Interesting Ball Movers From PL’s First 5 Games

jonny

 

A bit of leftover stuff from the United/Liverpool article  led me to flesh it out and take a look at the different positions so far and players who are progressing the ball well compared to how often they lose it via pass. I broke the pitch down into 7 positions, these could be done better because it’s simply where you play your passes from: so Mitrovic looks like a midfielder for example.

It’s simply a chart showing how often players progress ball into a forward zone (which yes I know, is often arbitrary) via pass or reception compared to how often they lose the ball via pass. Snip20180919_32

A list of players who progress the ball often and rarely lose possession compared to their position-mates, with the ones who are very interesting/surprising to me in italics

Attackers

Sergio Aguero

Roberto Firmino

Mohamed Salah

James Maddison

Midfielders

Aleksandar Mitrovic

Mesut Ozil

Jeff Hendrick

Demarai Gray

Jack Wilshere

Lucas Torreira

James Milner

Abdoulaye Doucoure

Deep Midfielders

Nampalys Mendy

Marouane Fellaini

Granit Xhaka

Defenders

David Luiz

Harry Maguire

Jan Vertonghen

Willy Boly

Conor Coady

Mamadou Sakho-king of these types of viz

Wolves 3 at the back might be sort of boosting the numbers of Boly/Coady

Fullbacks

Ryan Bertrand

Jonny (a guy I  loved at Celta, still showing up at Wolves. I once called him Wendell but for a cheaper price)

 

Snip20180920_64Snip20180920_63Snip20180920_62Snip20180920_61Snip20180920_60Snip20180920_59Snip20180920_58

Advertisements

Manchester United vs Liverpool Ball Progression Analysis: The Jesse Lingard Effect

jesse-lingard-680034

2018-04-28-158-liverpool_stoke-600x400

This is kind of a test, to see if I can write down and explain something I’ve found quite illuminating to myself. I’ve been looking into ball progression a lot (like here) and wanted to try and get a holistic view of how teams moved the ball from basically their own half into the danger zone. From there shots come into it more and it requires a different type of analysis. Right now I’m breaking the pitch down into separate zones and looking at how teams move through them up the field. Snip20180919_32

I looked at each teams 8 attacking players first off (front 6 and 2 fullbacks). I looked at their usage rate by zone, basically where they play their passes from:

Snip20180920_54Snip20180920_53

A few things stand out here to me. First, I misspelled Salah’s name, but give me a break I’m out of practice and it was very late at night. Second, I correctly spelled every other name, so give me credit (except Mané depending on how strict you are with French in English).

Both nominally feature a front 3 but when Lingard is up front for United he plays clearly a step behind Sanchez and Lukaku, with an offensive touch profile more like Keita than one of the other 5 forwards. United in general don’t get as far forward as Liverpool, the fullbacks and the Fellaini/Fred vs Milner/Wijnaldum comparison both come out with the Liverpool players further forward. And Lukaku is really dropping deep a lot, he has a higher share of touches in zone 4 than any other player has in any zone.

 

Next I wanted to look at how the teams are advancing through the zones, it’s fine to complete 100% of your passes within the zone or backwards but you eventually have to move the ball forward if you want to score. So I looked at the average pass each player made by zone and then how often he progressed a zone. So taking an example here of Manchester United starting in zone 6:Snip20180920_45

 

The colors tell you how often they complete a pass moving the team forward, so Valencia in this case progresses the ball into forward zones 43% of the time he passes, the league average from zone 6 is 32%. Fellaini is at 43% as well, Fred 33%, Shaw plays conservatively at 27% and Pogba plays very short passes advancing the ball 21% of the time. It’s conceivable this is him dropping back under pressure, but that would take a little more work and is a separate post.

 

So then I just looked across the different zones as the two teams made their way toward goal…the images are below.

 

This slideshow requires JavaScript.

 

 

This slideshow requires JavaScript.

 

My takeaways:

-Look at Zone 4 for Manchester United and you will see why I called it the Jesse Lingard effect. Not only does he drop deep, but he’s dropping deep to keep the ball deep. Lingard’s most common zone he plays in is Zone 4 and his average pass is actually backwards. He progresses the ball only 12% of the time from that zone, less than half of the league average of 25% and even lower than strikers Lukaku/Mane/Firmino who presumably often don’t have players in front of them to pass to. Keita for example who probably had the most similar profile to Lingard, advances the ball 33% of the time from Zone 4.

