An opinion poll on Scottish independence yesterday suggested the gap between the Yes and No vote was just five percentage points.

Every time this happens, and the difference between both scores is so close as to be, given the margin of error, virtually neck-and-neck, a quaver of excitement and/or fear ripples through the media.

Depending on whose side you are on, such a poll portends the imminent arrival of an independent Scotland, or a constitutional catastrophe that could end up sweeping the prime minister from office. So it was again yesterday, with politicians and commentators popping up on TV or on Twitter, extrapolating all sorts of conclusions from this one set of figures.

Regular readers will be familiar with my refrain that polls portend nothing. They are merely snapshots of opinion, not predictions of outcomes. The most constructive way of interpreting polls is to try and place them within a trend. Sometimes there’s no trend to be found, which can still be useful. But there’s usually some kind of pattern that can be discerned, into which the latest poll can either be fitted neatly or discarded frustratingly.

Yesterday’s poll placed the Yes vote on 39%, the No vote on 44%, with 17% undecided. From the graph below, where I’ve plotted the monthly poll averages since last August, you can see these latest figures don’t fit in with the trend at all. The top line, in dark grey, is the % score for a No vote; the bottom line is the Yes vote:

Something appeared to be happening a few months ago, with the Yes vote rising slowly but steadily from a low of 30% to 37% in March. Then it seems to have stalled and, most recently, fallen back slightly.

The No vote, meanwhile, has shuttled around between 45% and 50% for the past eight months. It fell at the same time the Yes vote rose earlier this year, but the average has yet to bounce back to its high in September 2013 of 53%.

Precedent suggests there is no way the Yes campaign will win the referendum. Look at the trend: the Yes average has never once come within the margin of error of that of the No vote.

Yet this referendum is without precedent. Look at how the No average fell eight percentage points between September and October last year. There are now just under 100 days left until the referendum. The real campaign has only just begun. And – according to yesterday’s poll – 17% of voters haven’t made up their mind.

It’s stating the obvious, but it bears repeating: either side could still win.