The Mid-term report series turns its beady attention to the centre back position. This part of the team has been a problem for Celtic for many years in terms of injury and lack of continuity. This season saw Ajer and Simunovic prepped to start the European qualifiers by being installed as first choice at the end of last season. This when the superior Benkovic was fit.
Despite best intentions, and characteristic of Celtic’s profile in this area, Ajer and Simunovic only appeared together for 90 minutes 5 times. £7m man Jullien came in permanently after the 3-4 Cluj disaster and following yet another Simunovic break down. He and Ajer have been partnered in every game since except for Partick Thistle and Hamilton Academical at home.
Indeed, the Ajer-Jullien axis is shaping up to be the first centre back pairing to both reach over 40 matches worth of minutes since Denayer and Van Dijk in 2014/15. Since then Boyata appeared in 39.25 worth of minutes in 2015/16 and Ajer 38.59 worth of minutes in 2018/19. Given Celtic are playing over 60 games a season and this is an area that works best with partnerships, that’s been a problem.
Defensive Performance
Focussing just on defensive actions, the following plots Defensive Action Success Rate % DASR%) and Possession Won from Defensive Actions %.
I have included all the centre backs from the Treble Years who have played more than 20 matches worth of minutes. I have also added Van Dijk and Denayer from 2014/15 as this was the last sustaining partnership over a season, and also, in my opinion, the most recent gold standard for centre back pairings.
Whilst it is still the case no one can get to Van Dijk’s DASR% of 84%, Ajer and Jullien are clearly operating at a higher level than any recent pairing. As can be seen above, 16/17 was a bit of a low point for centre back performances hence why Toure was brought in and Celtic conceded 50 goals the 54 under Rodgers.
17/18 saw the emergence of Ayer and the improvement of Boyata. Last season saw Boyata and Benkovic be the best performers despite Ajer and Simunovic ending the season as the selected pairing for expedient reasons.
But 19/20 Ajer, alongside new Bhoy Jullien are threatening to eclipse the performance of Denayer and Van Dijk as a pair, if not individually overtake the imperious Dutchman.
Why is this? Well, the continuity factor cannot be discounted. Celtic have had notorious bad luck with centre back injury but those two have now played nearly 30 matches together. There is a case the SPFL in particular is less competitive this season in terms of depth. Most clubs live within their means and therefore are at the mercy of player contracts. Clubs like Hearts and Hibernian have made bad managerial decisions and are therefore weaker. Aberdeen are declining after a successful cycle. Clarke left Kilmarnock. Also, there are no Champion’s League Group matches to potentially deflate performance. Neither was there is 2014/15 for Denayer and Van Dijk, interestingly enough. It may also be the case they are a lot more solid than vociferous critics believe.
Of the current centre backs Jullien is by far the more dominant defender with 2 more challenges/intercepts than Ajer per 90m and nearly 4 more than Bitton and Simunovic.
Defensive Error and Saves
I’ve documented many times that Bitton is error prone as a defender – he loses concentration and wanders out of position. Playing at the back can be a bonus against weaker opposition as his distribution is so good. It is astonishing to me however that he has deputised for Brown on so few occasions. Surely this is simply because of Simunovic’s injury and the need to have him as backup. This season, Simunovic also goes over my 0.5 threshold for acceptable numbers of defensive errors.
Nothing has divided my timeline on Twitter more than the decision to award Ajer Player of the Month for December on A Celtic State of Mind (ACSOM), a wonderful site I contribute modestly to. I can almost hear the gnashing of teeth from certain quarters as they see Ajer has the lowest error rate. Why so? Of the 11 errors I have tagged Ajer with, 5 have come in the last two meetings with The Rangers. Those higher profile performances were punctuated with uncertainty as well as being fresh in people’s minds. Some see errors against The Rangers as being more important than against anyone else – “Old Firm-itis” I call it. This in turn leads to selective memory where all the other performances are forgotten.
Bitton also has the most missed tackles (0.2 per 90m) whilst Simunovic has none. Simunovic and Jullien lead the way on possession lost in their own defensive 3rd. I find this is a result of those who effect more Clearances.
