I did a deep-dive here on passing and dribbling stats from last season: https://andthenthehex.wordpress.com/2016/04/20/evaluating-2015-mls-passing-and-dribbling-success/ I found some interesting stuff during the research. For example, I hadn't spent much time looking at dribbling success rates. Here is the player-by-player overview: What I'm most interested in is the repeatability of this stat. Are players like Nagbe and Poku exceptionally good at taking defenders on 1 v. 1? Or is there a natural convergence toward a 50% success rate, and they should actually be taking defenders on more often?
Really good stuff. Even though some of the conclusions were where I think you would get to intuitively, you have put them on a solid empirical footing. I'll give it another read through and post some other questions that you might want to explore. Thanks for linking it
Thanks! I'd definitely appreciate more thinking on this one, particularly two areas: 1) The dribbling thing mentioned above. By the way, Castillo's (and Accam's) numbers are outrageous, even in a global context. This season, here are the league leaders for some other countries (per game, not per 90): England - Zaha, 3.7 Italy - Perotti, 3.6 Germany - Douglas Costa, 3.5 France - Boufal, 4.3 2) The platonic ideal for passing accuracy by position, and placing that in context of a team's ability to create chances.
I would like to see chances created included somehow. Perhaps something like "successful actions" (passes and dribbles) and their correlation with chances created. That way I think you'd see who is flashy but wasteful and who is substantial and successful.
Great stuff. Because dribbles and dribbles per 90 are correlated your plot looks how you expect, but I think you really want dribbles and not dribbles per 90. Start with your hypothesis that each outcome is 50/50. You should see the relationship you plot and in fact you could calculate curves that show say the 90th percentile of the distribution for each number of dribbles. It would probably be easier to just calculate the numbers and look for outliers. For example, given the total number of dribble attempts for Darlington Nagbe, what is the probability that you would observe a 68% success rate if the true underlying success rate was 50%. (bearing in mind that, if you have enough players, you will always generate outliers).
Thanks, and good thoughts. I totally agree that there ideally is some end product associated with a successful dribble. What I need to locate is a source for sequential data. Sticking with Castillo...right now, I can determine how many assists, key passes, shots on goal, fouls suffered he created. However, I can't determine how many of those occurred immediately after beating a defender.
A dribbler's success greatly depends on his position on the field. For example, even toward the end of their careers, both Zidane and Riquelme were exceptional in maintaining control of the ball but they did it in the middle of the field where they had not only a passing outlet but space to dribble into. Contrary, someone like Garrincha was usually pinned to the sidelines and was often double- or even triple-teamed. Additionally, Nagbe, a deep lying central midfielder, will usually possess the ball on the own half of the front lines, as it were, with the opponent density largely in front of and yet away from him while Castillo or Givinco are positioned much closer to the opponent's goal and their positional marker. With the above in mind, I find it much more useful to break down the stats according to a player's position, i.e., wingers vs. wingers, strikers vs. strikers, deep central midfielders vs. DCM, etc.
Remember how Jeff Agoos was one of the best defenders in the first decade of MLS - but was almost always the guy getting juked or burnt in the highlights? It would be easy for one's impression of Agoos to be at odds with the statistical evidence of his contribution. So maybe you can find some stats to prove or disprove this assertion, but when I used to watch MLS highlights - or rather the appalling lowlights that make you cringe with embarassment as an MLS fan - I didn't consistently see that guys couldn't dribble or pass, or that they had uncultured soccer brains. Most negative stereotypes about MLS teams are oversimplistic overgeneralizations. But something about the water or our culture or probably Obamacare results in American soccer teams not being able to clear a ********ing ball to save their dicks. That seems true again this season.
Agreed, and I've tried to address this somewhat with the color-coded positions (as an aside, WhoScored data is brutal for determining a single position for a player.) With that in mind, guys like Poku, Powell, and Nagbe stand out for their success rate/adventurousness relative to their peers. Of course, if I were working for a club, this would be step 1. With the smaller number of total dribbles per season, it should be easy enough to evaluate all attempted dribbles on video. One could then overlay defensive pressure and field location metrics to add more of the context you mention above.
Unfortunately, I don't think we're getting this data unless someone had the desire to retroactively chart games. MLS signed with Opta in 2011, and it looks like the earliest available public stats are from 2012 or 2013. Even in Europe, the "enlightened era" starts around 2009. Footballoutsiders.com had the same problem, and literally has a team of volunteers providing detailed game recaps for late 80s/early 90s NFL games. If you're interested, I wrote a broader article on the role of stats and analytics in soccer here: https://andthenthehex.wordpress.com/2016/03/02/analytics-scouting-and-mls-attacking-tendencies/
BTW, comparing C. Ronaldo to Messi (you may have heard of those two ... if not, look them up in Wikipedia). An opponent's defensive belt is about 30 yards deep, from its own penalty box line (18) and extending to about 50 yards away from goal. Ronaldo will play the last 5-10 yards and often line up at the shoulder of the last defender. His dribbles are very often one v one and toward the open net. Messi will often line up the first 5-10 yards, those furthest away from the goal, then dribble into the heart of the defense, thereby attracting 2-3 additional markers. His plan is then not to dribble through everyone but to pull the defense out toward him, destroy its shape and play two types of passes - against the grain/cutback to Suarez darting in on the right edge of the box or across the field to Neymar at the left edge of the box. Messi is rarely ahead of the ball; Ronaldo is always ahead of it.
Lots of good stuff. OPTA ran an ELO evaluation on aerial challenges in MLS to determine who was the best in the league and it seemed better than just raw success rate. I'd love to see it applied to dribbling and tackling.
You can see an immediate correlation between individual speed and a blue circle outline. This helps! More blue circle outlines.
Tommy McNamara is the best dribbler in MLS. Graham Zusi is the set piece taker and Servando Carrasco is the best passer. Enuff said.