Points may not reflect ability of clubs outside the elite
The Soccer Boffin's weekly dose of betting wisdom
I worry for Tony Pulis, the likeable manager of West Bromwich Albion. He achieved his best Premier League finish of tenth in his first season for West Brom’s new owner.
If Guochuan Lai is like many other owners he will expect an even better finish next season. But another rise is less likely than a fall, and if that happens Lai could turn on Pulis.
Everything in Pulis’s garden looks rosy, which is when an admirer peeking over the hedge should wonder what might be lurking unseen in the flower-beds.
There is little to choose between most teams in the Premier League.
One can finish above another without being better, just because results broke more kindly for them. Next season, as likely as not, positions will be reversed. But when that happens owners and fans can get cross.
I will show in a moment that the results of Premier League games between non-elite teams are similar to what we should expect if they were all identical in ability.
Southampton this season finished eighth, one below the European places. Watford finished 17th, one above the relegation places. But Southampton won only six points more than Watford. Over 38 games, six points might signify almost nothing.
The average gap between eighth and 17th places in 38-game Premier League seasons has been 16 points. But even a large difference in points between teams might not correspond to any difference in ability.
Take, for example, the years 2001, 2002 and 2003. The teams who finished seventh in those years were Sunderland, West Ham and Everton. Where did they finish 12 months later? Sunderland 17th, West Ham 18th, Everton 17th. One went down and the other two breathed huge sighs of relief.
There have been 20 teams in the Premier League for 22 seasons, from 1995-96 to 2016-17. I isolated games involving teams who both finished between seventh and 17th – outside the top six but above the relegation line.
Eleven teams finish between seventh and 17th. Every season each plays the ten others home and away, a total of 20 games. I counted how often in those 20 games teams won different numbers of points and plotted the answers as the blue line on the graph.
Then I conducted an experiment. I imagined 11 teams of identical ability playing each other home and away.
In every match there would be the same chance of a home win, the same chance of a draw and the same chance of an away win. The chance of a home win would be greater than the chance of a draw, which would be greater than the chance of an away win, but only because teams gain an advantage from playing on their own ground.
Although there is no difference in ability between these teams, some will end a season with more points than others. Some will be luckier than others – they will get more than their fair share of favourable decisions and fortuitous bounces of a ball. I ran computer simulations of 10,000 seasons. I counted how often in those seasons teams won different numbers of points and plotted the answers as the yellow line on the graph.
You will see that the yellow and blue lines are similar. The blue line is spikier, as we should have expected, because it represents a much smaller number of games. The blue line in the middle is higher than the yellow line, indicating that an even larger proportion of teams gained middling numbers of points in real life than in the computer simulations.
If there had been significant differences in skill between the real teams we should have expected the opposite – the blue line to be lower in the middle and higher at the edges. Extreme points totals should have occurred more often.
I am not saying there is never any difference in ability between non-elite Premier League survivors. I am saying that most seasons there is not much difference.
We can also see this if we look in another way. Most Premier League teams who are not relegated win between 42 and 57 points from 38 games. Teams with 42 points in one season averaged 47 in the next.
Teams with 57 points in one season averaged 47 in the next. And so it was, more or less, for all the others in between.
Typically there was no difference over the next season between teams who won 42 or 57 points in the previous season, even though some had just accumulated 15 points more than others.
On average teams with 57 points finish seventh and teams with 42 points finish 15th. Owners, fans – and, yes, journalists too – assume the higher finishers are better. That is why I worry for Tony Pulis.
45 record gives Baggies unlikely top-ten hit
Recently West Bromwich Albion’s payroll has represented between 3.3 and 3.4 per cent of the Premier League total. Teams with such a payroll, in my experience, average 44 points. West Brom’s points totals in the last three seasons, starting with the oldest, were 44, 43 and 45.
Tony Pulis has managed the club for two-and-a-half seasons.
In five seasons with Stoke he won between 42 and 47 points.
Crystal Palace in 2013-14 won 45 points – 38 in the 26 games after Pulis arrived.
Pulis achieved his best Premier League finish of tenth with the Baggies this season, but he did it with 45 points, which is roughly what they have been paying for and what he has delivered for them and others.
Unless West Brom’s payroll rises significantly as a proportion of the Premier League total it would be unreasonable for anyone to anticipate more points.
Do they know, or will they remember, that before this season no team with 45 points had finished as high as tenth?