Defensive 3rd Wins and Recoveries
These two metrics speak to recovering the ball and especially within Celtic’s defensive 3rd.
Jullien leads the way by these metrics which speak to anticipation and positioning. Also, he is often one who engages first when the ball is launched forward by the opposition.
Ball Progression
Plotting total number of Open Play Passes and indicating how many of those are Pack Passes – forward passes that take out opponents.
Bitton is clearly the most prolific passer and over 9 of his passes per 90m take out opponents. As mentioned, it is a useful tactic when there is little defending to be done.
Jullien’s passing profile is otherwise the more interesting. He completes the fewest Open Play Passes (44 per 90m) yet the second highest Pack Passes with over 8. Whereas Simunovic complete 94% of his passes and the others are all over 90%, Jullien completes 89%. He puts more risk into his passing and that will be one of the features Lennon likes – he gets it forward that bit quicker and more often.
Ajer, of course, progresses the ball on the ground by dribbling. He completes an average of 6.06 ball carries per 90m (Pack Dribbles + Progressive Runs). Only Frimpong, Hayes, Morgan and Johnston complete more in the squad. Bitton is next on 3.91.
Conclusion
Ajer is the only remaining member of the 18/19 treble defence. As I have shown in the midterm reports for right and left back, the current incumbents are not yet an upgrade on Tierney and Lustig. Within this context, Ajer and Jullien have forged a partnership that has seen rare continuity, but more importantly, defensive performance fundamentals not seen since Denayer and Van Dijk. Some will scoff at that comparison but the numbers are clear and from over 30 appearances each so a good-sized sample.
Neither is perfect. Ajer has improved his manhandling of opponents but his left foot is as weak as ever, he goes to ground too often and jump timing can improve. Jullien tends to turn up for the big games a coast through the smaller ones – 10 of his 13 defensive errors have come against smaller SPFL sides. Jullien also tends to allow himself to get riled by unlikely opponents far too easily.
If they were better, they’d be at Barcelona. As Celtic centre backs, they are on course for a golden season. And no, the article does not claim either are individually on the same level as Van Dijk!
Let the early ball do the work instead allowing the opposition time to set up their defense, along with when this centreback looses the ball is caught way out of position to carry out his duty, defend. Maybe have a look at defensively failing to mark and positioning hence allowing opposition chances. At 6′ 5” you would think this player would be successful defending in the air, unfortunately not. Good defending when the opposition is attacking you suggests staying on your feet, not on your rear, which will give you more chance to block and be physical. Of course he is young, unfortunately we are being heavily pressured by our competitors and at this time we require a solid player to with with Julienne. I hope he can focus on doing his job. Do we have time.
With Alan’s permission, I am again adding some additional data. Assume everything in the first few sections is from SPFL-only except when I designate otherwise. I found a new “toy” this past week which focuses on Champions and Europa League data, and will include a separate section on that data.
Our centre halfs are truly unique in the SPFL offensively, as they produce xA and xG from Open Play p90 at levels which are much higher than league average- particularly Ajer. The combined xA+xG from Open Play p90 for each is Ajer 0.09 and Jullien 0.07. To break that down a bit, they have comparable xA but Ajer has double the xG vs Jullien, which is likely from his runs into the box. Ajer’s xA+xG is 200% higher than the average of SPFL centre halfs, while Jullien’s is 135%. By comparison, Rangers’ centre halfs offer very little in open play, with none of their three (Goldson/Helander/Katic) offering any xA – literally zero. Helander is also zero on xG from Open Play, while Goldson is 41% above league average and Katic is 29% below league average. From an open play perspective, our centre halfs produce like the average SPFL defensive midfielder, and in some ways, superior to Brown- which is not flattering to Brown.
