For the past 8 years, the Spanish/Barcelona possession-based philosophy cast a long shadow over the soccer world, inspiring imitators and reactive trends alike. Of course, there are many ways to attack, and teams such as Atletico Madrid and Leicester City have thrived recently without dominating possession.
While possession dominance hasn’t correlated to MLS success, it is still important to have individual players who can successfully complete a pass and retain possession when needed. The expectations for how many passes a player should complete will vary by role on the field and a team’s tactical plan. Let’s look at multiple passing statistics from the 2015 MLS season for a better sense of the league norms. Our base for this analysis is 302 players from the 2015 MLS regular season, all of whom played at least 800 minutes (data from whoscored.com).
The first chart shows total passes attempted per 90 minutes by the percentage of passes completed. The bigger the circle size, the more total passes that player attempted during the season. I’ve also color-coded these by position, with forwards in light-blue, attacking midfielders in blue, midfielders and defensive midfielders in tan, and defenders in red.
The upper-right of the chart features the metronomes. Osvaldo Alonso, Wil Trapp, Dax McCarty, Gonzalo Pineda, and Andrea Pirlo all attempted more than 70 passes per game, and completed more than 80% of those passes. There is a moderate correlation between passes attempted and pass completion percentage (R2=.37).
I also looked at the types of passes played. For example, a holding or defensive midfielder in one scheme might require this player to circulate the ball quickly, while another plan requires frequent diagonal passes to the wing. Here is passing completion rate by short passes and long passes (any passes 25 yards or longer), and filtered by players who attempted at least 30 long passes during the season.
On average, a player in this data set attempted 670 short passes and 96 long passes. Said another way, less than 13% of the average player’s passes are long balls. Players in the upper-right are players with above-average passing completion percentages for short passes and long passes. I found Michael Parkhurst’s performance particularly impressive, considering he attempted 326 long passes (3rd most in the league.) Attackers in the upper-left like David Villa, Obafemi Martins, and Octavio Rivera are also distinctive. These players likely play more difficult short passes in attack, decreasing their overall passing percentage, but they have the skill to connect on a long ball.
With the rise of the counter-press, there’s additional emphasis on not turning the ball over with numbers committed to attack. Whoscored.com defines “dispossessed” as being tackled by an opponent without attempting to dribble past them. When dispossessed, a player loses the ball before even attempting a pass.
The ideal analysis would determine who best retains possession under pressure (as Cruyff said, anyone can play with 5 meters of space.) For a partial look at ball retention with available data, here are total passes per 90 by times dispossessed per 90.
Defenders tend to get dispossessed less often than midfielders and attackers; few defenders lose the ball once a game or more while in possession. Any defender losing the ball once a game or more often while in possession needs to protect the ball more. There is a wide range of dispossession rates among attackers and midfielders. For example, attacking midfielders Pedro Morales and Federico Higuain played roughly the same number of passes per game, but on average, Morales lost the ball one additional time per game.
So far, we’ve looked at total passes, pass success rate, short and long passes, and players losing the ball while in possession. Now let’s look at dribbling ability among MLS players last year. Fundamentally, a player in possession is either going to pass, dribble, or shoot, and the first two occur everywhere on the field. Evaluating passing and dribbling together can also provide some indicators on a player’s tendency. (For those interested, shooting and finishing is covered extensively here, here, here, and here.)
With our same player sample, here are the total dribbles attempted per 90 by the total passes attempted per 90.
The immediate impression is that Fabian Castillo and David Accam will dribble your face off. Kwadwo Poku, Juan Ramirez, and Kekuta Manneh were the only other players to attempt 5 or more dribbles per game last season. The adventurous red circle in the middle is Alvas Powell, with nearly 4 attempted dribbles per game. Among players developed in the US, Khiry Shelton (just to the right of Powell) stands out with 4.4 attempted dribbles per game in limited action last season.
We don’t just want our players to attempt a dribble, though – we want them to get past a defender. Here is total dribbles attempted per 90 by the dribble success rate. Like shots on goal, dribbles attempted occur far less frequently than passes attempted. However, I kept all players in this chart to highlight the trend.
Whereas players who attempt more passes also complete more passes, dribbling success rate converges around 50% (the league average was 49% in 2015). That Castillo maintained a 50% success rate while attempting the most dribbles in the league is remarkable. Future analysis should look at dribbling success rate over time to determine how repeatable the skill is, and if, hypothetically, players with high success rates like Poku or Darlington Nagbe should attempt even more dribbles.
Finally, we can tie this together. Thom Lawrence recently introduced a great metric, looking at the time until a team takes or concedes a shot. Part of what prevents a team from taking a shot is retaining possession (Manchester City a prime example from Lawrence’s analysis.) Let’s apply this principle at an individual player level.
Here, we have a base of total passes attempted and total dribbles attempted. A successful action is completing a pass or a successful dribble. I’ve grouped “lost possession” as any time a player does not complete a pass or dribble, any time they are dispossessed, or any “unsuccessful touches” that concede possession through an unforced error.
Again, positional context is important here. In general, forwards are touching the ball less and turning the ball over more. An overlay of end product, like goal-scoring chances taken or created, would determine if the successful actions are worth the turnovers. Nat Borchers rarely turns the ball over on the backline. At the lower-right, our original group of five central midfielders varies slightly by total lost possession, with Alonso and Trapp separating themselves from the other three.
- Players who attempt more passes tend to complete a higher percentage of their passes.
- Evaluate the types of passes a player makes. In particular, certain forwards are likely better passers than their overall passing accuracy shows.
- A player’s success in passing or dribbling is best evaluated against others in the same position.
- Attacking a defender through dribbling happens infrequently. Actually getting by that defender occurs about half the time. Exploring the risk/reward equation of dribbling and repeatability of successful dribbling could yield new insights.
- Future research could look at a sequence of events for attackers, to determine if the higher lost possession rates are a byproduct of chances created or just wasteful.