Back in February Lincoln City were in the midst of a remarkable run of form that prompted me to write a series of posts regarding whether they were actually as good as the results were suggesting. To recap the Imps began the season with a 21 game stretch during which they scored 19 points and followed that with a run of 7 games that yielded a further 18 points.
I found that two things were behind the upturn in fortune. The first was that their woeful save percentage had regressed to a league average level and the second was that the teams shooting percentage had gone through the roof (it was an unsustainable 48.2% over the seven game stretch). Using shot rates along with shooting and save percentages I looked at how Lincoln could be expected to perform over the remainder of the season. So how close was it in reality?
So why was the prediction so far off? It rests in the values I predicted for sh% and sv% over the remainder of the season. I’ve actually learnt quite a lot since I wrote those posts. For one, the average sh% in league 2 is 24.3%, not the 30% I thought. Lincoln’s sh% over the last 18 games of the season was 24.6%, exactly league average.
The second thing I’ve learnt is that sh% and sv% can vary wildly, even from one season to the next, so trying to predict the values is pretty pointless. Hence I was no-where close to the correct value for sv%.
I want to highlight just how historically poor Lincoln’s sv% was over those final 18 games. Recently I looked at the sv% of premiership teams over a stretch of 19 consecutive games. Out of a sample of 4000 there is only one occasion during which a team have a sv% lower than 54.7%. That was by Arsenal in the 2001-02 season (their sv% over the other 19 games of the season was 81.7%, comfortably above league average). At no point during the 2010-11 season did Lincoln have a prolonged number of games where they posted a sv% that is close to league average (not even for games 22-28). I don’t think anyone could have reasonably expected Lincoln’s sv% to be so poor over the course of a full season.
One further thing I’ve learnt is that shots on target (or even total shots) are easily the best predictor of future performance and thus the best indicator of their true talent level. By splitting the season into two parts just games 22-28 in one part and all others (games 1-21 & 29-46) into the second part I can see whether the team was better in the seven game stretch or the results were simply smoke and mirrors. I’ve also included their PDO to show how much luck they got.
|1-21 & 29-46||-1.00||0.74||0.442||23.88||57.99||819|
So their SOTR was virtually identical all season long and yet their results were vastly different. The explanation for the shift in fortunes lies in the teams PDO value, which saw a huge spike during games 22-28 which was mainly caused by an unsustainable sh%. Realistically it seems very likely that Lincoln’s strikers were simply very lucky for a seven game stretch as the alternative is that they found some new skill that allowed them to score at will for seven games which they promptly forgot. For reference a team with an SOTR of 0.444 will typically score ~54 points over a 46 game season.
So in conclusion the reason for Lincoln’s relegation can be attributed to their historically poor sv%. In the premiership team sv% appears to be ~85% luck and ~15% skill and I’d argue that league 2 has a smaller gap of skill between it’s best and worst teams so the luck component is probably higher (for example in the championship the figure grows to ~92% luck).
Over the course of the whole season Lincoln were probably a pretty bad team who also happened to get terrible luck.