The relationship between possession and shots
The graph below shows the relationship between average possession per game (along the bottom) and shots per game (along the side) for the 98 sides in the Premier League, Serie A, Ligue 1, La Liga and the Bundesliga.
All data is taken from the excellent WhoScored website.
As you might expect, there’s a fairly obvious correlation – the more possession you have, the more shots on goal you’re likely to attempt, which is hardly a revelation.
The graph is interesting, however, for two reasons. First, because there are clear differences between the five separate leagues. Second, because there’s a handful of sides that don’t fit the pattern, and a lot of variation amongst the sides who see a lot possession.
The sides who are significantly ‘higher’ on the graph compared to the line of best fit are particularly efficient with possession – they have more shots than you’d expect for the amount of the ball they enjoy. Those who are significantly ‘lower’ are less efficient – they see a lot of the ball but record relatively few shots on goal.
Of course, being more or less efficient is not necessarily ‘better’ – because the sole purpose of possession is not to score a goal. Possession can be used as a defensive tactic to play out time when a side is ahead, and can be used to tire the opposition, before attacking more directly later on. The intention here is not to ‘rank’ sides, but to show their different styles.
Here’s a similar graph breaking the 98 sides into their five respective leagues:
The most interesting pattern here is the Premier League, because so many sides are on the more direct side of things. Swansea, the black dot towards the bottom-right corner, are unique within the division, which explains why there’s been so much discussion of their patient passing game.
A line of best fit for each league also shows the differences:
But what about individual teams? Here, some sides are significantly higher or lower than what we’d consider standard. The most efficient sides average one shot from every 3% of possession, the least efficient teams need 5% to have an attempt.
The most efficient sides are:
- Manchester City – the light blue dot
- Real Madrid – the grey dot
- Tottenham – the light green dot (the top three are virtually indistinguishable)
- Chelsea – the blue dot
- Juventus – the black-and-white striped dot
Also of interest are QPR (horizontal stripes) and West Brom (vertical stripes), who both record significantly more shots than you’d expect for their average possession.
The ‘least efficient’ sides are:
- Ajaccio – the dark blue dot
- Koln – the green dot
- Swansea – the orange dot
- Evian – the purple dot
- Athletic Bilbao – the red-and-white striped dot
Also of interest are those higher up the possession scale – Bayern Munich are the red dot, Barcelona are the red-and-blue striped dot.
Again, the purpose is not to indicate that a particular style is better – simply that there are big variations. Swansea see more possession than Manchester City, yet have six fewer shots per game. Manchester United have as many shots as Barcelona, despite 12% less possession per game.
In general, more possession will mean more shots – but it’s not always the case.
I would like to see goals vs. possession…
Taking shots is a tactical issue. Scoring goals depends on individual aptitude.
So maybe a good compromise would be to compare shots on target to possession.
p.s Its a bit surprising to find Bilbao amongst the ‘least efficient’ sides.
ZM talks about efficiency in shooting, not scoring. I wonder whether they are actually quite efficient in goals/shots (if they score a lot while having few shots).
I’m not too sure we are talking efficiency in the real sense here as it does not tell us if the shots were actually on target?
I think ZM is merely talking efficiency in a loose sense i.e. relative to possesion.
I’m more of the opinion this is more a revelation of style and intent rather than how well a particular style works.
For all we know, many of these shots were probably 20-30 yard shots, woefully off-target!
Efficiency is always an abstract idea, it is always a relative measure (ie. dividing numerator by denominator), you only have to know what two values you are taking into your calculations. In this article ZM has stated that by efficiency he understands ability of teams to tranform possession into shots (any shots, not necessarily shots on goal), that’s all.
Absolutely — “efficiency” does NOT mean the greatest number of shots. It means the greatest number of goals per shot, or, perhaps, the greatest number of goals per possession percentage.
No, efficiency means the highest ratio of A/B. In this article, A is shots and B is possession. Goals are completely out of context here.
It’s a combination of goals, shots, possession.
