There's food for thought about Australian federal Labor's recent polling improvement in these comments about the US Presidential election from Ezra Klein's Washington Post Wonkblog:
There’s an argument out there that the idea of “momentum” is largely a conceptual error. Momentum means something in physics: A car rolling down the hill gathers momentum as a result of it rolling down the hill. But that’s not obviously true in politics. There’s no reason to think that a candidate experiencing a few good days of polls or news will, by virtue of that good news or good polling, experience more good news and good polling.
But I’d go further: I’d bet that a careful study of media mentions of a candidate’s “momentum” would find that they tend to presage that candidate losing altitude in the polls. That’s because while “momentum” may not be real, reversion to the mean is.
When a candidate has been overperforming where they’ve been in the race for long enough that the media has become convinced of their “momentum,” that’s good evidence that they’re getting an unusually good run of good news or good polls, and the race has drifted away from its fundamentals. And when that happens, it tends to mean that the likeliest thing the race will do next is return to its fundamentals.