PSV 2 – 2 Ajax: Does match data confirm the naked eye observations?

The thrilling encounter between PSV and Ajax, which ended in a 2-2 final score after Ajax came back twice, has been reviewed on 11tegen11 immediately after the match had ended. Now that the match data, provided by InStat Football, has come in, it might be interesting to revisit some of the statements made on the basis of naked eye observation and see if the stats point in the same direction.


A few points from the match review can easily be tracked.

  1. The draw was deemed a fair result in a match that saw both teams dominate different phases of the game.
  2. Preventing PSV to circulate the ball to Kevin Strootman had been highlighted beforehand as an important part of Ajax’ defensive strategy, which seemed well carried out.
  3. Ajax’ inside right wing role by Eriksen allowed Pieters a relatively easy build-up, leading to a left-sided dominance in PSV’s pattern of play.
  4. There was a distinct lack of defensive quality, with two particularly offensive minded midfield line-ups.


Let go over those points one by one and see what the data tell us…

1.       The draw was deemed a fair result in a match that saw both teams dominate different phases of the game.


Note the lines for shots, possession and attacks circling around each other during different phases of the match. Red: PSV, blue: Ajax.

Just like we did in the previous match data report, we’ll break down each team’s number of possessions into the number of attacks created with it, and the number of shots created with those attacks. Both teams will be shown to be remarkable similar in this regard, although, as is demonstrated in the above graphs, they dominated during different phases of the game.

Both teams created 75 attacks. PSV did so from 103 possessions, for an attack ratio of 73%, while Ajax did so from 111 possessions, for an attack ratio of 68%. And both teams created 15 shots for a shot ratio of 20%, with 9 attempts coming from inside the box for both teams. PSV managed to find the target with 8 of their shots, while Ajax were slightly less accurate with 6.


2.       Preventing PSV to circulate the ball to Kevin Strootman had been highlighted beforehand as an important part of Ajax’ defensive strategy, which seemed well carried out.

In the match against Legia, used to demonstrate that new summer acquisition Kevin Strootman is the most important element in PSV’s possession game, the defensive midfielder was shown to receive far more passes than his team mates: 15% of PSV’s completed passes was directed at Strootman in that game.

In the Ajax match, Strootman received only 21 passes, the third lowest number of PSV’s outfield players, behind Matavz (19) and Manolev (20). The total number of PSV passes was indeed a lot lower (303 vs 543), but the share of PSV passes finding Strootman dropped from 15% in the Legia game to 7% in the Ajax game.


3.       Ajax’ inside right wing role by Eriksen allowed Pieters a relatively easy build-up, leading to a left-sided dominance in PSV’s pattern of play.

Initial observation of the match suggested that PSV built the majority of attacks through Pieters, who found himself in a lot of space as Eriksen played a very narrow inside right wing role and often only checked his man when Pieters did advance with the ball at feet. Although Pieters did made more than twice as many passes as his counterpart Manolev (59 vs 29), PSV finished the match with 25 right sided attacks compared to only 19 left sided attacks.

So, somewhere in transition between Pieters’ possession and PSV developing an attack, a preference for right sided passes occurred. An eye-catching difference in this area is the passing game of Toivonen and Wijnaldum. Eleven of Wijnaldum’s 19 passes were directed at the three offensive players (Lens 8 ; Matavz 2 ; Mertens 1), while only three of Toivonen’s 18 passes reached a forward player. So, PSV’s right sided offensive midfielder Wijnaldum, played a significantly more offensive passing game than their left sided midfielder Toivonen did. This may help explain the data described in the above paragraph.


4.       There was a distinct lack of defensive quality, with two particularly offensive minded midfield line-ups.

Ajax' first half dribbles, showing a 100% success rate.

Measuring defensive performances has always been the Achilles heel of match data analysis. The most compelling story in this regard is perhaps Manchester United selling Jaap Stam, back in the nineties, when his number of tackles per game started dropping. Assuming that the defender was past his peak, Ferguson sold Stam to Lazio, with some of his best footballing years yet to come. Only later insight revealed that his improved positioning skill allowed him superior defending, without the need for risky tackling. With that in mind, it’s always important to take context into account when assessing the value of raw match data.

The PSV – Ajax game contains very interesting differences regarding both teams and particularly regarding both halves of the match. Ajax succeeded in completing an amazing 16 of their 17 dribble attempts, but all 10 of their offensive half dribbles were made in the first half. PSV, on the other hand, had only two completed dribbles out of 9 attempts in the first half, compared to 8 out of 12 attempts in the second half.

