8 thoughts on “Unexpected Findings of Possession Efficiency”

1. I think you missed what affects number of possessions. The number of possessions in a game should be roughly equal to the total amount of interceptions (including saves the goalies holds), successful tackles, passes out of play, shots off target, fouls by the attacking team, and goals, because each of these constitutes a change of possession (I probably left some out). The three most frequent will usually be interceptions, tackles, and passes out of play. Because of this, if a good team controls a game, with a high passing percentage, the number of possessions will be lower than an average game. And if a team plays a direct passing style, like Bolton do, the number of possessions will be higher. Hope this helps.

   • Certainly! I don’t think I articulated specifically what I thought effects fluctuations in possessions per game.

     Games with large possession counts have a higher “rate of play”, which is almost synonymous with playing a direct style of football. I did find it strange, however, that Fulham was further up than Stoke. This suggests that style of play is not the only factor in rate of possession.

     In other words, what I am looking at in this article is the efficiency of these different styles of play in different game situations.

2. Hi Devin,

   Congratulations on an excellent piece of work. As you know, I’m looking into possession-based analysis at the moment too, so I’ve got some thoughts I thought worth sharing here.
   To go back to basics before trying to answer some of the questions you raised here, what’s the definition you used for possession here? In my work I tend to look at any string of match event containing at least a deliberate action with the ball, i.e. a completed pass, a shot, a set piece, etc. And further, does a possession spell start over in your definition when a team for instance wins a corner, or does that constitute a continuation of the same possession?

   As always in analyzing football, context is everything and not every possession is the same. You can see that very well in your ‘losing situations’ table, where teams really show different patterns of possession efficiency. The same holds true for level match situations and leads. The score in the match dictates much of what’s demanded from a certain string of possession.
   Finally, also the team’s pressure on the opponent will influence the amount of possessions, as it influences the amount of time the opposition is granted with the ball. So simply counting the amount of possessions will always mix offensive and defensive aspects of the game.
   Again, thanks for this interesting post!

   • Hey Sander,

     I’m happy you found your way over to my blog, I always enjoy reading yours.

     To answer the question regarding my definition of possession (which is a question I expected), I treat possession precisely as you do (one deliberate action) – and currently am including set pieces as a continuation of the same possession.

     Hopefully that can help you dig a little deeper into my questions.

3. Excellent piece.

4. Bolton: presumably this is largely due to winning lots of headers and giving the ball away quickly once they get it, meaning a large amount of changes of possession.

   Manyoo, City and Chelsea: more passing, slower tempo, more emphasis on retaining possession, rather than getting the ball forward quickly. Fewer changes in possession.

   You can only get another possession if you give the ball away, so a team that gives the ball away a lot will have a lot of chances to win it back!

   I wonder if it would be possible for you to show not the average possessions per team per game, but the average total number of possessions for both teams in each club’s games. So you could see if, as I suspect, on average a Bolton game has far more possessions overall than a Manyoo game. It seems fairly obvious that it should, because (unless I’m missing something) a team has to have almost exactly 50% of the total number of possessions in any game, given than any change of possession necessarily gives the other team the ball.

5. Would it not make more sense to map time in possession to goals. As has been stated, number of possessions can increase for many different reasons.
   Using time in possession while ball is in play may show a correlation you expected.

6. Hey,

   Love your blog. Tremendous insights. How are you defining possessions? What data source are you deriving these #’s from? As you may have seen from my website, these are questions that I am struggling with right now and I am interested to hear another take. Your possession #’s are significantly higher than what I’ve found so far in MLS games. Perhaps the EPL is that much more direct?

Comments are closed.