Establishing which sides make the most of their possession in the SPFL

Written by: Jamie Kilday (@FitbaInScotland)

Back in May, I looked at how adjusting stats based on possession can give a better idea of how a team plays as opposed to just looking at stats as a ‘per 90’. The article can be found HERE but I’ll give you a brief summary.

When you look at possession stats for football matches as minutes rather than as a percentage you start to realise that for a large chunk of the 90 minutes the ball is dead. In most games the ball tends to be in play between 40-60 mins, once you realise this you can see how problematic using a ‘per 90’ metric can be as two teams playing a match where the ball is in play for only 40 mins will presumably have fewer actions than a game with 60 mins, however this is not always the case. As an example of how much this fluctuates, we are only 4 games into the 18/19 SPFL season and already we have Livingston v Kilmarnock which finished 0-0 from a combined total of 59:06 min of possession and Motherwell v Rangers finishing 3-3 with only 43:40 mins.

With all this in mind, I’ve decided to look at which team make the most of their possession. What’s the point of having a lot of possession if you don’t do anything with it? So let's see which teams are impotent in possession and which teams play…virile football…

...

…maybe lets say, have a high 'Possession Efficiency’ instead.

First, a few things to bear in mind;

  • All event data is from open play, no point rewarding a team’s possession efficiency for goals scored/attempts from set pieces.

  • With that is mind it's worth noting what validating the metric against total points could be problematic as teams win games with goals and some of those goals will be from penalties, direct free-kicks and set piece routines…obviously.

  • This is only one seasons worth of data so repeatably isn't clear yet. We'll have to see in May if there is anything more to come from this metric.

  • At the time of writing, we don’t have the data to analyse when a possession began and ended so I will be using averages. Another way this could be analysed if you had the data would be to look at how long it takes a team to get a shot away after first receiving possession.

  • As there is no real average number of minutes of possession across all clubs as it is dependant on the style of play and ability so I am using the ‘Per Possession’ metric to represent one minute of possession.

  • For context here is the average number of minutes of possession for each club in the 17/18 season.

Dashboard 1 (6).png

I’ll cover this over the coming season but for now, let's look at last seasons data.

Attempts/Chances/Shots

Dashboard 1 (7).png

Call it what you want, when it comes to attacking the goal here is how the 12 SPFL Premiership clubs compared. Celtic and Rangers not only averaged the highest amount of possession last season but were also the most efficient at converting that possession into a shot. Kilmarnock averaged around four minutes less possession per game than Hibernian and Aberdeen yet are third when it comes to shot efficiency. As I've said before, it doesn't matter how much possession you have if you make the most of it when you have it and Kilmarnock certainly do that.

Unsurprisingly Partick Thistle are rooted to the bottom after what turned out to be a disappointing season for the Jags. Ross County on the other hand were not terrible at creating chances when in possession.

Dashboard 1 (8).png

As I touched on earlier, this should be taken with a pinch of salt as not all points are earned from open play but early signs show there is certainly looks to be a relationship between the two. The top 5 clubs stand head and shoulders above the rest of the pack whilst Hearts who also finished in the top 6 look unremarkable sitting in the mix with Dundee and St Johnstone.

Aberdeen don't look great here, with what looks like a huge over performance, largely due to how they relied heavily last season on pieces to create chances. Over 22% of their attempts coming from a set piece (not inc Direct FK and Penalties). Motherwell had a similar percentage of chances coming from set pieces but their lower overall possession and very direct style meant they did not have the same perceived over performance.

Goals

Dashboard 1 (2) (1).png

Whats going on with Rangers then?

A large percentage of Rangers goals last season came from open play so it is hardly surprising they come out top here. With only a seasons data it’s difficult to say if a rate of a goal every 17:21 minutes of possession is sustainable but from Jason of @TheGersReport’s work looking into other metrics like conversion rates and shot accuracy it’s likely that a figure so high is rare. More on that can be found HERE.

As appears to be the way, the top 5 teams in the league occupy the top 5 (all be it in a different order) with a surprise team in 6th. Now it’s St Johnstone who are the only side from the bottom 6 with an above average goals per possession.

A side that are noteworthy here is Motherwell. One would expect that their direct style of football would have led to their goals per possession being high as they are not keeping possession for long (as is the case with their attempts) but 10 goals from set-pieces from the Steelmen shows that their long ball, low possession tactic isn't entirely fruitful in open play... losing Louis Moult is also a contributing factor to this.

Dashboard 1 (9).png

Hamilton had very much the league average Goals Per Min but were lucky not to get dragged in to the relegation play-off last season. In the same way that there is no point being in possession if you are not going to do anything with it, there is also no point in being really efficient at creating chances when you have possession if you can't win the ball. It could also be said that if you are not going to have much possession, any goals created on OP play will naturally lead to a high Goals per min, to see how lucky Hamilton were to get the goals they did is probably best to look at the thinking mans favourite metric xG. 

xG

Dashboard 1 (2) (2).png

For the first time Celtic and Rangers have been separated at the top and based on the other statistics in this article is probably no surprise that Kilmarnock have separated them. It will also be no surprise to anyone who followed the analytics community last season that Hearts are 2nd from bottom. Its widely acknowledged that Hearts over performed last season and considering the quality of the chances they created per minute of possession you can see why people think this.

Hamilton's open play xG per possession is less than their actual goals which would point to there being a slice of good fortune behind the number of points they managed to get last season.

Dashboard 1 (10).png

Of the three aspects, I've explored this one seems to have a greater separation amongst the top 5 teams in the league where's the bottom 6 have moved closer together. Hearts continue to be an anomaly, where as on the previous graphs they were very much in the mix with St. Johnstone and Motherwell, they have certainly dropped off. xG is widely regarded as being a better indicator of future performance than actual goals but as with all statistical theory, there will always be something that doesn't conform. The fact that on this occasion the non-conformist is managed by Craig Levein, a man who very much dances to the beat of his own drum is very fitting indeed.

Conclusion

Celtic dominated most of last season but are outshone by Rangers when it comes to Possession efficiency, this could be read a couple of ways but I'd say the most logical reasons for this is fatigue and game state. The stamina needed to consistently create chances when you have a high possession must be high, this has only been over a small sample but I would imagine that across multiple leagues over several seasons, the possession efficiency stats will plateau or even drop once you reach a certain average minute of possession. There is also the fact that Celtic spent a lot of their minutes in a winning position last season so the need to attack would be low.

Obviously this is a small sample and shouldn't be relied upon yet solely as a metric in its own right, I'm very interested in seeing how this develops over the season and if more data becomes readily available I will look to open this up across different leagues and seasons to see if there are any further trends to be explored...or not, its not unheard of for a statistical theory to turn out be a dead end (its probably more often the case than not), It could well be that this is all statistical noise that fades into the discourse.

Of the ten clubs remaining in the SPFL from last season only Rangers have changed their manager over the summer so I would expect no great changes in the style of play from the other nine. If it stays the case over the course of the season it could make the search for repeatability easier. Although there is fat chance of that happening, especially with new boys Livingston and St Mirren seemingly being managed by a timeshare scheme so far this season.

In my next article I’ll look at attempts/goals/xG conceded per possession against and see which sides limited the oppositions possession efficiency. It can be found HERE

This article was written under the influence of BBC Radio 6 Music

This article was written with the aid of StrataData, which is property of Stratagem Technologies. StrataData powers the StrataBet Sports Trading Platform, in addition to StrataBet Premium Recommendations.