General Politics Betting / Jack Houghton / 05 January 2012 /

The following article was originally published in the Press Gazette in July 2006. It has been republished here as it is referenced in this article.

Any student of political science will tell you why the 1940 US Presidential Election was important. Franklin D Roosevelt's decision to run for a third-term in 1940 was the catalyst for the 22nd Amendment of the US Constitution: barring a President from more than two-terms in office.

It would take a more dedicated student to tell you that it was also the last US Election where Betting Commissioners on Wall Street offered wagering on the outcome: fundamentally changing forever how journalists reported matters of public opinion.

Between 1884 and 1940, political betting was big business on Wall Street. Betting Commissioners, previously working out of pool halls and bars, moved into offices in New York's Financial District.

For brief periods more money was traded on politics than on stocks and bonds. In 1916, the modern equivalent of $165 million was bet on whether Wilson or Hughes would prevail.

But it is not the volume of betting that's of note. Rather, it's the startling accuracy of the betting market in predicting the winner of public votes. In the 15 elections between 1884 and 1940, the favourite on polling day won 14 times.

This prescience made the current odds of participants the mainstay of election reporting for journalists, with national and local papers carrying daily updates and analysis.

As the New York Times explained in 1924: "The Wall Street odds represent the consensus of a large body of extremely impartial opinion that talks with money and approaches Coolidge and Davis as dispassionately as it pronounces judgment on Anaconda and Bethlehem Steel."

But this method of gathering impartial opinion was in decline. The legalisation of on-course horseracing betting in New York in 1939, and an increasing feeling that betting on politics was unethical, meant that 1940 was the last Presidential Election to see widely-organised political betting markets.

Political journalists found themselves at a bit of a loss. With no barometer of public opinion available, reporting the likely winner of elections would be little more than a well-informed guess.

Cue opinion polls.

Heralded by statisticians as a "scientific" way of assessing public opinion, and heralded by journalists as a way of lending validity to copy, they have become the common language of elections the world over. But do they really deserve such a billing?

Recent history suggests they are not always a reliable indicator. In 1992, opinion polls running in to the election showed Kinnock's Labour Party holding a clear advantage over the seemingly haphazard Tories. On the morning of the election, five major opinion polls showed an average lead for Labour of 0.8 percent - which would have returned a hung parliament with Labour as the biggest party.

The reality was different. Eight percent different as it turned out. Opinion polls, so diligently reported in the run-up to polling day, were wrong. But not just wrong - monumentally wrong.

Professional pollsters blamed a number of things in the aftermath of 1992. Many cited a supposed late swing in opinion, others blamed high refusal rates. Few of them raised the more fundamental question - are polls really a reliable indicator of anything?

Polls are created by asking a number of people (usually around 1,000) a question. These people are assessed against a view of what a demographically representative sample should look like and their answers weighted accordingly. In other words, the pollsters try and balance ethnicity, age, occupation and the like to represent wider society.

They then do something mathematically complex - probably involving pi - and out pops an answer.

This is the problem with opinion polls. Mathematically jiggery-pokery does not represent real life.

Most opinion polls only have a level of accuracy of plus or minus three percent. A poll could show the Conservatives on 42% and Labour on 37%. The reality may well be that Labour were leading.

But the core problem for political journalists is that opinion polls take time to produce. You've got to write the questions, find the people to ask, collate the results. Then there's that mathematical jiggery-pokery. These things take time. Time that a tight deadline does not allow.

Questions over opinion poll accuracy, and an increasing need for broadcasters and journalists to be up-to-the-minute in their reporting, has seen things start to turn full circle and drive back towards the days of the Betting Commissioners.

During the 2004 US Presidential Election, CNN and other media outlets used the Iowa Electronic Markets (IEM) as an alternative guide to the likely winner of the popular vote.

These markets, operated by the faculty of the University of Iowa as a research and teaching enterprise, showed Bush as clear favourite from early September - which conflicted with major opinions polls that had the race neck-and-neck.

The limitation of IEM is its size. As an academic enterprise it has aimed to prove that markets are the most efficient way of aggregating all available information. But with less than 500 active participants and less than $100,000 traded on the last Presidential Election - it was hardly a vibrant market.

Vibrancy is not an issue in Betfair markets. Tens of thousands of punters bet on the outcome of Bush-Kerry on Betfair - with £20 million of bets matched. And these punters got the result of the main event spot on.

More impressively was how the markets successfully predicted the outcome of 11 out of 12 of the marginal races. The one that the markets got wrong was Ohio, where the Democrats were seen to hold a 50.1 percent chance of winning compared to the Republican's 49.9 percent. Perhaps excusable?

And UK media outlets are beginning to recognise the worth of Betfair markets as timely and accurate indicators of public opinion.

In the last month, after a series of Labour controversies, the Tories are now favourites to win most seats at the next General Election. But only just. They are reckoned to have a 51.8 percent chance compared to Labour's 47.5 percent. Given the overwhelmingly negative press that Blair has received in recent months, it's amazing that it's this close.

But what would you rather believe: a mathematically dodgy poll saying the Tories are further ahead, or a real-time poll where people are backing-up their opinions with money?

Some will argue that those betting on political markets have their opinions formed by the opinion polls they read about and what political commentators tell them - that one will naturally follow the other.

There is merit in this argument, but as the academics at IEM say: it's not that the betting markets are a direct equivalent of opinion polls (they're clearly not). It's that they're an aggregator of all the information that's out there - whether opinion, fact or supposition. This is where the Betfair markets are invaluable.

'.$sign_up['title'].'

'; } } ?>

:

:

:

?

: