An Unsettled Outlook

The fourth season review considers the centre backs. I have discovered that assessing centre backs is harder than any other position. I think this is harder because:

  • Even more so than with full backs, central defending depends on “actions” not easy to capture – i.e. where to stand. Most danger is avoided due to being in the right place. Positioning is paramount. There isn’t really a metric for being in the right place. It can perhaps be inferred through other data points (blocks and intercepts).
  • More than any other part of the field, this position requires teamwork with your partner when lining up in the normal back four. I will attempt to look at the partnerships to mitigate this. So, positioning allied to communication, and collective “organisation”.
  • Unlike full backs where there are major differentiators in creativity and attacking actions, centre backs tend to be more homogenous in their data results. Boyata is an exception with his goal scoring. But generally, you don’t get many assists not chances created from centre back, nor many goals. They compete for aerial and ground challenges, clear the ball, and conduct a lot of low risk passes. Unless we are analysing David Luiz or Fernando Hierro!

But on we go!

Appearances

Lustig’s data is included for the matches he STARTED as centre back. There were two matches he started at centre back and ended up at right back. And there were two matches he started at right back and ended up at centre back. I have excluded the latter. So, his appearance data won’t include any bench time or not in squad time – see the full back review for Lustig’s complete data “Right Back Where We Left Off”. But it is fascinating he played at centre back more than Toure.

Ambrose, Ajer and McCart are discounted from further analysis due to insufficient data. O’Connell is included – I wouldn’t normally consider anyone with less than 900 mins (10 games) but it will become apparent why it is interesting to include him.

Note that the following loan deals were agreed (all between January and May 2017):

Ajer Kilmarnock
Ambrose Hibernian
McCart Inverness Caledonian Thistle
O’Connell Walsall

Despite not finishing the season as a starter, Sviatchenko was the most used centre back. Out of 59 matches, he is the only one to complete 90 mins in more than 50% of them. Out of 5310 possible minutes, only Sviatchenko and Simunovic completed more than 50% of the available game time.

C:\Users\Alan\AppData\Local\Microsoft\Windows\INetCache\Content.Word\svia4.jpg

Simunovic started the season injured for the first eleven matches of the season. Sviatchenko was injured after two matches and missed the next four. Toure did not join Celtic until the third match of the season. He was semi regular until the home game against Borussia Moenchengladbach, when errors leading to both goals resulted in only one more start all season. Ambrose was given two starts in the opening three games then did not feature in the match day squad again.

Boyata only appeared in one game before January but was then a regular. He therefore missed all the Champions League matches. He was injured until the end of October, but only selected once until after the Winter break.

C:\Users\Alan\AppData\Local\Microsoft\Windows\INetCache\Content.Word\Boyata2.jpeg

Simunovic and Sviatchenko established themselves as the most regular picks from September through to the end of December. Since the return of Boyata in January, it has been a case of him plus one other, most regularly Simunovic. Injury has had an impact on selection in this position, but it has taken Rodgers all season to work out his preferred options and even now I am not sure he has set first picks.

Defensive Action Success

Regarding O’Connell. Firstly – *SMALL SAMPLE ALERT*. He only appeared in seven matches, and 545 minutes, the equivalent to 6.05 full games. I wouldn’t normally include a sample less than ten full games, but what fun are his numbers?

16.18 successful challenges and intercepts versus 1.65 per 90m against is insane. It is almost double the defensive work of any other centre back apart from Sviatchenko. In addition, his clearance rate of 8.26 per 90m is far ahead of any other player. As you can see below, he has the highest intercept rate and second highest number of challenges won per 90m. He executed more defensive saves than anyone apart from Toure, but, oh dear, that error rate. Defensive errors per 90m of 1.49 is off the chart, and, I know what you are thinking. Efe Ambrose across his Celtic career had an error rate 1.26 per 90m. With O’Connell, the other glaring issue was that he simply did not move his feet quick enough to react. His core strength and agility can be improved. At 21, if he can develop and improve his decision making, Bury may have acquired a star defender.