ZM shows how efficient possession was, not how efficient the shots have been. But for some teams possession is about defending, too. Others have just been poor teams.
real efficiency is about …. what? fouls versus possession, distance versus possession, sprinting versus walking?, shots on target per total shots, shots per goal, passes per goal, passes per possession ….
It’s all, and nothing. It’s about context and style. This stats can show us a little bit of the style teams play. But they don’t show us the context of a single game, a specific pass, or shot, or move, or tackle…
In the last Champions League Season (2010/11) Barca played more passes than any other team and had more possession as any other team. No surprise.
But they also played more passes when you calculate the passes per possesion (or passes per minute if you like) stats. Their passing was more intensive than the passig of any other team. That says something about their style.
Spielverlagerung.de looked into the stats of the CL 2010/11 Season (in German). I think it’s even more in this numbers, especially when it comes to defending.
http://spielverlagerung.de/2011/11/09/zusammenfassung-des-technischen-berichts-der-uefa-zur-clsaison-201011/
I’m not a particularly obsess with stats but in my opinion here,
a shot on target per possession ratio is a more relevant measuring stick.
I don’t think so. Shooting at target “depends on individual aptitude” (if someone above allows me to use his expression :>), while we want to get teams tactics – willing to shoot.
Instead of shots on/off target maybe there is a way to look at where the shots come from on the field. For example a team with little possession spraying lots of shots in from outside the 18 isn’t nearly as efficient as a team with more possession and substantially less shots all coming from within 10 yards of goal. I think this takes “individual aptitude” off the table.
That’s what I was thinking while reading this article. You can pepper shots from 20 yards out and miss the net all day… I wouldn’t call that an efficient side.
And sure, you can argue that shots on target/off target is and individual problem and not a result of team tactics, but I would say that it is the goal of your offensive tactics to get the best shots possible… If a team is missing the target, there is surely something wrong with the tactics used.
That’s why ZM only meant efficiency of translating possession into shots (any).
And that’s why this graph makes imho barely any sense, while a combined graph (like shots per possesion (%) vs. goals per shot would be perfect…
For me the key is which areas possession occurs most! if you take Swansea as the exeption above a majority of their possession is in their own have as they build up from the back, if a team has more possession in the opposing half they will clearly have more attempts on goal, if a team has 80% possession in their own half they simply will have less chances!
Nice work Michael, and great use of simple stats and graphs to make the point of the article very clear!
What I’d like to see is what possession does to the net number of shots. As you mentioned, possession works as a defensive tactic too, and teams like Swansea clearly use it that way. Showing net shots will better reflect the whole picture, not?
I had a look at the stats on possession versus shots conceded. Stoke and Swansea being the obvious teams to pick out:
Stoke
Possession: 39.4% (Worst in Premier League)
Shots Conceded: 13.9 (8th best in Premier League)
Swansea
Possession: 58% (2nd best in Premier League)
Shots Conceded: 15.5 (8th worst in Premier League)
which gives the impression that possession as a defensive technique doesn’t work especially well. Except that Arsenal, Barcelona, Juventus, and Bayern Munich (with Lille and Marseille sort of sharing the spoils in Ligue 1) have both the highest possession and concede the fewest shots in each of their respective leagues.
I think Arsenal, Barca, Juve, and Bayern benefit from the fact that teams are naturally cautious against them. There is something psychologically daunting about playing against those teams, so opponents are more careful.
Swansea benefits less because despite all their possession, the opponent may be more aggressive and counter-attack more prolifically when they do get the ball, simply because it’s Swansea and not Barcelona.
Moreover, a part of Swansea’s success in possession may be the fact that teams are less defensive against them than they are against the other teams with high possession, so they are given more space to pass and keep the ball. But that leads to more shots conceded since the opponent might have more men forward then it would versus an Arsenal or a Barcelona.
It would be interesting to see similar tables for sides when they were winning/losing.
Barcelona seem unbelievably patient at times but does their shot/possession ratio increase when they are behind.
If you include crosses do the numbers look much different. Sides who like to work crossing opportunities might have a more inferior possession/chance ratio just because they didn’t make the most of a good cross (which deserves to be counted just as much as ANOTHER speculative Bale or Downing shot from 30 yards!).
What about goals to shots – surely if you’re taking a lot of pot shots from distance and having loads of possession that’s not an indicator of a perfectly functioning side?
Well done ZM.
Without even telling us which dot represents which side, many observers would still have been able to tell the sides on the extremes of the graph.
Over the last few years(essentially post-Ronaldo) I have always thought Real Madrid the most un-Spanish of all the Spanish sides; they are much less patient than the other La Liga sides and tend to play a percentages game ala British sides where the idea seems to be “if you don’t buy the lottery ticket…”
RM look to get the ball quickly in their last 3rd and “rain” shots on goal.
I can almost bet (if statistics were available), that of all the top sides they have the least possesion in their own half.
ZM, I think one can determine, which is the most efficient team based on posession, only after considering – the shots on goal, goal scored, goals suffered as well.
As you have rightly mentioned, posession is used for both attack and defence, and hence.
ZM, does that mean Swansea is copying rather Bielsa, than Barcelona?
Now that’s funny!
Couple of observations:
This seems to back up the idea that, in attacking terms, possession has its limits. Barca dominate the ball but because of this teams line up so defensively that it’s very hard for them to to turn possession into chances (they’re just so good that they manage it anyway). Playing a more open game, like Real Madrid do, creates more chances by drawing the other side out.
Of course what you want to do next is cross reference this chart with how many shots CONCEDED per game. Because while sides that seek to dominate possession will probably have fewer shots than sides whose more open style of play creates more space to exploit, they will probably still be ‘up’ in net terms because they concede fewer chances.
And if you *really* want to waste time, you can then categorise the chances being created. So does possession lead to one-on-ones or shots from distance, clear chances or half-chances? And would this corroborate the feeling I have that, while Barca and their ilk concede very few chances per game, the nature of the way they play mean that those chances are almost always clear ones (ie break the press and you’ve probably got a one-on-one).
dont be so sure, in the last 4 league seasons Barcelona conceded less goals than Real Madrid.
That’s exactly what I was alluding to when I said in the middle paragraph: “while sides that seek to dominate possession will probably have fewer shots than sides whose more open style of play creates more space to exploit, they will probably still be ‘up’ in net terms because they concede fewer chances.”
From a complex systems perspective the Bayern and Barcelona dots might be weak signals of a pattern that suggest one can have too much possession. That would be an interesting angle to explore especially since neither of those teams won their league this year.
Another interesting statistic might be possession and actual goals scored for and against. Maybe a lot of possession means more packed defenses and more long range speculative shots. Maybe too much possession results in the greater likelihood of counterattack goals against them.
As indicated in the commetns above, it seems like Bayern and Barcelona use possesion more defensively, and successfully so. The question is, what’s the best balance between attacking and defending.
Maybe not necessarily so?
We need to know in which areas this possesion was attained.
I’d suggest if you are having more of the ball in the opposition half or even last third you are probing and looking for an opportunity.
The fact that this results in your team not being attacked much, should not make this a defensive tactic as this is merely the result rather than the aim of your possesion.
Also some teams are more content to wait all day to create that perfect opportunity, i.e. they are more concerned with the quality rather than quantity of chances created.
Of course if much of your ball-work is done in your own half with a few forays into the opposition half, then you are probably using possesion as a defensive tool
One could argue that possession in the opponents end means there will be less space and it will be harder to score and maybe easier to lose a goal to a counterattack. So a little less possession might mean more good opportunities to score.
I agree another interesting stat would be quality shots. That was a huge issue when the N. American hockey teams started playing the Russians and Czech’s they would out shoot them by large margins and lose the games handily.
Interesting that in the premier league there are 7 sides that have about 55% or above posession and all the rest have less than 50% with quite a large gap in between. I guess the 7 teams are the top 6 (less Newcastle) plus Swansea and Liverpool. which means the other teams are likely playing counter attacking football I suppose?
“In general, more possession will mean more shots – but it’s not always the case.”
ZM I love you but that’s lame conclusion
Btw (i should not ask you this after my remark but well…) do you consider doing an article on PSG’s kinda false nine system? I know they’ve disappointed you when you saw them but they’ve been playing without a center forward of reference for quite a few games now and it’s working.
