From Miguel Delaney
As a quick aside the point of this post isn’t to do with sample sizes however, in fairness, Delaney does later acknowledge the small sample size and say he should have written ‘appears to be true’ in the initial message.
I’d rather address the broader point as to the perception of Spurs being a team that is somehow cursed/falls apart/can’t handle the pressure at the end of the season. So firstly, as far as I can tell, Delaney has his maths wrong. Below is a split of the points per game for each of the seasons since ‘lasagne-gate’
I’m confident in my data set so lets assume ’08-09 has been overlooked on the authors behalf. If we were to sum up that table I’d say Spurs do markedly worse over the final five games in one season (10-11), slightly worse in one (09-10), equally in one (07-08), slightly better in one (08-09), and markedly better in two (06-07 and 11-12) – and, whilst the difference between the averages isn’t statistically significant, it goes against the authors conclusion.
Secondly let’s plot the five game moving average of Spurs points per game over the past six seasons:
So the moving average starts low, rises until the middle of the season, drops a bit, bounces back, and then stabilises at what looks about average by the end. Let’s put a couple of lines on here to signify where a) the average performance lies, and b) how good/bad Spurs would have to be over a six years sample of five game runs to reach statistical significance.
The line at the centre is the average number of points Spurs score per game. The upper and lower lines are two standard deviations from the mean. In a sample size of 33, we’d expect one or two points to fall outside the lines and that’s exactly what we see here.
What causes the climb then fall in the middle of the season? Well that might be worth further investigation – it’s around Christmas time right? Have Spurs had a rash of injuries there? Or managers who do/don’t rotate the players in the starting XI? Or have they just faced tougher opponents in general? Or more away games than home? With all of these factors and a small sample size, it’d be very difficult to identify a signal from the noise.
However to challenge the initial assumption, are Spurs worse at the end of the season? We can’t be sure, but if you’re going to use points per game as your determining factor then there’s more evidence against you than there is for.