In both teams, several players stood out with a remarkably low rate of ground tackles won, confirming the initial observation of offenses dominating defenses in this match. Some ground tackle success rates for Ajax: Blind (0/6), Van der Wiel (1/6), Janssen (3/10), De Jong (1/4). And for PSV: Pieters (2/8), Marcelo (1/4), Manolev (3/8), Toivonen (1/6).

The introduction of Enoh, who won all of his 7 ground tackles, shifting Anita, who won all six of his ground tackles, to left-back, did increase the amount of tackles won by Ajax, but still didn’t stop PSV from completing a much higher rate of their dribbles in the second half. PSV also created 12 of their 15 attempts in the second half, while Ajax’ attempt were quite evenly spread.

PSV's tackles. Note the big difference between the first half (left) and the second half.

In the end

In short, yes, regarding the four points mentioned above, the data do confirm the observations made during the match. But there still remain so many interesting observations that only come forward when looking at these data. Take for example the difference between Toivonen and Wijnaldum with regard to their passing preference.

Our aim for the near future is to regularly implement the use of match data like these, courteously provided by InStat Football, to structure the observations made regarding tactics and player performances. Should more coverage be available and hopefully in the near future, all Eredivisie match be covered, the necessary context will become available to interpret different data as it should be. At present, though, the intermittent use of match data seems a very helpful tool to structure naked eye observations. As always, please feel free to share your thoughts on this type of analysis, which is only just emerging in the Eredivisie.


This post could not have been created without the support of InStat Football.

8 thoughts on “PSV 2 – 2 Ajax: Does match data confirm the naked eye observations?

  1. Orange14

    I think if Sulejmani were healthy, Pieters would not have had as easy a time playing up the field as Ajax were very silent going down his side of the defense. Not starting a true winger and having Eriksen drift inside was a real problem.

  2. bart

    I was interested in your PSV tackle stats so I looked at the .png and came with the following …

    Number Player name – total : 1st half : 2nd half : own half : opponent’s half.

    2 Manolev – 3/6 : 2/4 : 1/2 : 3/4 : 0/2.
    5 Pieters – 1/3 : 0/1 : 1/2 : 1/3 : 0/0.
    6 Strootman – 6/10 : 3/7 : 3/3 : 5/9 : 1/1.
    7 Toivonen – 1/2 : 0/0 : 1/2 : 0/0 : 1/2.
    9 Matavz – 1/4 : 0/3 : 1/1 : 0/2 : 1/2.
    11 Lens – 2/4 : 0/2 : 2/2 : 2/3 : 0/1.
    14 Mertens – 3/3 : 3/3 : 0/0 : 1/1 : 2/2.
    18 Derijck – 1/1 : 0/0 : 1/1 : 1/1 : 0/0.
    ? Unknown – 0/1 : 0/1 : 0/0 : 0/1 : 0/0.

    Total – 18/34 : 8/21 : 10/13 : 13/23 : 5/11.

    Now, I’m not saying much about your conclusions but somehow your posted figures before the graphic don’t add up what is displayed in the graphic … which is correct?
    For example, there is not one Marcelo (#4 for PSV) tackle shown in the graphic. I can’t tell who’s that one tackle is which is shown in the first half as a Manolev tackle (missed) is over it.
    What would make it even more interesting is to show which Ajax players successfully avoided the tackles (or not.) This could show a correlation between the defensive/offensive strengths of the players; who gave who the most trouble or not at all etc. when taking on an opponent or in intercepting a pass etc. You are of course right about the Stam thing so we can’t get down to the nitty gritty but still …

    If you ever need any help computing the data or discussing about what should be shown/analyzed etc feel free to mail me, my English is excellent and so is my Dutch. Live in Spain so never really see a Dutch game live on TV though …

    1. 11tegen11 Post author

      Thanks for your comment, and also for the offer to assist in data computing… Should more data like these be available in the near future, a more sophisticated approach may be possible. In that case, I know where to find you 😉

      To answer your question. The confusion arises from the fact that the text refers to ground tackles, where it should actually read challenges in order to avoid confusion with the tackles shown in the .png image.
      InStat Football defines ground challenges as ‘all types of duels on the field, which is a summary of neutral balls claimed, air challenges, dribbles, tackles and losses of a ball when the
      opponent takes possession of it’. Tackles are defined as an ‘active action of a player who tries to tackle a ball from the player who possesses it’.

      The best data on avoiding tackles would be successful dribble made. Scored of Ajax players regarding dribbles made are Sigthorsson (6/6), Boerrigter (3/3), Enoh (2/2), Vertonghen (2/2), Anita (1/1), Janssen (1/1), Eriksen (1/1) and Van der Wiel (0/1). In total 16 successful dribbles of 17 attempts, 13 of which in the first half, as shown in the .png image.