I have previously shown that players with high defensive save rates also tend to have high defensive error rates. My theory is this:

  • Players with poorer positional or anticipation attributes find themselves in situations calling for dramatic “last ditch” defensive action;
  • Sometimes this comes off as a saving tackle or final block – “looks good for the cameras”;
  • Sometimes it does not and they are too late or poorly positioned to stop a shot/goal;
  • The converse also appears to be true – some players have a low defensive save rate and low defensive error rate. Lustig is a great example. If you are in the correct position, or anticipate forward movement or ball placement well, then you are not required to commit to dramatic defensive action (which may or may not be successful but because it is “last ditch” is inherently risker); and
  • Good defenders do not need to execute saving tackles and last-minute interceptions because they are generally well positioned and communicate effectively with their colleagues to ensure solid defensive shape. Simunovic is another with relatively low error and save rates, although he is occasionally rash attempting to win possession from unlikely positions.

This data also highlights another aspect of the art of defending and why the bare numbers require some context and explanation. O’Connell and Sviatchenko (and to a lesser extent Boyata) are defenders that I will call proactive. The proactive defender will try and anticipate the flight of the ball or where the pass is going. They will engage the forward physically when they are in possession. They therefore make a lot of active decisions – i.e. decisions resulting in direct action as opposed to holding position or jockeying the attacker. Not all the decisions are good ones, because they are human! The more actions committed to the more risk of a mistake or misjudgement.

Conversely, a more passive defender, like Lustig, engages less defensive actions per game. Perhaps through experience, or more conservative decision making (choosing not to engage is still a decision), they focus on position, body shape, team shape more than winning the individual battle. Thus, they overall reduce risk but perhaps do not affect as many eye-catching defensive actions. Simunovic is generally more passive, but occasionally is overly proactive. He is also capable of executing tackle of the season!

T-shirt available from Celts Are Here.

Toure is a defender with low defensive action rates overall. He is hugely experienced and relies on organisation skills and communication to ensure collective good defensive shape. Toure has the lowest challenges per 90m. He has the lowest rate of interceptions, but choses wisely as he is miles ahead of his peers in his interception success rate. His high error rate suggests that, perhaps, at 36, his legs cannot move as his brain would like at this level.

C:\Users\Alan\AppData\Local\Microsoft\Windows\INetCache\Content.Word\toure.jpg

Boyata is, based on appearances and selection pattern, our current number one centre back. Yet he has the lowest defensive action success rate. Firstly, he should be commended for reducing his error rate from a previous of 0.94 per 90m down to 0.51 which is only 0.03 more than Simunovic and Sviatchenko. Secondly, he has reduced the challenges and intercepts he is getting involved in from 9.52 per match to 8.45 per 90m. Here is a plyer learning to be more circumspect and not committing to as many proactive defensive actions.

You will note when we discuss the defensive pairing combinations below, that they tend to include one “proactive” and one “passive” defender as per my definitions above. Boyata and Sviatchenko would perhaps be hair raising (and only happened seven times), whilst Simunovic and Toure may not be assertive enough (happened once). The most common pairings were Simunovic / Sviatchenko and Boyata / Simunovic.

Boyata and Simunovic do not come out top on any single data point I have covered here and tend to be in the middle. When you consider also the defensive error and success rate, Boyata and Simunovic appear to be the “safest” combination, and football managers are inherently conservative. Shame about O’Connell though.

And finally, Lustig looks a solid bet for a future centre half as he enters his 30s and the demands of the Celtic full back position become onerous for him.

Crime Count

Lustig, and Toure, our more circumspect defenders, give away significantly less fouls when playing at centre back. A centre back can be expected to commit more fouls than most as sometimes, as the last line of defence, you must “take one for the team”. On the other hand, you don’t want centre backs conceding to many fouls as they will generally be in dangerous attacking positions.

Simunovic, although more of a covering defender than a centre back that likes to engage a centre forward, commits the second most number of fouls. He undoubtedly can be rash on occasions and commit to challenges with little prospect of wining the ball. As the youngest of the established centre backs at 22, this is a clear learning opportunity.

