It’s about this time of year that journalists and pundits start to look at the remaining fixtures and predict what’s going to happen: who will win the league and who will be in those coveted top 4 positions. On the face of it this is not the trickiest of tasks. Three quarters of the season has gone; it’s easy to see who is in with a chance, it’s not like predicting it before the season starts.
And yet predicting the results of individual football matches is fraught with danger. The beauty of football is that the best team does not always win a one-off match. Let’s face it, if predicting results was easy bookies would not spend so much time and effort encouraging people to have a go at it. Ray Winstone’s floating disembodied head could keep its opinions to itself.
Last season with seven games to go Tony Cascarino, who writes in The Times, had a go at predicting the outcomes of every match still to be played by the top 6 at the time. Cascarino was described as ‘our expert’. I must warn you that this is the man who praised Podolski for his workrate. His opinions are frequently at odds with mine, so I came at this wanting the opportunity to make fun of him.
At the time of the predictions Chelsea led the table by a point from Liverpool, who were two ahead of Man City. Arsenal were a further three behind, but six above Everton.
Things didn’t start well for the ‘expert’, as his first two predictions were Chelsea to beat Crystal Palace and Man City to beat Arsenal. Neither happened, with Palace winning and Arsenal drawing. But there immediately is the folly of this game: if top of the table Chelsea can lose to Palace then anyone can beat anyone.
I won’t go through every one of the 38 matches that Cascarino tried to predict, but his overall score was 23 out of 38 – so he got almost 40% wrong. Sunderland and Palace were particular problems: both beat Chelsea, and Sunderland drew with Man City, while Palace drew with Liverpool. Palace also beat Everton, and Sunderland got a draw against the same team. None of this was predicted by Cascarino, who expected the relegation battlers to lie down and die against the top six.
As it turned out the ‘expert’ did manage to predict Arsenal, Everton and Spurs in their correct finishing positions, though none of them with the correct number of points. In fact Arsenal, after a bit of a scare, finished a comfortable seven points clear of Everton, and there was none of the swapping positions back and forth with the blue Scousers that Cascarino foresaw.
In the top three the predictions on position were worse, but the points predictions better – other than Chelsea. The pressure got to the West Londoners, while City dropped a few points unexpectedly but still triumphed. Cascarino had predicted Chelsea would sail imperiously to the title, though he was bang on with the number of points for Liverpool, getting five out of their seven results correct.
So what does this tell us? Firstly, as should be obvious, trying to predict the outcome of individual football matches often fails. You only have the past to go on, and in football the future often arrives in a different form to that expected. Many matches are won by luck, and psychology also plays a big part – which team needs a win more, which is more tired (or believes they are), which is more ‘up for it’, which has the momentum of ‘form’?
This is why any statistics on individual games are largely useless. A team, let’s say Arsenal, can play their usual way, have most of the possession, many more passes, and take more shots than say West Ham, but hit the post six times and lose to a single opposition goal from a lone attack.
But with a bigger group of games it’s easier to predict who will gather the most points. Seven games per team is not a huge number, but at the end of the season you already have 31 matches of form to study, not to mention points gaps between teams, and a gap of more than a few points is usually unlikely to close or be reversed in seven games.
So Cascarino had far more chance of predicting the correct order of the top six in seven games’ time starting in March than he would with everyone starting on zero points in August.
If the season was seven games long, literally any team could win it. But we know that over 38 games the cream will largely rise. The bigger, richer clubs will generally be nearer the top and the poorer ones near the bottom.
Over 38 games every team will have some poor results, but the big clubs have time to claw back any ground lost to the likes of Southampton, who might start well but are almost inevitably less strong than a Man Utd or Liverpool over a season.
So does this actually make it easier to predict the final top four correctly in August than it does in March? Probably not, as there are always one or two surprises even among the big clubs – in August you won’t know which big club will have an off-season this time, whereas in March it’s probably become more obvious. But I did gather a long list of pundits’ and journalists’ top four predictions at the start of the season, so I’ll dig those out and have a look soon.
In the meantime I find it difficult to make fun of Cascarino as much as I was expecting for this subject, even with his 40 per cent failure rate. Results average out so that mistakes on single matches are less relevant, so even with being far from precise on results you can get teams in the right order. Then again, he predicted Chelsea to win with seven games to go and they came third, so perhaps he is a useless twonk.