“Skilsinho?” Redefining what a successful dribble is.
/By Kevin Shank (@Kev_Shank)
Skill moves are one of the many great things about soccer. Watching the players twist and turn with the ball seemingly attached to their feet not only makes for fun highlight reels, but losing a defender also gives an advantage to the attacking team. As exciting as it is to see a player nutmeg another, it is equally disappointing to see him take one too many touches or misplace the following pass, squandering the effort put into the successful dribble.
As a Philadelphia Union fan, I have seen many times where Ilsinho uses his Brazilian footwork to dance around defenders, gaining the nickname “Skilsinho” from fans. Early in the 2016 season Jim Curtin lauded Ilsinho’s skills saying, “He catches the eye. He is a great 1v1 player, beat guys off the dribble which is a great skill to have.” And Curtin is not necessarily wrong since Ilsinho’s dribble success rate of 44.19% is just above the league average (43.37%) and better than the likes of players like David Villa (42.55%). So does this mean that Ilsinho is a more effective dribbler than David Villa? Well, not quite since Ilsinho often falls into that category of players I described above who will dazzle then disappoint with his footwork.
While both Ilsinho and Villa successfully completed their dribbles, it is easy to spot the difference in their effectiveness as Ilsinho misplaced a pass and as Villa assisted Jack Harrison (0.4469 xA). Although these are just two examples of a good and bad dribble, we need a way to empirically measure effective dribbles for every player across the league. For the 2017 season, I assembled every pass, shot, foul, and defensive action to see what event happens after a successful dribble and whether that event was completed or not. Thus for my Effective Dribbling model, in addition to a successful dribble a player needs to complete the next action.
In this model, the league average for Effective Dribbling was 26.01%. In addition, I looked to determine player expected goal production; that is, a successful dribble that immediately turns into a shot or key pass. Although the Top 5* in xG+xA have around average Effective Dribbling rate, the scoring opportunities created show the danger they add after beating a defender. It is also interesting to note that players are out-performing their xG off of dribbles, like Fanendo Adi impressively scoring on all four of his shots after a dribble (only 1.23 xG). Furthermore, across the league, players have scored a total of 44 goals from Effective Dribbles, vastly out-performing the 22.52 xG from their shots.
*Players with 20 or more touches
| Player | Effective Dribble % | xG+xA | xG | Shots | Goals | xA | Key Passes | Attempted Dribbles | Dribble Success % |
|---|---|---|---|---|---|---|---|---|---|
| 0.35 | |||||||||
| 0.2 | |||||||||
| 0 | |||||||||
| 0.53 | |||||||||
| 0.28 |
Not only is Villa producing more than Ilsinho’s .168 xG+xA, his 29.79% Effective Dribbling rate is still better than Ilsinho’s 18.60%. So while Ilsinho’s footwork captivates the fans, his numbers leave much to be desired in terms of actual production. This then begs the question: what players have the highest Effective Dribbling rate?
Interestingly, the Top-10 success rates out of players with at least 20 dribble attempts are all central midfielders. This list isn’t filled with players whose footwork with fascinate the crowd, but rather players who will escape a challenge while pulling their team’s strings.
| Row Labels | Attempts | Adj Success % | Dribble Success % |
|---|---|---|---|
Here is an example of league-leader in Effective Dribbling rate, Dax McCarty, sidestepping from Cristian Higuita’s challenge and playing a pass forward, which is a play that one would expect from this list of midfielders.
While this model was able to identify dangerous and effective dribblers, it is important to acknowledge that all dribbles are not created equal. As one would expect, a dribble in the attacking half is harder than a player’s own half (40.1% and 51.8%, respectively). Producing an expected dribble model would identify how successful a player’s dribble was relative to his position. In addition, Jared Young has showed that all passes are not created equal, so it would be beneficial to include this, too. While these additions should make the model more accurate, the Effective Dribbling 1.0 still shows that sometimes it is necessary to delve deeper into the stats.