One of the reoccurring challenges of football analysis is that of attribution. What factors were most material to an outcome? We saw why it happened, but why and what was most impactful?

Specifically addressing this problem to the weekend’s Glasgow Derby: did Celtic’s defence play well, or was the Rangers attack misfiring? We will attempt to create a framework to assess that. Bear in mind, that this is a discussion mainly about the second half.

So passive was the Rangers approach in the first half that Celtic were put under next to no pressure (legally – one clear offside goal and one created after a simple foul being rightly disallowed) apart from a self-induced 35-yard shot from Rabbi Matondo in the 42nd minute thanks to Joe Hart and a late corner and header from Kemar Roofe.

READ MORE: Celtic's Matt O'Riley was the difference in derby win

In the second half, Celtic sat deeper, and in what was a must-win game for Michael Beale, the Blues threw the proverbial kitchen sink.

The Celtic defence

A short version of the article could conclude ‘Clean sheet for Celtic, job done.’ but we all know that says little of performance.

A look at the average position pass networks shows Celtic had to drop much deeper in the second half as a rotating cavalcade of forwards were thrown on in search of an equaliser.

The team was a good 20 yards deeper in the second half, with the two inexperienced centre-backs Gustaf Lagerbielke and Liam Scales essentially guarding the 18-yard box.

Also note that in both halves, Matt O’Riley was slightly deeper than Callum McGregor, forming a double pivot in the first half and a tighter line of three in the second. The attacking players largely maintained their positions keeping the counter-attacking options open.

Given the rag-tag nature of the Celtic backline (Alistair Johnston perhaps 80 per cent fit; Lagerbielke’s third-ever match and first derby; Scales arguably sixth-choice centre-back and Greg Taylor badly out of form), it is not surprising there were six big errors from the back players. This compares with the 2.08 average in 2022-23.

Where Celtic were perhaps fortunate is that the impact of those errors was not reflective of the volume, meaning last season those 2.08 errors resulted in an average 0.22 xG conceded. Whereas, on Sunday six errors led to 0.05 xG, a much lower average xG per error. Is this a case of Celtic getting away with it, or were the home side’s forwards incapable of capitalising?

Joe Hart was forced into three saves, with a post-shot xG of 0.89. Two late efforts from Danilo were particularly well dealt with. Hart averaged 1.58 saves per 90m last season, but you could argue this was a poor return of shots on target (three) from such an important game (none in the first half).

Celtic were tenacious as the volume and percentage success of challenges won and lost shows.

This game is compared to 2022/23 averages again.

This was a tie where Scales shone, especially in the second half. Asked to be primarily an 18-yard box defender, he excelled. Scales led the team with 11 successful challenges and interceptions, 10 clearances, and 21 recoveries, but also had the highest passing percentage success rate (90 per cent) and provided the most pack passes (13). In the second half, he only completed five passes but 15 defensive actions.

Naturally, Celtic need technical ball-playing defenders and Scales is never that. But whatever happens in his career, he will always have September 3, 2023, and that second-half showing.

Perhaps most telling though was the number of turnovers Celtic either won or were gifted by the home side. 20 packing recoveries compares with an average of 12.38 last season. The worst culprit was Todd Cantwell, turning it over six times, but to be fair he also had four pack recoveries, one less than Nicolas Raskin.

The pack turnover score of 160 compares to an average against Celtic of 135 last term. Credit is due to Matt O’Riley, who either forced or was the beneficiary of an incredible seven pack recoveries (last season he led the squad with an average of 2.3 per 90m). Not far behind was Odin Holm, with five in only 26 minutes! This gives hope that a midfield three of McGregor, O’Riley and Holm may potentially have the defensive and turnover capability to be competitive in Europe.

Finally, a word on Daizen Maeda. He was much more aggressive in his attacking in the first half, being willing to push James Tavernier back whenever he got the ball. As per usual, his final third actions were hit-and-miss. Defensively, Maeda did a sterling job of shackling the player who is - by some distance – Rangers’ most dangerous attacker. Rangers’ captain did create two chances with a low xG total of 0.07, but those were from set plays. He also provided two secondary assisting passes. His one shot was also from a set piece way outside the box.

Maeda completed 30 pressures, nine more than the next highest by any player on the pitch. He also completed seven counterpressures, two more than anyone else.

If we look at the home side’s average position passing network:

We see the extent to which Tavernier was utterly shackled and barely averaged higher than the halfway line even in the second half. Neither the manager nor the captain came up with a means to reverse that pattern of play. Some eye-catching Celtic defensive performances for sure, but how wasteful were the home side?

The Rangers attack

The Blues generated 1.64 xG from 16 shots but only three on target. Last season they averaged nine shots versus Celtic but with four on target generating 1.38 xG. Here is their shot map from the game.

You will notice that the StatsBomb xG model gives the Sam Lammers chance 0.75 xG. This is too high and would be accurate at the time he received the ball. However, in rounding Hart, he narrowed the angle considerably with the ball running away from him.

What is striking is that Celtic restricted home chances inside that danger zone in the centre of the box. There were a couple of low-quality headers from corners and crosses, but only really Danilo's slightly scuffed effort from near the goal. Again, credit to Scales and Lagerbielke for defending their box, but the continual diet of long straight balls and crosses from deep did not perhaps provide the ideal platform for high-quality chance creation.

22 failed crosses compared with 13 averaged against Celtic last season and 10 from all opponents. Three were successful. This failure to build through the lines is reflected in a packing receive score of 324, compared to an average of 390 in games against the Champions last season. Rangers were content to go very early and very direct to put pressure on Celtic’s rookie centre-back pairing.

My own qualitative assessment of watching early games from Rangers this season is that the ball rarely sticks up front with the array of new attacking talent brought in. Celtic won the ball back 28 times in their own defensive third versus an average of 11 last season. Partly, this was down to the tenacious defence but also the home side's lack of ability to retain control under pressure.

Summary

There is no doubt that Celtic collectively defended their box effectively in this second half.

Individually there were some standout performances:

  • Scales defended his box using all parts of his body.
  • O’Riley and Holm generated a huge volume of turnovers.
  • Maeda shackled the main danger man for the home side, and
  • Hart made the saves he was called upon to do.

However, the lack of final third quality, especially from the new signings for Rangers, was most telling. The very highly remunerated Roofe looks like the main threat despite all the new acquisitions.

Context is also key. The home side's highly passive out-of-possession and hyper-direct approach in possession yielded no meaningful chances at all in the first half and none on target.

In a game where the home side had to win given:

  • Rodgers’ injury-hit side and lack of availability of new singings.
  • Elimination from the Champions League financial bounty.
  • Sketchy performances hitherto.
  • The risk of a four-point gap this early.
  • The huge reward on offer for this season’s Champions; and most tellingly…
  • Beale’s inability to win a match of serious consequence thus far.

All the above made that first-half performance inexplicable.

READ MORE: The Celtic numbers: How Kyogo and his teammates won at Ibrox

Yes, they roused in the second half, but from around the 15th minute onwards, the home side's mood turned ugly and with no Celtic fans present to aim vitriol at, that invective made its way down the terraces and onto the home players. Celtic were very sloppy in that second half losing the ball 14 times in their own defensive third (last season's average was five). However, the home side did not capitalise and held to launching crosses and long balls into the box.

Coming back to the initial hypothesis, we are veering towards 40/60 in terms of attribution. Yes, Celtic had some excellent defensive performances but were very untidy and error-prone. The home side's attacking strategy was unclear and unvaried despite the changing personnel.

As Neil Lennon said at the end, Celtic will get much stronger going forward but where does a full-strength Rangers go from here?