Since I started writing about players last summer, I have wanted to get better at doing data visualization. I’ve loved working with data for two decades now, but while comfortable with a database or a spreadsheet, I am D-U-M-B when it comes to anything art-y.

Thankfully, there are these things called books, and people like Nathan Yau and Edward Tufte who write about the art and science of visualization. I’m still a complete and utter novice, but time spent learning has made me just good enough to steal some pretty things and start adapting them to football.

Radar Charts
A couple of months ago, Orbinho (Arsenal fan and Opta’s Content Director) linked a big list of data visualization award nominees. In that set of vis was this project, from last year’s NBA. I saw it and immediately thought “Sweet! That should work for football as well.”

Ah, but how? Most football stats are per game or per season, which makes it tougher to compare guys with different playing time…

*Does a couple of months work*

Aha! What if we normalize most of the stats by percent or per 90?

So that’s what I did.

Ballon D’or 2013
Regardless of my feelings about subjective awards (I generally loathe them, especially in football), the Ballon D’or made for an interesting set of players to talk about.  For once in my pieces though, instead of just talking about player production… we can look at them.

The charts below are based off stats for the completed 2012-13 seasons. Obviously that doesn’t line up perfectly with the silly award voting, but comparing full seasons makes waaaay more sense than comparing performance over a calendar year.

Ronaldo_1213_NPG

Ronaldo’s shot stats from this season are the boundary for the shots axis, meaning he has the most shots per 90 of any player in the database. In fact, of the 4.5 or so seasons of data I have access to, Ronaldo has all five of the top 5 sports in shots per 90. I’m surprised he hasn’t kicked his leg off. Luis Suarez is capable of shooting a lot, but no one even approaches Cristiano Ronaldo.

One of the things that surprised me is how low Ronaldo’s dribbling rate is, especially compared to his fellow nominees. My brain is burned with an image of Ronaldo that involves ridiculous stepovers and fast feet, but his dribbling at Madrid has never been nearly as high as what Messi and Ribery produced last season. Additionally, that 1.88 number from last year is the lowest in the data I have, and down almost 1.5 dribbles per 90 from his 2009-10 campaign.

Messi2013_NPG

As you can see in this one, Messi’s season last year is the Non-Penalty Goals per 90 boundary. It’s the best in recent times, and possibly the highest in a European big league in the modern era. Messi creates a ton of shots, is good at putting them on target, his goal conversion is great, and he’s an outstanding dribbler.

The area covered in the scoring areas on his radar chart is massive. From a statistical standpoint, this season was Messi’s finest ever. Add the goal and assist rates together, and you see his scoring production was 1.85 per 90 minutes across an entire season, a number we may never see again.

Not good enough to win the Ballon D’or though…

Subjective awards are still stupid.

Franck_Ribery201213

Ribery presents a fascinating contrast to the top two players.

Messi and Ronaldo profile quite a bit alike, in that they are high-scoring shot monsters.

Ribery is clearly something else.

While his goal scoring rate is quite good for normal wide players, and his accuracy and conversion rate are also good, it’s in the passing and dribbling stats where Ribery shines. That assist rate is among the best in Europe, and both Successful Dribbles and Key Passes are as well.

Ribery’s different role causes there to be completely different areas covered in the chart versus the other two.

Meaningful? Maybe.

Cool? Certainly.

To compare and contrast, this is what the stats look like in basic table format.

ballon_dor_table

Next Steps
Now that I have the template done (and huge thanks to Nat James for helping me create that), we can churn these out fairly regularly with updated data. Expect to see a snapshot or two of interesting players from each week on the basic template, either on the site or on our Twitter feed.

Another thing that I will start to explore soon are position-specific templates that help compare key stats for spots on the pitch. This basic one works reasonably well as is, especially for attackers, but there are plenty of other stats you will want to look at for defensive midfielders, fullbacks, defenders, etc, and I think we can find new templates to emphasize those.

Feel free to leave feedback on them in the comments or on Twitter.

