Defense is probably the most important element to focus on in football. It’s important for title winners. (*Glances toward Liverpool* Okay, most title winners.) And it’s really important when trying not to get relegated.

Yet our understanding of defense, or even the various ways we track it statistically, is awful. We have almost nothing to go by when evaluating team play except goals and shots conceded. Do tackles matter? Do interceptions? Clearances? Blocks? There’s research going on in these areas, but needless to say, this has been bugging me for a while and I’m not happy with the breadth of the data we currently evaluate.

Today I want to simply introduce a couple of additional defensive metrics we should be examining at the team level. This is not rocket science stuff – it’s basic info that up to this point most people have not had access to. Tomorrow I’ll start layering in some additional contextual info, plus adding some historic data, and maybe at some point we can draw conclusions. For now, just check it out and see what you think.

(Note: all of this assumes that I sliced and diced the data correctly. That may be a large assumption. I don’t have any real baselines to compare to, but these mostly pass the eyeball test.)

Opponent Passing Percentage
So we track how well teams pass the ball, but for whatever reason, the inverse information is not available anywhere. If passing and possession are part of attack and game control, then preventing your opponents from doing the same should also be important.

Here is the EPL table for opponent passing percentage through April 21st.


So this table looks at accurate opposition passes, total opposition passes, and the percentage, all on a per game average basis.

Seven of the bottom eight teams have been relegation candidates at one point this season. The other team in that bottom eight… are small favorites to win the league. Ooookay.

What about if we look at this same metric in Germany?


The top 3 teams here have either clinched a Champions League spot or are battling for the final one. But the bottom 4 are battling for European slots as well. Hrm…

How about Spain?


Rayo, Bilbao, and Barcelona all have pretty hefty deviations from the mean in the positive direction, but beyond that I don’t see much to draw on.

Right now we don’t know how much this correlates to other defensive metrics like goals and shots allowed. It may be that this tells you more about style of defense, and which teams employ a regular press, but it’s something for people to explore in the future.

Opponent Final Third Accuracy
Obviously what happens in the final third is the most important bit defensively, but different teams employ different tactical schemes to try and prevent their opponents from being effective. Does that show through in these stats?




Anyway, expect to see more about this type of stuff from us (and hopefully others) in the future.


  • Duncan

    I would never have expected Spurs to be there as of April 1st. How do the AVB and Sherwood eras differ?

    • Richard

      Although it won’t answer your question specifically in terms of the AVB/Sherwood difference on pass/cross prevention, for an excellent analysis on how the system changed, pop on over to Cartilage Free Captain and read Michael Caley’s piece.

  • Chris Gluck

    Ted, Great to see you looking into this in the European Leagues; alot more data to add to the effort for sure. I’ll will keep a close eye on how things play out compared to Major League Soccer.
    Best, Chris

  • Stephen

    Defence is a strange one to track but in my opinion what is missing is unit data. For example take your average defensive 4-4-2. Two blocks of four with the two front players either dropping deep to prevent midfield passes or slow down the oppositions attack to allow the blocks to position themselves. The front two players are important in the early phase of defence but when looking to stop and counter an attack, the first two blocks are important. I think there would be a strong correlation between a very good defence and the average distance between players in blocks and the distance between blocks themselves. Another cofactor would be the first blocks distance from their own goal.

  • ZDR

    I like the concept of opponent passing percentage, but conceptually (and as is clear in your tables), it’s too raw. Consider, Chelsea against Liverpool. Sitting very deep and allowing Liverpool to boss possession, you’d expect Liverpool to have a reasonably high passing percentage. Do we then say Chelsea are defending poorly because Liverpool are allowed to pass it around? I think there is merit here, but some nuance will be needed in future to account for such things as tactics and such.

  • Brandon

    Other aspects that may be of use when combined with the final table (the one that accounts for opponent passing % in the attacking third) would be goals against, location of goals scored, and chances allowed. That may give us a more complete picture of a team’s success at defending in the final third, as it is goals against that are the ultimate measure of a team’s defensive success. It is also beneficial to take into account the number of passes allowed. For example, Man City are in the bottom third for final third opponent passing but have the fourth fewest total passes attempted and second lowest completed against them, meaning they must still a good job defending as a whole. They may let you pass the ball, but not let you do much danger with it.

    Equally interesting is what this info reveals about how teams defend, particularly how they set themselves up. Liverpool, based on these stats, do a better job of disrupting play in the final third (lower opponent passing %), but allow more crosses than City, which would lead you to believe that they funnel attackers to the wings, whereas City seem intent on funnelling players inside where their big boys (Toure, Kompany, et al) can deal with those threats head on.

    What about considering opposition key passes? Could adding this metric be more revealing as well? A final note – though this system is new and admittedly not a complete measure of a team’s defensive success, it is interesting that the team with the best rate overall in the three big leagues is Bayern Munich, a team famous for its organization and possessional dominance.

    • Chris Gluck

      Brandon, I hope Ted doesn’t mind me offering up some stats here about MLS – I published an article specific to MLS a few weeks ago entitled Expected Wins – here is the link – and that speaks specifically to your questions. All told the R2 relationship is extremely strong throughout the course of primary data points in my Possession with Purpose. Overall I’ve been doing research on the other side of the pond for about two years on how well the attacking team attacks versus how well their opponents attack – that is the essence of my PWP Composite Index.

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