I believe xG from Headers p90 is an important metric for centre halfs, and unfortunately for Ajer and Celtic generally, the data confirms what most intuit from anecdotal experience- we aren’t great at set pieces in the air and Rangers are far superior. I’ll have some “aerial duel” data in 2nd block of info, which supports that Jullien is pretty good defensively in the air. but his xG from Headers p90 is only about 46% better than the league average SPFL centre half. To put this into perspective, he is effectively tied in this metric with grandpa Berra, who was just disposed of by Hearts. Ajer is objectively terrible – 59% BELOW league average for centre halfs. Rangers’ centre halfs dominate the league in this metric – with Goldson 137%, Helander 223%, and Katic 154% above league average. It is impossible to decompose how much of this is due to the quality of delivery, and surely some of it is due to how good Davis/Tavernier/Borisic are with deliveries, and that is reflected in their xA from Set Piece data. Regardless, this is a serious competitive disadvantage relative to Rangers.
xPassing data (basically looking at their pass completion % vs what the average of similar passes is) for each is comparable across the centre halfs from both teams- all about between 6.5%-7.5%. Jullien and Goldson stand out playing longer balls, with average pass length around 23 feet, vs the rest of the centre halfs closer to 20 feet averages.
Key Passes and Secondary Key Passes from Open Play p90 is another area where Celtic’s centre halfs dominate the league and their Rangers counterparts. Jullien average about 0.46 Key Passes from Open Play p90 and Ajer 0.32 – those are 215% and 117% of SPFL league average for centre halfs. Secondary Key Passes are the pass to the passer that sets up a shot on goal, and Jullien averages about 104% of league average, while Ajer is just 4% above league average. By comparison, Rangers are pretty terrible in these metrics. Helander hasn’t registered a single Key Pass from Open Play, while Goldson and Katic are 63% and 31% below SPFL league average. Goldson does lead the group with 129% of league average for Secondary Key Passes from Open Play p90. When I compare the creative passing data from Brown vs Davis/Jack, and then the centre halfs from both teams, it is clear that they use their central holding midfielders as creative playmakers, while we outsource a little of that to our centre halfs, but overall we are at a terrible deficit in this regard when we net it all out.
Given the length of this section and the nature of my new “toy”, I am going to provide the 2nd section in another reply post.
This past week I came across a company called smarterscount (www. smarterscout.com), which is a company that shares the data analytical models of North Yard Analytics. They offer a free (!) basic subscription where users can run 10 reports per day. Unfortunately, they don’t currently offer the SPFL except to Private Clients, but they do cover 45 other leagues, including Champions and Europa. After a few days of the 10 “credit per day” allowance, I’ve run all of ours and Rangers’ players using EL as a benchmark.
They run data algorithms for various models- they have an FAQ section here (https://smarterscout.com/faq) to explain what they are doing. To put it simply, they run their algorithms for each player then compare each player with the average and distribution of the league you set as a benchmark. I selected Europa League, so our players and Rangers’ are then compared against the average Europa League players by season and then assigned a score from 0(bad) to 99(good).
Alan’s work with defensive stats is FAR superior to this platform, IMO, as the data sample sizes are too small for Celtic/Rangers players. Also, their defensive models make assumptions/guesses which have a pretty good R square, but I trust Alan’s diligently sourced data more. With that being said, I think it is an interesting tool to compare on a RELATIVE basis – basically where do our players fit compared to other Europa League level players at their positions.
Their data models largely confirm Alan’s within the context of competitive positioning in EL. For example, Ajer is an elite dribbler for left centre halfs, and ranks 82 out of 99 in EL this season, while Jullien is only 13. In what they call “passing towards goal”, which is a pass of 10 meters or more towards the centre of the opponents goal, Ajer is just 24 and Jullien is 67. For ground duels when out of possession (basically defending against an opposing player who is in possession of the ball), Jullien rates at 73 and Ajer is 16.
I won’t recount everything available other than to summarize that Ajer’s defensive aerial data is decent in their models and Jullien’s is excellent. Also, Rangers’ Goldson (22) and Katic (31) are poor defending against those with possession on the ground and Helander (98) is elite. Their shape and tactics are tailor made to manage this weakness with tight compactness.
Most passes in the EPL… Van Dijk.
Second most touches in the EPL… Van Dijk.
Astonishing numbers.