When he was asked what was the most difficult thing about his job, Harold Macmillan, the former British Prime Minister, replied: "Events, dear boy, events."
There is plenty of time for events in the three weeks between today and the European election on May 22. No projection is ever set in stone, stuff inevitably happens, and in the end the outcome is entirely up to the voters.
In fact, the last few candidates are so closely bunched it wouldn't take very many votes to grasp the final seat from Jim Nicholson of the UUP and give it to Alex Attwood of the SDLP, or even – in a perfect storm – Jim Allister of the TUV.
Equally, Mr Nicholson's vote could increase. In 2009 he did better with Tory backing. Our model also assumes that Martina Anderson's surplus will mostly go to Mr Attwood on the second count. However, in 2009, when there was a 5,040 surplus accrued by Bairbre de Brun, the Sinn Fein candidate, it wasn't distributed because, under electoral rules, it wasn't large enough to change the order of candidates.
This time there are more candidates closer together, so Ms Anderson's surplus is more likely to be shared out among the others. A few more votes for the smaller parties could change things.
Technical considerations aside, public opinion can change. This is the first election contested by NI21 or its candidate, Tina McKenzie. She could perform well in a TV debate and bring out new voters, but that is hard to do, and we don't yet know how much airtime she will get. The same goes for Henry Reilly of Ukip, who could also benefit from his party's improved standing in Britain.
Alliance would say that Anna Lo traditionally outperforms her party's average. She pushed the Alliance vote up from about 6% before she stood (in 2003) in South Belfast to 20% in the last Assembly election. Her party will be hoping that she repeats the trick. On the right day she could pull ahead of Jim Allister and stay in the race long enough gather more preferences.
In TV debates leading candidates can slip on banana skins. Big parties who believe they have seats in the bag may resist debates which could allow the smaller ones to score points at their expense.
That leaves no excuse for voter apathy.
This election promises to be a close-run thing, yet our model assumes that only about half the electorate is likely to vote. If more, or less, people actually turn out, then anything can happen. Candidates can surge or plummet.
Those who stay at home thinking that voting changes nothing couldn't be making a bigger mistake. This is an election where moulds could be broken – or reinforced.