It’s taken a plenty of studying and hard work to get here, but I feel like maybe we’ve created (or more appropriately, “adapted”) a new way of looking at football stats that will be both approachable and useful for casual fans and analysts alike. If nothing else, you’ll see more pretty pictures on the site.

–TK

  • woocash

    Great work and interesting findings.

    Is there any way you could put these three charts onto one?

  • John

    Quite a few of the variables in the plots are not yet in the stat definitions section, would be good to add those for new readers.

  • david

    Are those stat systems for rating players something similar to the way they are rated in pro evolution soccer? could be wrong, haven’t played that game in years!!!

    • tknutso

      Apparently they are, though I only played FIFA myself. As noted in the piece, the first place I saw this used was that NBA vis.

  • Joshua Jones

    I’ve just written an article similar to this for WhatCulture (pending review) and it takes a very similar stance to this. Its about what the team of the year would be if it were based purely on key stats in various positions.

    Good work pal. The visualisation here is great.

  • Jalal

    Wow, That was truly amazing. Especially the overlay gives a great perspective. The difference in style between Ribery and Ronaldo occupying similar areas of the pitch is well represented.

  • Usman Abubakar

    I was highly interested I your findings. You did a great job.

  • Thomas

    Hi Ted,

    Love the effort and love the site. I check it all the time. I’ve been thinking about your graphs and, as feedback, here is the way I would go about:

    1. I’d lose the circles. There’s something that I think is not quite right in the sense that the performance shapes are octogonal (as they should be) while the reference shapes are circular.

    2. I’d also let go of the alternative shading in the back. It’s information that I find too little to be that important visually and, when you overlay performance on a single graph, it makes it all messy in my opinion.

    3. I’d keep only light octogonal reference lines on a white/transparent back and I’d add a reference shape of the position-specific means, or maybe medians, outlined in a darker/thicker line.

    Cheers

  • Hussain Dzan

    Been wanting to say how I love your site. Really great for Fantasy League players (not that I’m one of them), and makes such an insightful read in general.

    What would be helpful for these templates are ‘ideal’ stats gray-overs that gives a rough idea on whether the players reviewed are ‘World Class’, average or just duds.

    Keep up the good work!

    HD, Malaysia

  • kroki22

    I Don’t believe it, that Ronaldo was better in keypasses, I don’t beleive that Ribery had more successful dribbling.. Was ribery a better passer than Messi? Don’t tell me that.

  • kroki22

    don’t tell me, that Ribery gave more assists than Messi!

  • kroki22

    Not good enough to win the Ballon D’or though…I think it should have been enough for Messi to win the Golden Ball. Ribery wasn’t better than Messi! it’s nonsence!

  • Pinkybum

    Love the site and all the stats.
    However, radar plots are pointless unless you are really conveying related information and the shape and area really mean something.
    Somethings to think about:
    Does the more complete player have greater area?
    If the shape is bigger in one direction does that indicate the type of player they are?
    Are those characteristics closer together to reveal those characteristics?
    How can you more easily compare the players? Scrolling down it is hard to compare these charts.

    I am going to come back to this site regularly to watch the developments

    Many Thanks

  • William

    If I wanted to understand this in NBA terms, is Ronaldo like Kobe in high scoring, high usage rate or like Durant in high usage rate, transcendent scoring that makes you not care about usage rate?

    • tknutso

      Closer to the Kobe theory. At some point it harms the efficiency of your effort. Where that is, we don’t know yet.

  • Pingback: Making Data Look Good, but Still Conveying Pertinent Information | Stat Hunting()

  • Pingback: Do statistics have a role in football? | woodsport()

  • Pingback: Radar Love – Capturing Players in a Single Picture | 11tegen11()

  • http://www.ramimo.com/2013-NBA-All-Stars Rami Moghadam

    It’s a shame you couldn’t make these infographics your own and had to copy my design exactly. But whatever works…

    • Ted Knutson

      Rami,
      I’m sorry you feel that way, though I have given you credit for the inspiration numerous times throughout the process. The current versions of the football radars have had a ton of development done from your initial NBA versions in statistical presentation, but the basic design of the circle and alternating team color representation is certainly your work, and it’s excellent design.