-United haven’t quickly advanced the ball through zones 4 and 5 using central players: Valencia and Shaw have mostly been fine but Fred/Pogba/Fellaini/Lingard all progress the ball less often than the average player. That’s not necessarily a bad thing: David Silva is one of the lowest in the league, but he isn’t trying to often and when he does, it’s always complete: these guys are and often lose the ball much more than the Chelsea’s and Man City’s who are so conservative.

 

-Wijnaldum has a bit of the Lingard in him, he almost never progresses the ball but Keita and especially Milner have been successful, moving the ball forward much more often than the league average. Liverpool’s attacks can flow quickly through those two where United don’t have central players who have done that.

 

-Liverpool also have fullbacks who are successful, Alexander-Arnold and Robertson progress the ball even better than Shaw and Valencia.

 

-Most know by know, but James Milner is a hell of a player. I once wrote Milner was not suited for midfield and remember him roaming aimlessly early on at Liverpool but he certainly seems to have settled in this season and is arguably the best among all these big-name players like Keita and Pogba at progressing the ball forward.

-Firmino and Salah have been great so far passing from zone 3, unlike Lingard and Sanchez for United. Mane seems to have basically been a shot hog so far this year, probably can have one of those.

 

Next piece will build on this and add in ball losses to try and flesh out the analysis. Might look at Man City vs Spurs.

 

 

Final charts simply show progression% from a zone, so Pogbra successfully progresses ball forward on 15% of his passes from zone 3 while Fred does it 24% of the time.

Snip20180920_56Snip20180920_55

Do Huddersfield Have a Build-Up Blockage?

christopher-schindler-huddersfield-town-football_4110171

terencekongolohuddersfieldtownvbirmingham-ij9k1b9ty9l

 

I was crawling deep through the early-season data as one does on a Champions League Wednesday and found something a bit odd that deserves a little investigation. First, I had broken up the pitch into 9 zones, seen below:

 

Snip20180919_32

I was looking at how teams move from zone to zone as they build-up before they get deep into the box, basically from the back until they get to zone 3. There were standouts on both ends: Man City and Liverpool’s efficiency and Cardiff and Newcastle’s hopeless bashing but Huddersfield around zone 5 stood out oddly. Here a few graphs to show what caught my eye:

Snip20180919_33

Most teams play more passes from Zone 5 than Zone 6, Huddersfield play about half as many. So I looked back at what was going on in Zone 6 a bit:

Snip20180919_34

and found that Huddersfield simply aren’t trying to move the ball forward from Zone 6 at all. Man City and Chelsea are 2nd and 3rd lowest in this list but when they do attempt to progress the ball they are the 2 most efficient teams at it (93 and 86%). Huddersfield is 16th at 67%. I wondered if this was a problem just completely across the back so I looked at passes starting in Zones 7 and higher to see if a similar trend continued.

Snip20180919_35

And no it did not. Huddersfield aren’t trying a play it around the back type Chelsea/Arsenal/Man City strategy, but something odd is going on once they get it just off the back into zone 6. Sample size wise, there are 369 passes starting in Zone 6, so it’s not like we are looking at shot-level scarcity here in the data.

 

So why is the ball “sticking” in Zone 6 for Huddersfield? To investigate I looked at each teams main two passers in that zone. These were mostly center backs but there were a handful of central midfielders sprinkled in (Xhaka, Gosling, Doucoure, Stephens, McDonald, Noble, Milivojevic, Ralls, and Diame were the 9 midfielders with the remaining 31 being center backs). I charted how often they attempted to progress the ball into a forward zone vs how successful they were at it:

Snip20180919_36

Of the 40 players, Christopher Schindler is the most conservative. Under 30% of the time he attempts to move the ball forward (average is 45%) while he is also one of the least successful passers. Only Diame, Gosling and Holgate complete a noticeably lower% of passes. Terrance Kongolo, the other Huddersfield centerback, you can see is also one of the most conservative and least successful players.

 

So any answer as to why Huddersfield can’t get out of zone 6 has to include Christopher Schindler. I wondered if it was a tactical tweak that led to his numbers being thrown all out of whack, after all we are just 5 games into the season. Huddersfield played a back 3 vs Crystal Palace, Chelsea and Everton while have gone back 4 vs Cardiff and Man City. Maybe the back 3 vs back 4 numbers look starkly different. They don’t look any different for Schindler really: 20% of time he tries a Zone 6 pass he successfully progresses the ball forward no matter if he’s been in the back 3 or back 4. No matter if it’s been a back 4 or 3 however, Zanka has not had the problems that Schindler and Kongolo have…he is moving the ball forward basically where you see Doucoure on the graph above. He’s conservative but only slightly, while succeeding more than the average. The evidence seems to point to the fact that Huddersfield are putting too much on Schindler and Kongolo’s plate, and they simply aren’t good enough passers to move the ball forward into Zone 5.