Unsurprisingly, Sviatchenko, O’Connell and the adapting Boyata, as our more aggressive proactive defenders have much higher rates that Toure and Lustig. Compared to previous seasons, Boyata’s foul rate has increased from 0.91 last season. Stats, eh? Although, to be fair, his foul rate is hugely impacted from the last game at Ibrox when he was awarded a season high 6 fouls against, mainly for Garner throwing himself to the ground repeatedly (Boyata may have gotten fed up and left a couple on him).

Passing and Creativity

Centre backs should have high passing completion statistics. They are the last line of defence, are usually the player selected for physical attributes over high technical ball playing ability, and operate in an area of the pitch where to concede possession would result in an opposition transition against an unorganised defence and be very high risk. Therefore, central defenders complete a lot of passes as they are the least likely to be pressed by the opposition unless you play Barcelona. I noted in “Deeds of Boyata” how he had adjusted his game to play a much simpler style. To complete the highest number of passes and at the highest completion rate is still commendable. Boyata also weighs in with the best assist rate, admittedly from a very low sample.

It is noticeable that Lustig’s creative instincts have not been tempered when played at centre back. Although not registering any assists when playing centre back he nevertheless dominates the peer group in creating chances or attempting to. He has the lowest pass completion rate of 90% indicating he tries more higher risk passes when playing this position (his pass completion from right back is 86%).

Otherwise there is not a lot to choose between Simunovic and Sviatchenko. Toure concentrates on defending firstly, secondly and to infinity!

Goal Scoring

In tight games, and against defensively deep opponents, a possession dominant team like Celtic needs to have the option of a goal from a set play – a corner or a free kick played into the box. And Boyata has delivered that this season. Scoring at only 0.01 goal per 90m below Forrest and 0.02 below McGregor, Boyata has had good quality shots and exceeded his expected goals (xG). Some context: the go(a)ld standard here is Van Dijk. In seasons 2014-16 he averaged 0.15 goals per 90m. When Van Dijk scored 10 goals in 2014/15, he did so at the rate of 0.18 per 90m. Boyata scored 6 goals last season at 0.15 per 90m. Boyata’s 0.25 may be a high-water mark for some time.

When bursting back into the team in January, he memorably produced winning goals against both St Johnstone and Aberdeen at home in tough matches. His goal against Aberdeen was the only Celtic shot on target of the match. Towards the end of the season, his link up with Griffiths on corner taking duty created four chances in one match away to Aberdeen, with 1 goal. A similar chance was missed in the Scottish Cup Final.

No other defender has come close to this. Sviatchenko threatens the goal the most and is under performing his xG. His 1 goal of the season was memorable, however, being perhaps the longest range shot Celtic scored from, at home to Ross County.

Lustig is a goal threat whatever position he plays, and had 0.09 goals per 90m over the season. Simunovic is often the defender not to come up for corners and free kicks and consequently has a low number of attempts. Even his 1 goal required deflections to score against Dundee.

Partnerships

Celtic used different pairings in the first five matches, and only in the seventh and eighth matches of the season did the same centre backs play two matches in a row (Toure and O’Connell – a season really is a long time in football!).

Highlighting how much of an achievement it was to quality for the Champions League, consider the partnerships in place for each qualifying game.

Ambrose and O’Connell together away in Astana! Sviatchenko was injured for a spell early in the season, Simunovic until mid-September and Boyata out until late October. It is not surprising Rodgers brought in the experienced Toure as a short-term measure. Even though he only appeared in seventeen matches, he helped guide the team into the Champions League and through the initial domestic games. If he had only played the one match away in Be’er Sheva he would have earned his contract for the year.

The combinations used were:

No centre back pairing played together for more than four matches in a row. Simunovic and Sviatchenko did not start a game together until the fifteenth match of the season. Boyata and Simunovic did not appear together until the thirty-seventh match of the season. Thirteen different partnerships in fifty-nine matches.

