(Apologies to those with no knowledge of Victoria’s electoral geography – some of this post might not make a lot of sense to you.)
Coverage of the Labor vs Greens contest in the Victorian seat of Batman (in which – disclosure – I am strongly supporting the Green, who is a personal friend) has picked up on the fact that Labor’s David Feeney, having started as favorite, is now noticeably behind in the betting odds. Sportsbet now has him at 6-4 against, with the Greens at 2-1 on.
That’s an interesting if not decisive data point. But it motivated me to have a look at the odds for the Greens in other Victorian seats, and they’re fascinating. Here they are, as of lunchtime today (Monday):
20-1 on Melbourne
2-1 on Batman
4-1 Melbourne Ports
15-1 Corio, Gorton
20-1 Ballarat, Calwell, Chisholm, Corangamite, Holt, Hotham, Isaacs, Jagajaga, LaTrobe, Lalor, McEwen, Maribyrnong
25-1 Aston, Bendigo, Bruce, Casey, Flinders, Gippsland, Kooyong, McMillan, Mallee, Menzies, Murray, Wannon
The first few seem plausible; I’d say Melbourne Ports is a bit stingy and Wills if anything is a little generous, but they’re clearly in the right order of magnitude. Beyond that, however, things get pretty strange.
Not just in the sense that the Greens don’t really have a one in 21 chance of winning, say, Ballarat – it would really be one in many hundreds, if not thousands. That’s just how bookmakers work. What I’m talking about is the relative placing of the seats.
Scullin is a safe Labor outer-suburban seat; the Greens only got 7.5% there in 2013, well below their state average. What on earth is it doing at 7-1? Meanwhile Kooyong, one of their best seats last time with 16.6%, is way out on 25-1.
There’s essentially no relationship between the strength of the Green vote, or any other objective factor you can think of, and these odds. Corio and Gorton are poor Greens territory (worse even than Scullin), while Casey, Corangamite and Jagajaga are all above average.
Indi, at the bottom of the list, was indeed very bad for the Greens last time, but Goldstein, priced next to last, was one of their top ten.
There are really only two plausible explanations for this. One is that there has been practically no money wagered on any of these races, so the odds simply reflect the bookmaker’s initial settings – and that whoever did this at Sportsbet did it by sticking a pin in a list while blindfolded or otherwise incapacitated.
The other, more likely, is that these markets are all extremely thin, so the prices have been moved around by small amounts of money wagered more or less at random. Whoever is running for the Greens in Scullin, for example, might have had $20 bet on them by an indulgent uncle, and that was enough to produce a dramatic shift in the odds.
But if that’s what’s happening, then it sounds a note of caution about using betting odds to predict election outcomes.
As I said, I think the first few prices on the list – where the markets should be more robust, since they’re the only ones where any sane person would be risking real money on the Greens – are probably not unreasonable. But if we have to rely on our own background knowledge to make that judgement, then the betting market isn’t doing much independent work as a predictive tool.
And if you do choose to experiment with some of these odds, please bet responsibly.