Let’s explain it basically: 433/4312; menez and nenê cutting the center back to full back pass lane, pastore deeper but sometimes pressing the center backs.
It can be uninteresting when matuidi & sissoko are playing behind the front 3 because it makes PSG a basic high rythmn&turnover, counter-attacking team (still the front three can play freely with flash without worrying about bad decisions);
but when both motta and bodmer play they become a nice side to watch, with composed & well thought attacks (rare in Ligue 1).
So if you ever see a PSG game without hoarau and both matuidi&sissoko starting you may enjoy it, maybe against the battling for europe Rennes or an hypothetic championship deciding game against the always well playing Lorient
+
This reminds me of Villas-Boas’s discussion of match-control with possession. I’ve always wanted someone to elaborate more on that concept, how match-control and possession relate and how best to exert match-control. The issue has fascinated me immensely, especially of late.
You control space when you don’t have possession and you control tempo when you are in possession.
I think it’s possible to control space when in possession (let the opponent run) and control tempo when not in possession (press the opponent).
But the first usually leads to goals (controlling the ball and being able to use space) and the second leads to winning the ball.
So if a team wants to have long stints of control over space and tempo in possession, it will create chances and cancel before scoring (but keep the ball and kreat a new chance).
If a team wants to have long stints of control over space and tempo without possession, it will hunt the opponent but don’t win the ball, even though it would be easy to win possession.
Both are situations where one team is so much better that it controlls the opponent anyway and would be able to win the ball or score but prefers to play cat and mouse (with and without the ball). And I don’t think there are much teams that don’t want the ball (for counter attacks) or that don’t want to score if possible.
Sometimes you spend time plotting data in the hope it will show something interesting only to be disappointed. This was one of those times.
You must not be able to read then.
a great chart would also be “possession” vs. “shots conceded”
+1
What we need is a comparison with other measures. How does shots on goal compare with balls played into the penalty area; with territory in the final 3rd of the field; with ratio of passes played forward to those played back; with significant runs made off the ball? I think your work is interesting enough for ferreting out styles of play between teams; between leagues; but, as you hint at, it doesn’t necessarily mean anything in terms of shots at goal and winning games.
I think you’re on the right track. Possesion vs shots is nice, but its just a start. If you are truly looking for an ‘offensive (possesional) efficiency’, Possesion vs Territory (or chances created) might more informative. This is a very good article, though. I’d love to see more detailed and descriptive statistical analysis for soccer.
There’s a PPG(points per possession) stat at NBA which is considered as the most important stat by itself.In football the stat whould show how the possession ended – did they win a corner, throw in, set piece, with a shot, with a cross etc.
Then we could really make something out of the numbers, pretty sure Bilbao is crossing a lot and hence the low number of shots, cross into a box is a high EV play not that much lower than a shot(on goal, not on target only), so we miss a very important piece of information right there.
There’s a lot more to the relationship of possession with “X” in football, and definetly something worth of thiking about, but possession – shots, is kind a meaningless.
I agree that this is what’s needed – or at least some sort of error bars on this denoting shots on target.
But what really intrigues me is how PPG stands for points per possession?!
He made a mistake he mixed up the abbreviations for Points Per Game and Points Per Possession. His points still hold though.
Why is everybody missing the point ZM was making? I guess that must be frustrating for the author.
The stat you are talking about is an indicator of a real effort/danger created by the team. ZM’s stat shows the tactics of that team, their decisions, their plan. You are talking a posteriori about what happen on the pitch, while ZM tries to find out what the team is thinking a priori (if these two terms ring a bell). Example, from 30 shots of some team none could be on target, but it doesn’t change the fact that the team’s tactic was to transform possession into shots from distance, only thing – they had weak shooters.
Not necessarily.
It makes more sense to talk of efficiency if the measured event is a result/intention of your actions.
If a team possesses the ball as a defensive tactic then it doesn’t make sense to measure their shots as they are simply concerned with not conceding shots themselves or losing the ball.
In that case to see how efficient the strategy is it might be better to count the number of times they lost the ball or shots conceded.