      Thanks again for your in-depth comment. More than welcome!

      1. bart

        ah ha … well maybe InStat Football should change their definition a touch since their “ground challenges” seem to include air challenges as well 🙂 maybe a word like “recuperations” would cover it all a bit better … anyway … on to the interesting stuff …

        now, if we have the successful dribbles then we could assume that it has been met with some sort of challenge, maybe even a tackle (failed) how ever feeble it may be … so if we flip the Ajax dribble .png it fits nicely over the first half PSV tackle .png so we can see for example that #9 of Ajax (Sigthorsson?) dribbled by #2 Manolev at the center top of the penalty area and also by #6 Strootman just to the right (on the PSV .png) … interesting, now if only he could start doing it against Real Madrid we’d have a game (2-0 currently) …

        anyway, watching the game so don’t have time to look a bit further at the graphics but cheers for the clearing up of terms,

  3. MvW

    PSV plays 4-2-3-1. You could also call it 4-3-3 (Dutch coaches and media have a habit of doing so) , but is definitely not the 4-3-3 formation (with Toivonen en Wijdnalum als ‘left and right midfielder’ that has been suggested.

    Toivonen ‘normally’ takes up position centrally in front of two controlling midfielders; when PSV win the ball, Strootman tends to be the playmaker who drops deeper to collect the ball, while Wijnaldum is the player who makes more forward runs (Wijnaludm’s role has been termed the ‘pendelfunctie’ by Twente coach Co Adriaanse last Sunday: Fer or Janssen play that role at Twente). Toivonen will move more to the left than the right, so that he makes space for Wijnaldum’s forward runs.

    Therefore there average positions show up as if PSV play 4-3-3 with Strootman as th anchorman. But that is not the defensive organisation, which define stheir formation: to see that, you should look at snapshots at which the team ‘re-organises’ itself, for example when one of the keepers has the ball. See for example, this snapshot against ADO

    Toivonen moves very much along the axis, and has to go with ‘his man’ ; he also has to help out ‘in the air’ where needed; this means his average positions is around the centre circle, but don’t let that fool you in believing that he does not play ‘in front’ of Strootman and Wijnaldum.

    The reason that Toivonen finds relative few forward passes is because he tends to play just in front of Strootman and Wijnaldum, often with his back to the goal; he is the central point to combine with, while Wijndum and Strootman can much more play with their face to the goal; it is then only naturakl that the latter two play more direct passes to the forwards.

    This is all very plain to see in the live games if you ask me!

    I very much appreciate someone bringing in stats like this, but I think they should be used ‘with care’.

    1. 11tegen11 Post author

      Match actions of PSV's midfield, taken from the PSV - Legia match

      Please look at the above image, or click here for an enlargement. All dots represent match actions, things like passes, challenges, fouls, etc…
      It is clear to see that Strootman operates centrally, slightly behind Toivonen and Wijnaldum, while covering a wide range of the pitch.
      Toivonen and Wijnaldum show a comparable pattern of match actions, at least in terms of their spread over the pitch. Look at their remarkable mirror image in terms of number of actions on the left, central and right side of the pitch. These numbers are presented below the pitch diagrams.

      I share your caution on average position diagrams, as player movement across the pitch is not represented, and positional switches ruin the validity of the average position image. Images like these, representing the ‘area of activity’ of certain players do present a much more accurate image. Look at the above ones and draw your conclusions on PSV’s midfield.

  4. MvW

    Those are great stats.Thanks for plotting them.

    I do find them in line with the 4-2-3-1 I see on the field, on how I described it above. In particular Wijnaldum is a ‘controlling midfielder’ with great freedom to go forward; this is also how he has described the role himself early on in the season, and similar words have come from Rutten. As I said before, Toivonen does go more left than right ta make space for Wijdaldum’s runs. Strootman does goes everywhere when PSV have the ball. So when we have the ball you are likely to see Strootman as the midfielders furthest back, and Wijnaldum and Toivonen almost on the same level.

    Would you perhaps have the same stats for ADO – PSV? This was the game in which Rutten ‘switched’ but in my opinion it was not the formation that changed, but the execution. In particular Wijnaldum dropped to midfiled and Lens went to play striker, while Labyad played on the right side.

    This game was, for the first hour or so, almost perfectly executed by PSV. It would be very interesting to see where the midfielders were taking up positions in that game; in my memory they were extremely disciplined in that sense in that particular game.


Leave a Reply