Forget the polls, the punters hold the key to election predictions

Fine politicians like Denis O’Donovan were not the only ones to lose out in the election. Bookmakers lost out, too.

Celtic Bookmakers’ owner Ivan Yates became a household name and instantly recognisable face, courtesy of his showcasing by the producers of the Frank Opinion TV series. Perhaps because he made the idea of taking a political punt so interesting, more people, this time around, bet on individual candidates than ever did before. They bet more money, too, some of them wagering scary amounts of cash on individual candidates. And they did amazingly well out of their gambling.

On the morning of the count, two things were clear. The first was that Ivan Yates was richer in profile but poorer in pocket. The second was that the gamblers’ approach to results prediction had a lot more going for it than the opinion poll approach.

Opinion polls pick a thousand people and ask them who they’ll vote for. Bookies deal with a much larger group of people and ask them who they think EVERYBODY will vote for. The indications are that the second group may deliver a better crystal ball of what’s going to happen: the wisdom of the crowd may beat the carefully calibrated and weighted wisdom of the thousand. (Fianna Fáil would say that the wisdom of the crowd also beats the wisdom of the media, because the media decides who everybody SHOULD vote for and gets that wrong, too. But that’s FF for you.)

My actuary friends (of whom I have one) tell me that the Who Wants to Be a Millionaire? TV programme showed group intelligence beating individual every time.

A contestant who didn’t know the right answer to a question on the programme could do one of three things. 1) Remove some of the questions, which improved the odds of getting the answer right. 2) Telephone a friend, selected before the programme as being the most knowledgeable person known to the contestant.

The third option was to poll the studio audience.

According to research into the American broadcasts of the programme, nerdy friends of contestants got the right answer 65% of the time. But the studio audience — collectively — got the right answer 91% of the time.

Now, let’s be clear. Individuals within the audience didn’t do better than the nerdy friends. Some did a lot worse. But when the computer scrunched a studio-full of answers into a single total, that total was better than the brainboxes could come up with. The strange thing about the wisdom of the crowd is how accurate it is on matters of FACT. Starting nearly a hundred years ago, American sociologists and psychologists did a series of experiments to find out just how good group intelligence is.

Experiments asking hundreds of volunteers to guess the weight of a series of

objects, or the temperature of a room, found that the group estimate was as-near-as-dammit accurate, even though only five out of (say) 200 of the volunteers guessed the right figure. The bigger the number of volunteers, the nearer they got to the fact.

One of the researchers, Francis Galton, found that mutual ignorance played a part in excellence of prediction. In other words, where members of the group didn’t talk to each other, didn’t share their guesswork, the whole group did better. Interaction between group members changed the results.

He also found that while every group holds a couple of people who are very good at guessing an outcome, they won’t necessarily be the same people each time, so the best way to get good answers is from the totality of the group rather than from the individuals within it.

James Surowiecki, author of the bestseller The Wisdom of Crowds goes further, holding that the average opinion of a group, members of which don’t know what other members think, will always be more accurate than a consensus reached through discussion.

“You can’t find collective wisdom via compromise,” he says.

In theory, then, the best way to predict election results is not to go for the personal opinions of the few but to seek the general predictions of the many. Particularly if the many are prepared to lay their money down in the hope of doubling or trebling it when the election actually happens.

That’s the way America is going, right now, with several institutions trading bets on contests like the presidential election. Some of them take bets in play money. Some of them take bets in real money. Some of the real money markets are tied to universities and to the research into futures conducted by those universities. The Iowa Electronic Markets has an enviable track record of getting election results right, beating 600 opinion polls in the last four presidential elections.

These “decision markets” are not quite the same as bookies shops. They allow clients to buy a share, rather than make a straightforward bet. But the end result is the same: good prophecy. An Irish website called TradeSports.com has a current affairs section, the trading on which pretty accurately predicted the date on which Saddam Hussein would get his comeuppance.

Currently, in America, if you want to bet or trade on Hillary Clinton’s chances of getting the Democratic nomination for the presidential election, you can put down $45 and — if she succeeds — get $100 back.

To win $100 on Barack Obama’s chances would require expenditure of a little less than $30. But — significantly — just a month ago, a punt on Obama would have cost a mere $19. The wisdom of the crowd suggests that Obama’s gaining ground when it comes to his hold on the imagination of the electorate.

Which, in turn, begs an interesting question. In future, will some of the millions fundraised by candidates, traditionally used (in Ireland) for posters, pamphlets and pints and (in America) for TV advertising, be diverted into trading on the political futures markets and into bookies shops, if only to shorten the odds on a candidate and improve their perceived chances of winning?

Three things are certain, as of this point.

The first is that the bookies’ odds, as a predictive force, are going to figure hugely in all future elections in this country.

The second is that university faculties with a bit of imagination are going to link up with Paddy Power or Celtic Bookmakers to examine betting as a factor in and a measure of voter behaviour.

And the third is that political parties, during the three weeks of an election, are going to learn to dread the release of bookmakers’ odds as much as they dread the publication of the results of opinion polls and focus groups.

Because the science of “political futures trading” says the aggregate judgment of the crowd is better than the judgment of the cleverest individual within that crowd.

The crowd, in short, is cleverer than the most informed pundit. Or pollster.