I should probably preface this with an apology.

Dear Fulham Fans,

I am so sorry.

–Knutson

Okay, now that that’s out of the way, we can get down to business. One of the things I have been doing lately is creating all sorts of radar charts to examine and compare player stats. Many people seem to like these. Some analysts have complained about certain elements (and there are notable weaknesses), but in general, I still think they are highly useful and add a level of approachability that you just don’t get from posting numbers.

Anyway, almost immediately after I released the first midfielder charts into the world, I wondered if we could use the overlays to compared coverage of entire midfields. I was not the only one, since a number of smart people also popped up that same day asking the same question. That’s what we’re going to test out today.

Because I don’t know if this idea will work, we’re going to start out by testing some extremes. Like say, one of the greatest midfields of all time…

versus this year’s Fulham.

Whee!

Barcelona 2010-11
I’m just posting the overlays here up top, but for those of you who want to see each player’s radar, I have added an appendix down at the bottom so that you can go nuts.

Barcelona_1011_Overlay

You can quibble a bit about the composition of any midfield, but like peanut butter and jelly, or Beyonce and Single Ladies, Busquets, Xavi, and Iniesta is one of the greatest combinations of all time. (Of all time.) Yes, there were plenty of times where Iniesta started as one of Barcelona’s wide forwards in a 4-3-3, but there are also plenty of times where this was the starting midfield combo that season.

As you can see from the chart, each player has a fairly significant, and distinctive shape.

Xavi is the masterful playmaker, with extreme scores in passing accuracy, key passes, long balls, and completed throughballs, who never commits fouls or gets dribbled past. His defensive contribution looks low, but it’s pretty solid for a guy whose job is not defense, and whose team controlled something like 70% of the possession that season.

Iniesta is intriguing, as I don’t think I have seen a shape quite like his. Scoring contribution (Non-penalty goals and assists per 90) is lower than I would have expected, but he has an excellent dribbling rate, and scores well in the creative passing stats (that throughball number is also great, and only looks weak next to the likes of a Xavi, Totti, or Cesc Fabregas). Additionally, his defensive stats are outstanding for that type of player, and he rarely fouls and never got dribbled past.

This is the first time I have encountered a shape like Iniesta’s.

Busquets looks an awful lot like the charts of Mikel Arteta or Michael Carrick, with like a 10% bonus added everywhere. He is the prototype midfield destroyer and recycler, but he’s also a capable offensive passer as well. He’s also only like 22 years old in this season, which is pretty amazing.

Fulham
Okay, now because I wanted to compare extremes to test if total midfield shape overlays would be useful at that level, I have chosen the team who has one of the worst midfields in the Premier League this season. Formation is a bit different in that Fulham have varied between a 4-2-3-1, 4-3-3, and 4-4-2/1-1 this year, but it basically always involves Sidwell and Parker and usually Kasami in some capacity (the central attacking mid in the 4-2-3-1). I could put the occasional 4-3-3 partner of Karagounis in here, but that just makes it worse.

Fulham_14_Overlay

You know all those big spikey swathes of coverage Barcelona’s midfield has? Fulham has none of that. In fact, Fulham gets basically nothing offensively out of their midfield. The lack of performance there is shocking. Sidwell and Parker do post solid tackling and interception rates, but that’s probably a result of the fact that opposing defences are always running at them.

It’s also interesting to note that all three of the Barcelona midfielders complete more long balls than any one of the Fulham players, something you don’t really expect if you buy in to the Barcelona equals short passing mythos, but then again, they complete so many passes a game, some of them are bound to be longer, right?

Anyway, initial test complete. Barcelona’s midfield overlay covers a ton of area on the radar chart, with multiple players looking fairly amazing across a number of metrics. Fulham’s overlay covers very little area, with almost no offensive output, and little good stuff in general beyond some ability to tackle.

