Why Celtic must improve defensively under Brendan Rodgers

Celtic conceded 51 goals in 50 matches across all competitions. 27 were conceded in 38 league matches and, of course, a much higher proportion conceded in the Champions League. 1.02 goals against per game is the second worst in the ten seasons I have been collecting data with only the chaotic 2020-21 season (1.08) being higher. 1.12 xG conceded per game is the highest since 2016-17 apart from the 2020-21 season.

A theme of the campaign was a failure to kill games off, often conceding goals when in promising positions. On the face of it, there’s much to do defensively so let’s smash into it.

Recap

As I covered here, Celtic are playing a much higher line under Brendan Rodgers, by as much as 1.5 metres. Secondly, the off-the-ball pressing is far more aggressive than we saw under Ange Postecoglou.

Both aspects can be considered defensive pillars, or building blocks, that we hope can be refined and improved upon with further cohesion and better players. These aspects also introduce risks. Given that platform, how does the defensive performance compare?

Opposition passing

Opposition pass volume was highest under the relatively passive defensive framework of Ronnie Deila, then under Rodgers decreased year on year. Under Lennon, it increased again, albeit percentages completed sharply fell.

Last season we are back to almost identical Rodgers year one numbers – 240 completed passes at 72 per cent completion. Given that possession is Rodgers' primary form of defence, we can expect that pattern to repeat with year-on-tear volume decreases in opposition passing.

Opposition creative passing

Is Celtic becoming easier to play through? We can assess the answer by mapping the volume and impact of pack passes by the opposition. Pack passes are forward passes that take at least one Celtic player out of the game. The pack passing score indicates the number of players bypassed and their position (i.e. more points awarded for defenders bypassed).

Opposition packing is far more voluminous and effective than under Rodgers’ 2017-18 and 2018-19 teams. I’d expect he knows this, and remediating will become a priority for the season ahead.

We are not quite at early Postecoglou levels whereby his ultra-attacking team were too easy to play through. But the numbers have risen from the second-season treble winners. Rodgers’ current team does press aggressively, which leads to pack passing opportunities for the opponents. In Postecoglou’s second season, long-ball-orientated sides in the SPFL struggled to break that press.

Does this increase in playing through Celtic’s lines lead to better chance creation? Firstly, let's consider crossing volume and success.

Rodgers’ final season saw the lowest volume of successful crosses if not the lowest percentage. Last season saw the highest volume of successful wide balls into the box and as a percentage of total efforts.

This, of course, has two major considerations. Firstly, allowing crosses into the box – the job mainly of the full backs and wide midfielders to stop at source. Secondly, the crosses increasingly find a target in the Celtic box. That is mainly the job of the centre-backs. On both counts, Celtic are allowing the opposition into their danger zone. Expanding on that point...

This view brings us back to packing and specifically the number of Celtic defenders taken out of the game by forward passes – this is the most vital statistic in the packing stat pack. I’ve combined it with the volume of successful passes into the danger zone – the area of the Celtic penalty box between the goals and the edge of the area – the zone of maximum opportunity.

For seasons with records, last term saw the most defenders taken out of the game by opposition passing, and the highest volume of successful opposition passes into the danger zone. Celtic were far too easy to create danger against. The packing data indicates the balance between attacking full-backs and cover defence. It reflects the increasingly direct nature of much of the league’s teams in terms of playing styles. Early, longer passes may take defenders out of the game. Finally, as reported, Celtic’s average defensive line is as high as ever.

The remedy is a mixture of familiarity with a high playing line, and better organisation especially amongst the fullbacks in decision-making on when to go forward and when to form part of the rest defence. Quicker and more anticipative defenders help.

Finally in the opposition creativity section the quality of chances they are managing from pack passes. That is, from forward passes that take defenders out, what is the average expected assist value per 90m?

Sides found it increasingly difficult to create chances when Postecoglou’s Celtic relentlessly hounded them into their own defensive third. Last season saw a record value of 0.356 expected assists following pack passes against the Celtic defence. It is indicative of the ease by which the champions were able to be played through relative to prior seasons.

This isn’t indicative of crosses or set plays lumped into the box, but the ability of the opposition to play through Celtic’s lines to create a shot. Again, indicative of overall defensive shape and alignment deficits.

Opposition goal threat

Did the increasing ease with which sides could play through Celtic manifest as better goal-scoring opportunities? Firstly, let’s consider the total volume of chances the opponents created and the number of touches they had within the Celtic penalty box.

20223-24 saw the highest volume of both metrics since records were kept. 13.6 possessions of the ball in the Celtic box and 6.82 chances created by opposition passes per 90 minutes. This is a far more porous Celtic defence by these metrics than the disastrous 2020-21 season. Did this manifest in better shots?

Before I get to that, notice the tight arrangement of data points at the bottom in terms of the lowest average expected goal value per opposition shot. Those were Rodgers’ first three seasons. This has not been the worst for the quality of opposition chances – 0.127 xG per shot is better than each of the prior three seasons. But last term was by far the worst for the sheer volume of opposition shooting permitted – 8.82 per game.

I am more concerned with shot quality than volume. Whilst this view may appear grim, focus your mind on the average xG per shot. Again, this is a data point I am sure Rodgers is painfully aware of and we can see steps being taken to reduce that potency within the opposition's chances for next season.

This view summarises aggregated data for opposition attacking threats. Firstly, the final third effectiveness is the percentage of times the opponent either kept the ball, got a shot off or won a throw/corner within Celtic’s defensive third. CAT score is short for Celtic Attacking Threat but in this case is the aggregated attacking threat for the opposition.

Last season saw the most efficient opposition final third ball retention. And saw the highest average attacking threat score. In short, Celtic are easier to play through and as a result, the opposition have seen more opportunities to score. The saving grace is that chance quality has not been super high.

Weaknesses

The weaknesses in Celtic’s team were identified last Summer, and if anything were exacerbated by recruitment. The overall shot map comparison between Celtic and their nearest domestic rivals for SPFL matches summarises it well.

Cast your eyes between the two maps back and forth. Two trends should become obvious.

1. Celtic allows shots far too close to their own goal.
2. Celtic allows high-value shots overwhelmingly from the left of their defence.

An ageing, relatively immobile goalkeeper and a porous left-hand side are no secrets. But this map shows the impact of those weaknesses. The reasons for the relative decline in Celtic’s defensive play are more complex and bring in the wing play and the midfield too.

Summary

Rodgers will undoubtedly recognise Celtic has been too easy to play against, and through, that the opponents are too easily creating chances and that the left-hand side is a concern.

There is nothing new in any of this and it opens the door for significant scope for improvement next season.