Graphic: top of the drops

Hull City. I need you to think for a little bit about Hull City. I’ll come on to the rest of the relegation battle in a moment. But only if you stick with me while I ride my Hull hobby-horse.

When the season began, Hull had a 46 per cent chance of going down. In other words, they were — just — more likely to avoid the drop than to be relegated. Then they began the season in tremendous style.

Their performances seemed to defy the odds and in individual games they most certainly did. It was highly unlikely that Phil Brown’s team would defeat Arsenal at the Emirates Stadium. But although individually their results seemed extraordinary, they are a lesson that, despite the cliché, you really can’t take football one game at a time.

Football match results cluster. Let’s say that a team are capable of winning 50 per cent of their games. They will not win and lose alternately. They will achieve results in clumps. So when teams go on a run, you have to be very cautious about whether they have really improved or are simply having a cluster of good results that will later be cancelled out.

Computer modelling using goals and shots on goal allows you to make this judgment. And the Fink Tank has always believed that Hull’s early results flattered them and that they would come back down to earth. The only question now is whether the season is going to end before they make it all the way down to the drop zone. There is an 80 per cent chance that it will.

So Hull’s progress is eloquent about the importance of always, always, always judging teams on their class and not form.

Now, as promised, the rest of the race. Dr Ian Graham, Dr Mark Latham and Dr Henry Stott have been computer simulating the rest of the matches repeatedly. At the beginning of the campaign this yielded what, to me at least, was a startling discovery. It suggested that there was, as you would have read in these pages in August, “a strikingly high chance of Newcastle United being relegated, which we give as an eye-catching 20 per cent”. This has risen to a perilous 57 per cent. The chance of two teams from the North East going down is 66 per cent (Middlesbrough have a 70 per cent chance of being relegated).

Much more surprising to us than Newcastle’s difficulties have been those of West Bromwich Albion. They now have a 96 per cent chance of going down.

This is 90 per cent higher than we judged it was at the beginning of the season. On the whole, using Coca-Cola Championship data to estimate Barclays Premier League chances has worked well for us — it helped us to predict Wigan Athletic’s flying start and that of Reading — but not, it seems, on this occasion (although overall the model is outpacing the bookies).

We have never regarded Blackburn Rovers as likely to go down and we now regard them as relatively safe (relegation chance 5 per cent). Sam Allardyce was correct to gamble upon going there. And Portsmouth, who should not have been anywhere near the zone to start with, can feel reasonably pleased that there is a 93 per cent chance they will play in the Premier League again next season.

Manchester City still have a tiny chance of relegation — 0.17 per cent. And that is precisely why the Fink Tank does not start fiddling subjectively with the stats to adjust for something like Robinho’s arrival. It is far better to let the results speak for themselves.