Boyata and Simunovic appeared to be the first choice by the seasons end although they have never played more than four consecutive games together. Not only did Lustig play infrequently at centre back, he also played with five different partners in seventeen starts there, and only played once with three players. O’Connell appeared with four different players in seven starts.

Partnership Defensive Action Rate

On the face of it, the season ending partnerships have the lowest defensive actions success rates, and the “occasional” partnerships the highest. Also, the partnerships with the highest defensive save and error rates score highest and those with the lowest defensive save and error rates score lowest. Small samples are not helping here, so bear that in mind.

Another view might start to explain what is happening. Here are the goals conceded and shots given up by each partnership, per 90m:

On this basis, the table is flipped on its head. What this suggests is that those partnerships with most defensive actions to perform will come out on top in defensive action rate. It is the same principle as using shot on target save rates to judge goalkeepers. On that basis, the goalkeepers from the poorer teams face the most shots and their goalkeepers tend to make the most saves. Teams that are dominant and face fewest shots have less saves for the goalkeeper to make and hence by not being as busy have less opportunity to build impressive numbers.

If we look at the opposition faced by the partnerships, it may help to explain this. Toure and Sviatchenko only played together five times – H v. Aberdeen, H v. The Rangers, A Barcelona, H v. Manchester City and H v. Borussia Moenchengladbach. Not only are those matches amongst our hardest of the season, the Barcelona game is clearly an outlier, resulting in a 0-7 defeat. This match was also remarkable in that the centre backs did not lose a single challenge – 11-0, as Barcelona played through and around the defence. They also succeeded in performing 21 intercepts as the dominate Catalans swarmed forward. Remarkably, if you discount defensive errors and saves, the centre back pairing was 32-0 in successfully executing defensive actions – and we lost 7-0! In short, the defensive data I have viewed in isolation, does not explain well the reason for the defeat. The centre backs conceded 0 fouls which begins to hint at the reasons as Celtic sat off Barcelona and allowed them time and space to construct moves between the lines both vertically and horizontally. Barcelona created 8 clear chances and scored from 10 shots on target. Great players given time and space to play.

Lustig and Sviatchenko played together six times and only once at home to Lincoln Red Imps. The five away games were to Inverness Caley Thistle (2-2); Ross County (4-0 but Ross County had numerous chances); Borussia Moenchengladbach (1-1); Partick Thistle (4-1) and Hamilton Academical (3-0 but in many numerical ways one of the hardest games of the season and with 10 players for much of it).

Boyata has not appeared in any Champions League matches and therefore his partnership data is only inclusive of domestic matches. Hence the lowest goals per game against and shots against, defensive save and error rates. Celtic have only conceded 10 goals in the 21 matches Boyata has started.

Summary

I would conclude that the data favours players having a lot of defensive actions to perform. Where there are few defensive actions to perform, one or two lost tackles can skew the overall defensive action rates. Hence the need to consider the fullest possible picture such as goals against rate and number of shots given up.

Considering the individual and partnership data I would summarise:

  • Boyata is improving, playing a simpler game, making better decisions, but is untested by Rodgers in the Champions League.
  • Simunovic is the youngest, is susceptible to injury, can be rash, but at 22 is immature for a centre back and has great potential. He only appeared in two Champions League matches this season.
  • Sviatchenko suffered a late season wobble in confidence, but when called upon performs at a similar level to his peers, and remains an experienced option.
  • Lustig shows great potential as a centre back, but continues to be the first choice right back.
  • Toure provided stability in the early part of the season and helped steer a variety of partners through the qualifiers. His performance in Be’er Sheva was worth his annual contract. However, with 6 defensive errors in his last two full matches, and at 36, there are doubts whether his career will be extended.

The second conclusion I would draw, therefore, is that we still don’t have an irrefutably compelling partnership such as Van Dijk and Denayer. None of the current partnerships, nor the players in them, have significant Champions League experience. Whilst centre back may not be a “problem” position for Celtic, there remains the potential for considerable improvement, and the need to settle on a definitive “first choice” partnership.

Leave a response

Your email address will not be published. Required fields are marked *