texas deteriorated so did his electability index. In other words, the markets were indicating that losing a primary not only dents your chances (however slimly) of winning the nomination of your party, but it also dents your chances of winning the general election IF nominated. That's the bad news for the Democratic hopeful. The good news from his point of view is that as things stand he's still almost twice as likely (according to the Betfair markets) to win the Presidency as his all-but-certain Republican opponent. That, it seems, is the size of the challenge currently confronting Senator John McCain of Arizona.

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

Betfair Politics Zone