Victoria is the fourth Labor government in recent years to seek a third term.

In two – Queensland 2004 and Tasmania earlier this year – the opposition scored a significant swing (around 4%) but no joy in terms of seats; just three gains in Queensland, none at all in Tasmania. NSW (2003) was even worse for the opposition, with no net gain of seats and a small adverse swing.

The question for tomorrow is whether Ted Baillieu can break the pattern: no-one thinks he can win, but the Liberals hope to make the sort of gains that would set up a realistic challenge in 2010.

The polls, however, show results very much in the Queensland or Tasmania range. ACNielsen on Monday had Labor leading 54%-46%, a swing of just under 4%. Galaxy today in The Herald Sun says 55%-45%.

Yet many commentators are suggesting something bigger could be on the cards. William Bowe, the Poll Bludger, is picking a net opposition gain of eight seats, which would be (just) enough for Baillieu to claim a moral victory.

Peter Brent at Mumble expresses his uncertainty by saying “We should be in for a thumping Labor re-election. We probably are, and yet …” (his ellipsis).

According to the pendulum, a swing of 3 or 4% would yield a gain of seven seats: there is quite a gap on the Labor side, with nothing between 2.7% and 4.4%. But I expect that those seven marginals will swing by less than the state average, and unless the overall movement is unexpectedly large then as many as four of them could hold on.

There are also vulnerable Labor seats above the 4.4% mark; those most often mentioned have been Prahran (4.4%), Eltham (4.8%) and South Barwon (5.0%), but others including Mordialloc, Bentleigh, Morwell and Frankston will be worth watching. My guess is Labor might lose two out of this group.

On the other hand, there is also a concentration of very marginal non-Labor seats: ten with margins of 3% or less. Even with a moderate anti-Labor swing, it’s quite likely that one of these will fall: watch out in particular for South-West Coast, Bass, Caulfield, Box Hill and Mornington.

In total, I expect the opposition to make a net gain of about four or five seats — not enough for Baillieu to claim victory, but enough to keep his job and get the whole leadership cycle in the Liberal Party going again, à la NSW.

If he does significantly better, he will have set a new benchmark for opposition success.