Betting on it

There has been some interesting chatter recently about the betting market for the NSW election.  I always take very careful note of what the betting agencies are offering - after all, they more than anyone have no partisan interest, and the amount of money they have riding on the result means that the odds they offer are (presumably) untinged by emotion or sentimentality.

It is in the bookmaker's interest to make sure that the actual favourite is priced as the favourite - if not, it is likely they will loose money.

The one thing that must be borne in mind, however, is this.  Once a bookmaker takes a disproportionate amount of money on one "runner", they will shorten the odds to prevent a monetary loss.  This does not necessarily mean that the prospects of this runner are any better than they were a day ago - but because the bookmaker now wants to discourage betting on this runner, they will shorten the odds.

Whilst the betting company has no sentimentality, the betting public certainly does.  For example, I would expect that when Australia plays in a sporting event the Australian betting public bets more on Australia than, say, a bookmaker in the opponent's country.  Consequently, all things being equal, the odds on an Australian win should be shorter with the Australian bookmaker.

Having said that all that - I doubt it makes a huge difference for political betting, although divergent odds on different major betting company websites suggests that that view may be premature.

Anyway.  I've made a spreadsheet on google docs which you can see here: http://bit.ly/gdecpv

It lists the NSW electorates, names the current member's affiliation, and then lists the winner as predicted by Sportsbet.com.au.

Frustratingly, Sportsbet is only offering odds in 38 electorates, and they're listed in black in the "Sportbet tip" column.  It's not even only the "close" competitions that are in the market - of the 38 markets, 8 have one runner where the odds offered are $1.05 or less.

I've filled in the remaining 55 electorates using Anthony's Green's tips and some guesswork, and they are all marked in red. Surprisingly, there are some very tight competitions that are not on the Sportsbet market - there 7 seats presently held by the ALP with a margin of 10% to 15% that have no market.  

Of those 7, six are left in the ALP column.  One (Kiama) is left marked for the Coalition because Anthony Green said so, apparently because a redistribution makes it s bit more Liberal.

So, what did this spreadsheet reveal?  When the Sportsbet market is combined with my guesses, the ALP win 26 seats, the Coalition win 61, the Greens win 2 and 4 seats are won by Independents.

Of course no one expects a uniform swing - Kogarah has a margin of 17.7% and Sportsbet tips it to move to the Coalition, whilst Toongabbie with a margin of 14.5% is tipped to be retained by the ALP.

That said, if we get out Anthony Green's Election calculator out and plug in numbers, what swing spits out the ALP winning 26 seats?  Roughly, 12.2%.  That's about a 60/40 two party preferred split.

Even if I move all 6 of the 10% to 15% ALP seats into the LIB column and make it 67 to 20, the swing is still less than 15%

60/40 a shellacking, by any measure.  Probably too much for Labor to be back in power in 4 years time.  But the swing "predicted" is still far less than the numbers we're seeing in the polls, which which have tipped swings of as much as 18%.

Of course the only poll that matters is the one carried out in polling booths on 26 March.  But I think we learn far more looking at the Sportsbet market than we ever will looking at a Newspoll.