Would it make sense to say Inter’s possession in that 2nd leg semi against Barca was highly in-efficient because they didn’t convert what little possession they had into shots?
We do get ZM’s point though but, we recognise the limitations of his assumptions
However, there’s no such confusion talking efficiency when measuring goals relative to shots as I can’t think of any other reason for taking a shot, regardless of whether it’s OFF or ON target!!
Given extensive studys on it, its meaningless, reggressing to the mean for about 50%.
Interesting stats. But i would prefer a goal per possession or even more a pts per possession graph. Esp. b/c gaining pts is the main objective of every match and thus more relevant than shots, which are just one subgoal in order to gain pts (i.e. winning matches).
Also shots differ – quality wise. A shot from 30 metres count as much for this analysis as a shot from 2m on an emtpy goal. And you also see the flaws when considering the Barca philosophy (rather increase the quality of goal-scoring opportunities than the quantity of it).
There is a lot of fanning out once you cross the 50% mark.
That makes the analysis more difficult since it implies that once you cross that threshold other factors (i.e. playing style/team philosophy) have a greater effect on shots taken than possession.
In other words, imagine breaking the data up in two sets: below 50% possession and above 50% possession. IMO, you might find that possessions explains a lot less of the variance in shots taken in the second set (above 50%) than it does in the first set (less than 50%).
For Spain in particular, I think there will prove to be no statistically significant relationship between possession and shots fired above 50-52% possession.
Good point. I would suggest that Spain stands out because it’s the league where teams are most likely to use possession for reasons other than to create chances.
Nice work michael, i suggest you do a linear regression on the data so the last line isn’t so vague, easy to do in Excel.
Good article. I’d like to compare that to the nature of the shots (distance, taken in front of or behind the defence) to see how good teams are at creating dangerous chances. I think it could also indicate the teams’ style of play. For example a team that take most of there shots from behind the defence and have low possession would probably rely on counter-attacks and through balls.
Just a thought
I think better than shots on target if you wanted a crude measure but not individual aptitude would be headers and shots outside the box excluded
I decided to do some research on the so-called high-efficiency sides:
Real Madrid
Juve
City
Tottenham
Chelsea
I find one thing in common- they each have one player that produces a lot of so-called “Key Passes” according to whoscored.com
For Juventus, that’s Andrea Pirlo, who is second in Europe in terms of key passes. For Chelsea it’s Juan Mata, who is 5th in Europe. For Madrid it’s Mesut Ozil, who is 6th. For City it’s David Silva who is 8th. And for Tottenham it’s Luca Modric who is 12th.
On the other hand, Arsenal has no one in the Top 15 in terms of key passes, while Barca has Messi, who is 15th- not spectacular. Ajaccio’s top key passes man can’t even pull of 2 key passes per game.
So perhaps you need a player like Pirlo, Mata, or Ozil that can cut a defense with a smart pass if you want to achieve high efficiency. A kind of fulcrum that opens up the opponent and provides opportunities to shoot.
Interestingly, there seems to be no relationship between the number of players taking lots of shots and efficiency.
Real Madrid’s shots are mostly taken by Ronaldo, but Juventus, who also are very efficient, have quite a few players that take a similar number of shots.
So you don’t necessarily need a player that takes it upon himself to fire (like Ronaldo), but you do need a fulcrum like a Pirlo, Mata, or Ozil.
The German paper die Zeit put a interesting article on their web page: Jedes Tor kostet 703.460 Euro / Every goal costs 703,460 Euros
It’s about the efficiency of the Bundesliga.
http://www.zeit.de/sport/2012-05/effizienzrecher-bundesliga-tore-geld
(link to the interactive graphic http://www.zeit.de/sport/fussball-kosten)
Bayern “payed” 1.298.701 Euro per goal.
Borussia Dortmund 512.500 Euro.
And SC Freiburg only 293.478 Euro.
They also calulated the costs of a victory in the Bundesliga (2.8 million) and every successful pass was worth 3.344 Euro.
Best online manager game. Play football and hockey and win Champions League! Follow us on this link: http://www.powerplaymanager.com/r384368