Having previously looked at vertical passing in relation to Everton last season, one of the main issues was with defining a vertical pass. The method I used said a pass couldn’t be more than 10 units either side of the start point and must be forwards. This meant a lot of passes that weren’t actually vertical managed to slip through the cracks though, particularly short distance diagonal passes.
Since then I’ve tried to implement a new method using polygons and checking if the end point of a end pass lies within the polygon made. To do this I defined the polygon/triangle as the start point of the pass and the second two points being at the end of the pitch 20 units either side of the start point.
Now, there is a decent amount of assumption going on. In R I used the ‘point.in.polygon’ function and I’m mostly just hoping that when given the 3 points it creates the triangle I want it to. Then, once I had these 3 points I just had to check if the point lay within the polygon.
The points had to be ‘strictly interior’ to the polygon in order to be deemed inside and therefore vertical. To be honest, I’m not 100% on what to think about a point ‘lying on a relative edge’ to know whether to include it or not, while it’s highly unlikely that the end location is one of the three defined points.
If all goes to plan though, if a pass is made from the centre circle it will be deemed vertical if it falls within the badly drawn triangle below:
This is a huge improvement from the previous area, but also meant that the numbers were significantly down. Due to the numbers being down I’ve also been slightly more lenient on the wide area rule. I’m only counting passes that don’t end within 15 units (15%) of either touchline. This meant the numbers were down further but tended to weed out the full-backs who can rack up vertical passes by moving it up the line slightly.
Unfortunately I haven’t had chance to roll this method out across older seasons (I do it a really stupid way which involves me looping through all the rows of passes), so I haven’t had chance to see if what I said about Everton last season holds up or not. I’ll probably look into this another time, as it’ll also be interesting to see if there’s any changes under Sam Allardyce.
Before getting into some highlights, a few questions I was thinking while doing this were ‘does it matter?’, ‘will this help assess players or teams?’ and ‘is it linked to performance in anyway?’. And the answer to these seems to be no. It has practically no link to xGD from what I checked and there’s plenty of talented players – and lots with high xP Ratings – who don’t make many vertical passes.
It seems more a stylistic choice than anything else. Particularly when looking for defenders, it can help break down the types of passes they make. So do you want a ball-playing centre-back who recycles possession, sends long-balls to the flanks or one who is comfortable bringing the ball out of defence and making line-breaking passes? This is probably the most useful it’s been.
An example case of this is within Serie A, between Napoli’s Kalidou Koulibaly and Lazio’s Stefan de Vrij.
They both have exceptional xP Ratings, Koulibaly with 1.132 and De Vrij 1.194. However, given the different styles of play between the two sides the number of vertical passes they both attempt is pretty different. De Vrij attempts 2.516 p90, Koulibaly 9.786. This could be a factor when it comes to signing a new ball-playing centre-back, looking at how they distribute the ball rather than just can they.
So, two highlights.
Stuttgart’s Impressive Young Defenders
One of the first things to stand out was Stuttgart’s impressive two 21-year-old defenders – Timo Baumgartl and Benjamin Pavard. Baumgartl (pictured above) has represented Germany at under-21 level while Pavard recently earned his first 2 senior caps for France against Wales and Germany.
According to TransferMarkt Pavard has spent time at centre-back, right-back and midfield this season, though the majority of his appearances have come from centre-back it could mean his numbers are slightly skewed. He’s unlikely to make vertical passes that are counted from right-back, so if we only counted his time at centre-back he might fare slightly better.
So, what makes these two so impressive?
For players with more than 450 minutes, Baumgartl has attempted the 8th most vertical passes p90 and Pavard the 24th most. Filtering for players who’ll be 24 or younger at the end of the season (born after 1/6/1993) sees Baumgartl attempting the 3rd most vertical passes p90, behind only John Stones and Kevin Akpoguma (who has some great numbers too) and Pavard with the 7th most.
The graph below shows them in relation to other defenders:
Looking at vertical passes into the final third sees the two shine again. Baumgartl attempts the 5th most passes in the final third p90 and Pavard the 18th, while filtering for under 24’s Baumgartl is first and Pavard is 4th. The below graph shows how they relate to other defenders for vertical passes into the final third:
While Pavard attempts less vertical passes than Baumgartl, they completely practically the same amount. The Frenchman completes 4.143 p90 and Baumgartl 4.615 p90. This is reflected in the xP Rating of the two defenders, while Baumgartl’s 1.075 is still great, Pavard’s 1.130 is the 8th highest for defenders under 24 with more than 500 attempted passes.
