Betfair Official Blog

Prediction Markets: Why Betfair's betting market data can be used alongside opinion polls

Click here to view market

Mike Robb tells us why Betfair's prediction markets add an extra dimension to the analysis of political events

Opinion polling is a great thing. Above all else, it gives the minions something to read on a slow news day. But there can be no doubting that we all sit up and take notice from time to time, no matter how wrong they often turn out to be.

Over the course of the last two months I remember two such occasions. Going into the New Hampshire Primary Barack Obama was said to have a near unassailable lead, well into double figures. But we all know what happened there now, don't we. Perhaps this is a unique example whereby everyone and everything got it so wrong, however, including the polls, betting markets, commentators, academics and, yes, even me. I gave Clinton no hope of winning, though to be fair this was largely on the back of the aforementioned polling figures. In fact, thinking about it in the cold light of day I still can't work out how she did it.

Another poll that made me squint to make sure my eyes weren't playing tricks on me (I'm getting old, you see) was in January when YouGov gave Ken Livingstone just a one point lead over rival Boris Johnson. One point! For a man who many thought had more hope getting the lead role in the Moscow State Circus to be just one percentage point behind the well-established, experienced and, amongst many Londoners, well liked Livingstone was astounding. From that poll onwards I have started watching the London Mayoral race with far greater interest.

The thing is that prediction markets and opinion polling are two very different phenomena, but both can be used together as tools to employ in analysis, comment, general discussion or even pub talk. Prediction markets add a different perspective to polling for many reasons and, in my opinion, should be used with the same validity and force as opinion polling in the mainstream media.

How can prediction markets be used alongside polling figures?

As has already been documented in this blog, it is easy to change betting odds into an implied percentage chance figure. This is the basic thing that many editors and journalists need to get their head around - betting odds essentially are an implied probability, and so 'what the 1,000 people in Mr. MORI's poll said' can quite easily be seen alongside 'what the 1,000 people in Betfair's prediction markets said'.

The two are not the same thing, but they are equally useful sources of analysis and opinion.

What are the advantages of using prediction markets?

There are many things prediction markets can add to the analysis of political events. Two bonuses in particular, however, are that they are both free and instantaneous.

Cast your minds back to the scandal last year when the data disk holding the personal details of more than 25 million people was lost by the HMRC (bless them - only a government department would think it is a good idea to store this kind of thing on an unsecured little disk, probably one that had recently stored a copy of the latest Killers album that the boys and girls of the Inland Revenue didn't want to fork out for).

If you actively track the prediction markets, you could have instantly seen that between Monday 19th November 2007, when the news broke that this data had been lost, and 1300 GMT on 21st November (just after Prime Minister's Questions when old Gordie took quite a beating) there was a two percent decrease in support of Labour, and a corresponding two percent increase in support for the Conservatives, to win the most seats at the next election.

How many hours or even days would it take to amass an opinion poll to judge this? Even if you did, by the time the questions have been asked (and, by the way, to a very limited number of people) the moment is over, and the results are less influential anyway. In addition, how much would this have cost? I don't know the answer to that, but the phrase "too much" comes to mind.

You could always have just logged onto Betfair and got the data straight away.

Right now (1600 GMT, 22/02/2008) this is what Betfair's prediction market is saying:

Who will win the most seats at the next election?
- Conservatives 57%
- Labour 42%
- Lib Dems 1%

In the same time the world's media editors have even started the thought process behind deciding whether or not to commission an opinion poll they could have logged on to Betfair to see if there has been any change... For free.

Click here for the Betfair Politics Zone