 

Perusing World Cup Passing Data: The Fascinating Case of Mario Fernandes

So I was looking for the “best” and “worst” pass of my precious little types of pass* to center a piece around them, but when I looked at the top and bottom 10 passes for players I found a can’t miss hook: Mario Fernandes was on both lists. The “Russian” right back (he was born Brazilian and Putin decreed citizenship on him) was most famous for missing the goal entirely in the shootout against Croatia in what was a devastating loss for the neutral who surely was pulling for the Russians, after all they’ve been through with such a home crowd behind them roaring them on, flying all over the field, giving their all. But I digress; this is about Fernandes. He was known for that painful penalty miss; but maybe no longer after this article catches fire.

*each pass is assigned to one of 30 “pass types” for easier summarization and analysis

 

The best pass list generally features players from the best passing teams: Toni Kroos is on there, Sergio Ramos, and Jan Vertonghen twice. No surprise really. But then for some reason, two different Russians are on there, both crossing the ball from the right wing. They both are completing crazy high rates of this pass:

Snip20180808_8

Then here he is, back in his own half, putting up numbers like he’s playing for Cardiff facing Man City. Snip20180808_10

So what is going on? My first thought was basically the clustering was too broad and the Russians were sort of gaming the system by playing short or easier type passes than the normal pass in cluster 29 while doing the opposite in cluster 5.

 

 

World Cup Most Passes Gained, The Rest

Full explanation of the method and breakdown of attackers is here. Now we move to the rest of the pitch

 

50-55 Yard Range For Pass Origin, 78 players

Dominated by the Spanish.

Snip20180814_60

Snip20180814_61

Snip20180814_62

  1. Jordi Alba +30 completions (+8% vs avg)Snip20180814_63
  2. Sergio Ramos +30 completions (+6% vs avg)

Snip20180814_64

3. Gerard Pique +25 completions (+8% vs avg)

Snip20180814_65

 

Interesting to see the differences between Pique and Ramos at centerback.

 

More interesting is who we see at the bottom

Jordan Henderson -12 completions (-4% vs avg)

Snip20180814_66

I’ve long felt improving on Henderson’s passing might be a way for Liverpool to make that next step, maybe it’s true for England as well. The midfield of Alli/Henderson was a passing disaster for England and their most glaring offensive weakness.

Snip20180814_67

 

 

 

Defenders, 55-60 pass origin, 84 players

Snip20180814_69

Snip20180814_70

Snip20180814_71

Not sure exactly how much of this is an English strength and how much is it sort of compensating or picking up or taking away completions that would have gone to a midfield. Maybe playing 3 at the back makes it easier for these guys to complete passes but then robs Henderson/Alli of options making their numbers look worse.

 

Stones

Snip20180814_72

Maguire

Snip20180814_73

Walker

Snip20180814_74

 

One interesting standout as the highest % gainer was Guillermo Varela. Not only did he stand out among his position mates, he was Uruguay’s best passer by miles. No unauthorized tattoos during this Cup, just nice and neat passing.

Snip20180814_76Snip20180814_75

 

Deep Defenders, >60 yard origin for passes, 62 players

Snip20180814_79

Snip20180814_80

Let’s look at the 2 Colombians, because that is a fascinating difference. First, Yerry Mina (+8 completions, +5% above average)

Snip20180814_82

and now Davinson Sanchez (-10 completions, -5% below average)

 

Snip20180814_81

 

Sanchez was much worse at playing long balls, and short passes as well. It does make me wonder if it is connected to what I wrote in my first post on the WC series about Colombia where they simply didn’t have a left-side in attack (3 attacking midfielders played 7 attacking passes into box from the left side vs 48 from the right). So when Sanchez looked long, there literally was no one there. Compared to when Mina looked forward, he had an abundance of options to look toward. Without video it’s just a theory, but the numbers are incredibly stark here.

 

There is another pair of defenders here (along with one from previous group) who stand out in their team totals. Anderson Santamaria and Luis Advincula (from Peru) really stood out compared to their teammates)

Snip20180814_85

First Santamaria:

Snip20180814_87

and then Advincula

Snip20180814_88

and so did Romain Saiss of Moroccco.Snip20180814_83

 

Snip20180814_86

World Cup Most Passes Gained By Position, Attackers

So this is a simple post, mostly numbers and charts and minimal analysis. How I calculated “most passes gained”:

 

  1. assigned every WC pass into one of 30 “types”, seen below:Snip20180808_11

2. Calculated completion% for each type.

3. Compared player’s completion% for each type pass to the overall completion% to see how many “completions gained” he had. So for example, pass cluster 4 (the short pass forward in the right attacking half is completed at 72% rate. If a player attempted 10 of these and completed 8, he would be credited with 0.8 completions gained.

