At Arsenal’s recent AGM Arsène stated that based on points accumulated so far, he expected the Premier League to be won with a total of 82-86 points this season. He appeared to be basing this on the fact that the top three were all on 20 points after 9 games, which is half a game off a quarter of the season. Here they are:
If any of the three leaders continued to rack up points at exactly the same rate of 2.222r per game, they’d have 84.444 points after 38 games – right in the middle of Arsène’s estimate, so clearly he sees this as a fairly linear progression. He may or may not think the eventual winner will come from the trio of Arsenal, Man City and Liverpool, but he does seem to think that points are accumulated fairly steadily through the season. But is he right?
Last season Leicester won the League with 81 points.
After 9 games the table looked like this:
So on the basis of the leaders’ tally being 21 rather than this season’s 20, Arsène might have expected a higher final total of perhaps 86-90 (2.333r points per game). So that doesn’t work – but that’s not entirely unexpected. Remember when Man Utd won the first 10 games in a season? They didn’t end up with a total of 114. However, if you take the top three instead of just the top one, and average out their points per game after 9 games, then you’d expect a final total of 83. On that basis I’d expect AW would have gone for a range of 81-85 – in which case Leicester’s winning total would have been within the expectation.
In 2014-15 Chelsea won the League with 87 points:
and after 9 games the table was:
Going just on the leaders’ tally of 23, the Arsène range would probably be 95-99, but using the top 3 and averaging, the Arsène range would be 81-85. Chelsea’s 87 points is between these ranges, so suddenly the Arsène method looks slightly less impressive.
I’ll give him another chance: 2013-14 saw Man City crowned champions with 86 points.
As you can see, after 9 games Arsenal led with 22 points:
City themselves were languishing in seventh on 16 points. The Arsène estimate based on leaders only would be 90-94, but on the top three average his guess would be 85-89. Once again the Arsène method proves successful, with the final tally being nicely in that range.
I’ll keep going. 2012-13: Man Utd were champions with 89 points.
After 9 games the position was:
again giving a first place only prediction of 90-94, and a top 3 prediction of 88-92 – Arsène wins again! (Well, you know, on predictions.)
2011-12. Here’s the table after 9:
If Arsène was naïve enough to look only at the leaders, his range would be 104-108. But he’s cleverer than that, so on the top 3 average he’d come up with a range of 88-92. And that was the year when the two Manchester clubs were first and second, both on 89 points. Once again Arsène’s Nostradamus-like powers are plain. His method has worked four times out of five.
2010-11 – 9 games in:
And the final table:
You know the calculations by now: Arsène range: 77-81; Winner’s total: 80. Tick.
Finally (as there is a limit to my devotion to research, if not my curiosity), 2009-10:
Arsène range: 85-89. Winner’s total: 86. All good.
So you can correctly predict the champion’s total within a range of 4 points by the Arsène top 3 method* six times in the last seven seasons! No wonder he sounded confident.
*Well I say the “Arsène top 3 method” – I’m making assumptions about what his method actually is, but this does seem a fairly reliable way of predicting the total the champions will get.
ps: I’m aware there are many complex statistical models that try and predict this sort of thing in a far more ‘scientific’ way than this. but it’s probably quite hard to find one that’s more accurate in practice!