So, now we’ve established they’re both great passers of the ball, can they defend?
I’ve mentioned before I’m not great at judging individual defenders based on stats, so I’m not going to dive into their tackle and interception numbers. However, while their current league position doesn’t flatter them, they have conceded the 6th least goals in the Bundesliga this season with 17 in 14 games which is encouraging.
Having an experienced defender such as Holger Badstuber alongside these two youngsters could be good for their development and it’ll be interesting to see how their development continues and if any top clubs start sniffing around. See some clips of the pair below:
Lewis Cook’s Start To The Season
After Bournemouth signed Lewis Cook last season and he didn’t feature much, I initially thought it was due to Eddie Howe’s strange record of signing players only to not give them sufficient minutes. However, it was a recurring ankle that troubled Cook in his first season with the Cherries. After a successful summer captaining England U-20’s to a World Cup win though, he’s started the 2017-18 season brightly.
Cook has still only played 465 minutes this season, so these values should be taken with a pinch of salt, but his numbers are impressive.
Messing around with filters show just how impressive Cook’s numbers are.
There are only 9 players (>=450 mins) in the top 10% for attempted vertical passes p90, attempted vertical passes to the final third p90 and completed take-ons p90, Cook is the youngest of the these 9. The other 8 players are Lionel Messi, Andres Iniesta, Philippe Coutinho, Paul Pogba, Naby Keita, Mousa Dembele, Luis Alberto and Gabriel Pires.
For players under 24, Cook is 2nd to only Naby Keita for attempted vertical passes p90 and comes in 8th for attempted vertical passes to the final third p90.
If you want a midfielder capable of resisting pressure and progressing the ball forward, through either passing or dribbling, Cook seems a good match. This graph looking at vertical passes and take-ons shows as such:
Then filtering for under 24’s he looks even more impressive, as does Naby Keita:
Interestingly, his xP Rating is 1.009 so he’s completing pretty much exactly what’s predicted. This is a bit behind players he’s in a similar area to – such as Naby Keita and Lucas Torreira – but is at least above 1.
Given his time out and age it may also be something that disappears with more minutes under his belt and it’s definitely not a concern. It’s only worth pointing out as it’s the only area he falls short of those players around him.
Bournemouth uploaded a video showing his every touch against Leicester which is interesting watch, showing just how capable he is under pressure:
While Bournemouth haven’t had a great start to the season, it could be a good place for Cook to continue to his development. Not under the same kind of scrutiny or pressure as if he would be at a top six club he can focus on improving and getting minutes. After a call-up for the recent friendly against Brazil, he may even have an outside chance of getting on the plane to Russia in the summer if he can maintain this level of performance.
Splitting passes up can be fun and lead to some interesting findings. While it may not be performance related and playing more vertical passes isn’t a quick hack to get more points, they can be useful tools and players who are capable of vertical passes may be desirable for clubs.
Manchester City, Napoli and Bayern Munich tend to have the individuals who excel at vertical passes, which shouldn’t come as too big of a surprise. The first method had Kalidou Koulibaly and Jorginho as the defender and midfielder with the most vertical passes and that holds true in this new method. Cesc Fabregas, as usual in passing numbers, gives a good account of himself. Whilst Kevin de Bruyne, David Silva and Fernandinho all impress in midfield for City.
In defence Jerome Boateng is in a league of his for getting the ball into the final third, while partner Mats Hummels plus Dortmund pair Marc Barta and Sokratis also impress.
Interestingly, Arsenal play the 2nd most vertical passes per game in the 4 leagues used, behind only Manchester City. Shkodran Mustafi, Granit Xhaka, Alexis Sanchez, Mesut Ozil, Aaron Ramsey, Nacho Monreal and Laurent Koscielny all attempt nearly double the average (~2.9) of players with over 450 minutes this season.
I’m sure I’ll use this again in some way when looking at players and teams in the future, while also looking to continually improve the method.
*Data up to date as of 7/12/2017