At the last federal election, the idea of picking results by looking at the betting markets was still in its infancy. This year, however it’s all the rage: Simon Jackman’s blog has a good discussion of the pros and cons.

There’s not much doubt about who the punters think will win, but it’s been remarked for some time that individual seat betting doesn’t seem to match the overall odds. On Centrebet, for example, the seat-by-seat markets have Labor favorites in only 74 seats, not enough for a majority. Even Portlandbet, which currently seems the most favorable for Labor, has them ahead in only 78 seats.

But there’s an interesting variation available on Portlandbet. You can also bet on whether or not the Coalition will win a certain number of seats per state, and that shows a rather different picture.

Take New South Wales, for example. The Coalition currently holds 27 of its 49 seats (counting Calare). Based on the seat-by-seat odds, Labor is favored to pick up four of those – Parramatta, Lindsay, Dobell and Eden-Monaro – which would give an expected total of 23.

But the statewide odds suggest that the punters expect the Coalition to win only 20 seats: the option of 20 or fewer is paying $1.80, while 21 or more is paying $1.90.

Queensland tells a similar story. The Coalition starts with 22 seats, of which (according to the individual odds) they are set to lose four. The statewide odds, however, show a clear expectation for them to end with 17 or fewer seats ($1.67 compared with $2.10).

In both states, the punters think the government will lose more seats; they’re just not sure which ones. Which, given the difficulty of picking individual seats and the tendency for fluctuations to cancel out, is an entirely reasonable way for them to think.

Making a few minor assumptions (and taking an educated guess at South Australia, which was missing from the market this morning), it looks as if the statewide odds tip Labor to win at least 82 seats – still not landslide territory, but enough for a comfortable majority.