Barcelona_Fulham_Overlay

With that out of the way, expect occasional follow-ups doing midfield output comparisons, and potentially looking at attacking overlays as well. Every article won’t contain these going forward, but expect to see them regularly for a while, as I keep testing and refining the concepts.

A Whole Heap of Bonus Radars

Xavi_1011_done

Iniesta_1011_done

Busquets_1011_done

Scott_Parker_14partial

Kasami_14_partSidwell_1314_part

 
Opta-Logo-Final-Cyan

 

Opta

  • toshack

    Ha, ha Ted, you’re a cruel man.
    My Fulham pal, who I usually forward all Fulham stuff on StatsBomb to will cry and feel sad or tragic or hysterical or heartbrokoken or dejected or apathetical or possibly all of the above…
    It’s a good thing you apologized!

  • marlon

    For a full midfield comparison, might it be better to combine the players stats into one radar, rather than overlay each individually?

    • toshack

      No point to combine. With the overlays you will see the whole picture and also se how they complement each other (or not).

  • Derek

    what is the point of these maps? to make charts look cooler? is 83% passing equivalent to 4.2 interceptions and 2.8 successful dribbles? not necessarily so I don’t exactly understand why they are put on the same axes.

    • miki

      Yeah, I’m at a loss too. I just don’t see anything on these charts that I wouldn’t see by looking at numbers, and even worse, some of them give the wrong impression.

      Take the Busquets’ and Parker’s as an example. The difference in area covered by their respective graphs are small compared to difference in player quality. For instance, Busquets makes double the number of interceptions of Parker (which is huge in terms of the player’s defensive performance) but it only makes a small difference on that graph.

      • Derek

        yea I agree they are actually worse than simply listing the stats. these maps imply more area=better player by the way they are set up despite little evidence of that being true and the scaling being seemingly arbitrary.

        • Hussain Dzan

          Ted,

          As far as scaling is concerned, if we were to normalize all active (maybe 5 full games per season) players in European top leagues, and get the values that separates the normal from the elite, I would think that the graph would make more sense.

          We can then plot the total number of each attribute instead of individual numbers, and thus have a fair reflection on how Man City would fair against La Liga for example (notwithstanding the fact that they play against different oppositions compared to Barca and Real).

          That being said, the logical next step is to have the charts done for UCL games..

          • Derek

            there’s no way to tell normal from elite when there is no indication to the importance of each category on the map. if “elite players” have 1.1 throughballs per game and “below average” players have 0.6 and “elite players” have 3.6 interceptions compared to “normal players” with 1.8 that still tells you very little about how that helps their team.

            I haven’t it seen proven anywhere what any of these stats are worth much less if 2.8 key passes are equal to 83% passing (which since key passes are contained within passing % is kind of a stretch to compare the two anyway). there needs to be years more research and team context applied to these maps before they tell you anything about quality of the player and are anything more than something cool to look at.

            I still think way more information gets across by listing the stats and not making the false equivalence between all these categories

  • Errorr

    Just further evidence for me that Busquets is perhaps underrated. His skill set is so much a step beyond most players at a position that is probably undervalued compared to players who score goals.

    I also have a feeling that Mascheranno might be almost as good except switch the interceptions for tackles if one is to believe the talk that he might be one of the best tacklers in the world.

  • Errorr

    Also, the argument for an unified radar for a set of players is that it helps alleviate the visual discontinuity of comparing a set of spikey players to ones with looser roles but lesser skills. The area is largely influenced by the arbitrary positioning of the stats. In the examples above the skikey dribbles to dispossessed part of inesta when combined gives a weird lack of overlapping when combined with busquets who doesn’t ever dribble past but also doesn’t get dispossessed. Of course Iniesta getting dispossessed in the opponent third is going to happen a lot while if Busquets loses the ball it is a disaster.

  • Ed

    Great charts apart from one thing: there is no reason to rotate the labels of the different quadrants. “Fouls” etc. This just makes it difficult to read.

Improve Performance and Productivity in Your Club:
State-of-the-art Football Analytics