 

I never view these as “player skill” but more how a team expresses itself.

 

Attacking Players (Those whose passes started within 40 yards of goal: 16 players with 40+ passes)

Very few players in this group.

Snip20180814_45

  1. Mesut Ozil +12 completions (7% above average in percentage terms)

Snip20180814_43

The bluer the color, the better than average that type of pass was for Ozil.

3. Raheem Sterling +4 completions (3% above average)

Snip20180814_44

Now this is a fascinating case. Sterling was the hot-topic on stats twitter during the World Cup as he was bashed for producing essentially nothing by many fans and credited for linking and producing much of England’s attack with his movement by much of stats twitter without evidence in response to what they viewed as racially tinged attacks over the years. This map kind of gives credence to both sides. While Sterling rated as one of the best “forwards” in passes gained, they all come from the blue marks, basically in midfield. Then when it’s time to move the ball into the box Sterling was a big negative. Now of course all these maps (as with ~everything in soccer stats) are about the entire network around a player and not individual profiles, and that should be in mind for every player, but this map revealed a lot to me.

Attacking Players Level 2 (Those whose passes started 40-45 yards of goal: 54 players with 40+ passes)

Snip20180814_47

Snip20180814_48

A big reminder of positional effects here and limits of this type of model telling you about individual passing skill without intertwining all kinds of other effects. Good thing this isn’t measuring passing skill but illuminating the World Cup.

  1. Isco +27 completions (+6% vs average)

Snip20180814_49

 

Heavy volume in the mid-attacking left side.

 

2. Jesse Lingard +18 completions (+8% vs avg)

Snip20180814_50

Actually seemed to pass better the closer to goal he got.

3. Philippe Coutinho +18 completions (+6% vs avg)

Snip20180814_51

A varied profile here as he was all up and down the left side, but extremely strong at passing into the box.

Now for the bottom of the list.

Ante Rebic -22 completions (-17% vs avg)

Snip20180814_52

Antoine Griezmann -19 completions (-9% vs avg)

 

Snip20180814_53

 

Team effects, guys out there for shots? Bad passing numbers can’t be gotten around. I could not believe the transfer rumors for Rebic after watching him closely due to my DraftKings lineups hinging on him a couple times, these numbers make me even more puzzled.

Attacking Players Level 3 (Those whose passes started 45-50 yards of goal: 80 players with 40+ passes)

 

Snip20180814_54

Snip20180814_55

Snip20180814_56

  1. Toni Kroos +28 completions (+9% vs avg)

Snip20180814_57

2. Axel Witsel +17 completions (+6% vs avg)

 

Snip20180814_58.png

 

Quite different profiles for Witsel and Kroos, probably caused by Germany’s extreme field tilt basically at all times.

 

Hopefully I can find some players who stick out amongst their teams in a future post. The problem with diving into data like this is you get 20 ideas for every one post you put out. Next though I keep going further back from goal. Dividing wide players from central players is not something I am doing for the sake of simplicity now.

Perusing World Cup Passing Data: Time and Game Effects

Before I get to hopefully future pieces on possibly: Senegal, Mexico, Mario Fernandes, Dele Alli, Thomas Muller and more we look at a bit of game state and time state effect on passing patterns in the World Cup.

 

A reminder, the 30 “passes” you see on each field below are just types of passes. Every pass played is assigned to one of these types of passes for the sake simplicity of analysis.

 

Ahead

Snip20180810_40

I think this is kind of surprising. I attempted to correct for team quality here by comparing “expected passes” of a certain type when ahead/behind by looking at teams tied rates. I didn’t expect long balls and own half passes to be so clearly more common when ahead.

Behind

Snip20180810_39

 

 

Late/Early

 

The next two graphs are the same numbers plotted differently. Both represent how often passes are played early in a tied game state compared to how often they are played late. Early is minutes 1-20, late is minute 70 on.

Snip20180810_36

Snip20180810_35

 

The later the game gets, even if tied, the further forward the center of gravity gets from the perspective of the attacking team. This isn’t new for me really but it’s interesting to me nonetheless and gives me chance to link to own piece in past and improve 10% on what seems to be a fine viz to me but usually leaves 25% of people looking at it a bit confused. Teams pass it back in their own half a lot more at the start of games when the game is tied than they do late in games. Why do they do this?