Why there is a gap between Celtic and Rangers this campaign

Both used statistics from the game to show that there was not as much of a gap between the sides as the scoreline or the sentiments from hordes of bluenoses on fan forums and podcasts would have you believe.

Clement, of course, needs to protect his scarred players and build positivity where he can, whilst Souness is a fan. Nevertheless, it is encouraging when data enters mainstream debate.

Stat box

Let’s start with the summary Statsbomb produces.

Expected goals (xG) have become relatively mainstream. It is most effectively used over a long period as single game xG is subject to all sorts of variance and context as will be shown.

A further complication is the diversity of xG models. StatsBomb has Celtic a healthy 2.18 to 1.31 up on xG. Opta had it at 1.55 to 1.24 in the Champion's favour. My rudimentary model had the away team coming out on top 1.32 to 1.70. If we take the average of all those, Celtic scored 1.52 and Rangers 1.42. Certainly, much closer than a three-goal scoreline suggests.

Souness pointed out that his favourites achieving 47 per cent possession at Celtic Park was an indicator of performance parity and given Celtic’s prior average of 74 per cent possession, perhaps we can acknowledge that. Indeed, their pass accuracy was one per cent superior to the Hoops’ efforts. Furthermore, as he correctly states, overall Celtic managed 14 shots at goal and the Light Blues’ 13.

England hopeful Jack Butland only made one save in the match! What was Celtic doing, sleeping? Meanwhile, great Dane Kasper Schmeichel was forced into three saves. Phew! So, do the Rangers legends have a point?

Shots

We’ll start with the overall chance creation and shot data. Firstly, here are the shots from both sides.

The key piece of data, if we are measuring quality, is the ExpG/Shot (open play only) to the top left on each team’s maps. Celtic’s efforts averaged 0.18 xG per attempt whilst for Rangers it was only 0.12. Therefore, there is a significant difference in the quality of the chances created with the home sides being higher calibre.

Rangers’ best chance fell to Ross McCausland in the 90th minute when an Alistair Johnston block hit Callum McGregor and ricochetted fortunately to the young Northern Irishman right in front of goal. An alert and sprawling Schmeichel save protected another Celtic clean sheet. It was already 3-0 at that stage with only added time to play. One would expect the challengers to be in “kitchen sink” mode.

Let’s consider the game state as regards shot outcomes. Here are Rangers' efforts when the game was level.

Rabbi Matondo’s early header was the only meaningful effort before Celtic took the lead. Headers are always lower xG due to the technical difficulty in heading the ball, and whilst he was under pressure, he also seemed more concerned with not being hammered than in scoring. Here are the Rangers' efforts when behind.

Most of their efforts came in the second half. The shot count was 10-6 in the away team's favour after the break when Celtic were 2-0 ahead. The only slightly reddish one of note (denoting higher xG) was the aforementioned McCausland added time chance. Again, we’d expect a reaction and for the team losing by two goals to take some risks in attack.

Whilst the game was level, Celtic attempted the following efforts.

They scored with their first meaningful chance, a cutback from Johnston met by Daizen Maeda whose shot went under Butland. From that position, Celtic’s shot map when winning:

Five of those chances came after the first goal and before halftime, with no opposition efforts in between, and resulted in a comfortable two-goal cushion at the break.

The xG race chart shows that once Celtic got over the shock of losing the ‘Ten Minute Cup’, they put in sufficient effort to be comfortable in the game. Once Celtic had their three-goal cushion, they did not generate another shot, Rangers had five accounting for half their xG (0.7).

Therefore, in terms of game state, aside from the first ten minutes, Celtic built a lead and then were content to manage it. Rangers' chance creation was mainly when chasing the game, but their efforts fell well short of a siege on the home goal.

Packing

The packing data I collect covers a multitude of critical on-ball activities. Including forward passes provided and received; dribbles that take players out of the game; turnovers and recoveries that result in players being taken out of the game. Here is the xG and packing race chart from my dataset.

The green line is Celtic’s cumulative overall team packing score as it changes per ten-minute game segment. The blue line covers the away team. Again, aside from the opening ten minutes, Celtic had a larger packing score, and this only narrowed once Celtic had established their three-goal lead in the 75th minute. Similarly, the xG patterns only change from the same game time.

Context is key here. When the game was tight, Celtic produced dangerous moments and built a healthy lead. Only when chasing a deficit did the away side threaten, and then not very effectively.

Pressing and passing

There is no doubt Rangers pressed the Celtic back line effectively in the opening ten minutes. Indeed, over the match, Celtic gave the ball away 12 times in their defensive third. However, from those turnovers, they generated only 0.25 xG.

Many of Celtic’s better contributors on the day had bad moments in their defensive third with McGregor getting caught in possession three times, Greg Taylor the same and Paolo Bernardo twice. This is where Souness was correct in his appraisal - that Rangers lacked the quality in the attacking third to take advantage.

Celtic had the majority of the possession and given the lead they held from early in the game, did not have to press too aggressively. But this was another facet where quality overcame quantity. Rangers executed 212 pressures to generate 31 regains – 15 per cent of their pressures resulted in winning the ball back. For Celtic, it was 33 regains from 145 pressures – a 23 per cent success rate.

As Clement acknowledged, Celtic players have been together as a unit for some time now and this superior pressing effectiveness results from that. Both sides completed more passes in their own half than their opponents. Celtic completed 159 out of 443 completed efforts in the opposition half – 36 per cent. For Rangers, this was 145 from 389 or 37 per cent.

But, again, Celtic, from the 17th minute, were comfortably defending a lead, whilst Rangers should have been pressing high and hard and aggressively into the Celtic half. Towards the end of the first half, shortly after the second goal, the Celtic backline and McGregor were passing the ball around as if it were a training exercise.

Summary

Celtic, in five matches this season, has scored seven goals by the 20th minute and ten by the 40th minute of matches.

With only one conceded in the League Cup versus Hibernian, this has allowed the Champions to control matches and keep opponents at arm's length with controlled possession whilst simultaneously not being required to throw bodies forward overly aggressively. So it was on Sunday. Celtic produced an attacking threat when they needed to and exerted control for the majority of the contest without being under undue pressure.

The Rangers fans know this, their pods and forums are clear. Clement must protect what are probably mentally bruised players, physically almost had the Ibrox reception committee gotten hold of them.

But Sunday’s game was a classic case of focusing on quality, not quantity in the data.