While watching the Brazil/Germany World Cup semifinal, I was thinking about how possession is calculated. We only see possession as a percentage figure, but that doesn’t provide a very interesting look at how possession changes throughout a match. What would this data look like over time?
I started recording when possession switched from one team to the other by using the “lap” feature on a running stopwatch on my iPhone. At the end of each half I recorded the length of each lap, assigning the even numbered laps to team A, and the odd numbered ones to team B.
But before I get into the results of this data gathering, I should explain my definition of “possession” that I used to determine when I would hit the lap button. For lack of a better term, I’m calling it “Real Possession,” but that’s not to say my version is any more “right” than what the official stats do.
How do we define possession?
I do not know how FIFA defines possession, and there are many questions as to what actually defines it. If a goal kick by Team A is intercepted by Team B winning an aerial duel, but the resulting header lands at the feat of a player for Team A, did possession switch between Team A, to Team B and back to Team A? Or was it always in Team A’s possession? What about when a defending team wins a throw-in deep in their own zone, only to have the throw intercepted by the attacking team? Did the attacking team ever really lose possession?
I tried to take these situations into account when deciding when to hit the “lap” button on my stopwatch. When Team A attacked Team B but eventually out the ball out for a Team B goal kick, I didn’t switch the possession to Team B unless the goal kick resulted in Team B controlling the ball. In general, I defined controlling the ball as having the ball at your feet or completing a controlled pass to a teammate.
Just like controlling a goal kick, simply intercepting a pass or a cross with your feet or your head isn’t enough if your first touch goes right back to the other team. And similarly, possession doesn’t change in my eyes if the ball is bouncing from player to player and team to team in a frantic scramble without any real control of the ball.
What this does is provide a look at “Real Possession” or a team’s control over the game at a given time. I’ve visualized how this played out in the surprising 7-1 thumping Germany gave Brazil which you can see in the image above.
Looking at Brazil/Germany
With Brazil in yellow and Germany in black, you can see the different times throughout the match when one team controlled possession for large chunks of time, or when they gained control only to give it back seconds later. You can see that Brazil got off to a terrific start in terms of possession, controlling the game for roughly 72% of the first 7:30 of the game, even keeping the ball for 143 straight seconds (2:23). But eventually the half became more even as the teams traded short and medium chunks of possession.
Even after Germany’s first goal, things still remained even on possession terms, but you can see the life leave Brazil as Germany scored its second goal and the following three goals of the half. One area of note (see below) is the time between Germany’s 2nd, 3rd, and 4th goals. Between those goals, Brazil only managed to hang onto the ball for a total of 33 seconds after kicking off at the center circle. What’s even more impressive is how Germany produced 4 goals on 5 possessions, all around 6 minutes of play.
Later in the half, you see Germany begin to dominate possession as Brazil react to their collapse. For a 10 minute period starting shortly after Germany’s 5th goal, they begin to completely control the game with 82% of the possession during that time. This includes huge chunks of 190 (3:10) and 155 (2:35) consecutive seconds of control.
In short, the second half was more even as Germany took its foot off the accelerator a bit, and Brazil perhaps tried slightly harder by producing better chunks of control at the beginning of the half. But you can see Germany closing the match out strongly with three large periods of control ranging from 143 (2:23) to 207 (3:37) seconds in length.
Both teams had large patches of possession, resulting in 30% fewer possessions per team in the second half from the first. This is also evident from the increased average length of possession for each team in the 2nd half. See the table above for a detailed breakdown of how this all played out over the course of the match.
I find this way of visualizing possession interesting to see, and could make for a better viewing experience when calculated in realtime during a match and displayed from time to time on TV. It gives you an instant view of the flow of a game in a way simple total percentages can’t.