Are they though?

They only reflect probabilities, they don't dictate outcomes. If the odds on outcome X are as low as 1.2 with all other outcomes at odds of (say) 10 or more, you might think "well that's a sure thing then" but it isn't, it just means that if you run the same election (or whatever) 100 times (say), roughly 83-ish times outcome X will occur. Variance is A Thing - but as you run that election a thousand or a millon times it "should" get closer to that roughly 83% chance.

Of course in the real world every election is only run once. So it's not possible to evaluate with precision how efficient/accurate the markets are for any one election.

Even if something has a 95% chance of happening, on average if you try to make it happen twenty times, you will fail once.

However in general the political betting markets do a bloody good job and are generally really useful indicators. They fall down when the market doesn't have a lot of money in it or with "wierdness" like the Reform votes recently (because % support does not translate well to # seats under our electoral system, especially round the 10-20% support mark).

I do think it's interesting how all the polls and prognosticators reckoned Labour had a 20%+ lead over the Tories whereas on the night it was more like 10% - the models are supposed to account for "shy Tories" and the like but this suggests, well, they don't!

Posted By: Old Man, Jul 12, 12:32:05