This is the final review of the Brendan Rodgers’ return season. I have been comparing performances in various aspects to Ronny Deila’s sides, Rodgers’ first term, Neil Lennon then Ange Postecoglou’s. Having considered team style, we looked at defending, creating, and opposition threat. Finally, let’s consider Celtic’s pointy end - the attack.

In the ten years I have collected data, Celtic have managed over 100 goals in all competitions. This season saw the second-lowest number of matches (50) yet the champions converted 115 times. 2.3 goals per game surpasses Postecoglou’s initial term (2.28). I make that point as I feel Rodgers’ first season resembled the 2021-22 campaign in many ways given recruitment and injury difficulties.

So, not bad then? It’ll take more than simple goals per game to convince you.

Shot accuracy and conversion

The following shot-related metrics have been around for many years and are like possession statistics in being useful in building an overall picture, but we’ll need more advanced metrics to understand what is happening under the covers.

There is no real correlation between “success” and these metrics and so it is unclear whether we are seeing manifestations of variance or “luck” or examples of “great finishing”. For example, the highest on-target conversion percentages were 2019-20 and 2022-23. Those were Odsonne Edouard’s and Kyogo Furuhashi’s most productive campaigns. This illustrates that from an attribution perspective, it is a bit of both.

Overall, this season was bang on the averages – 39 per cent of shots on target, 34 per cent of on-target shots converted to goals, and 13 per cent of all shots converted to goals. So, we’ll need to dig deeper.

Expected goals

The following illustrates the volume of shots (total shots per 90 minutes) and the quality of them (xG per 90m).

Rodgers’ first spell was characterised by a high volume of shots but with a relatively low average xG per shot. His team contained the likes of Stuart Armstrong, Ryan Christie, and Patrick Roberts who were all fond of a “dig”. Lennon’s teams had a lower volume of shots with similarly average xG. Under Postecoglou, the age of data analysis in shot production was upon us. Shot volume was average but the quality of the chances being taken to shoot was high in comparative terms.

Rodgers’ side has continued in that vein albeit although shot volume has increased slightly, the average quality per shot is down. Given the quality downgrade in the squad that is understandable. Nevertheless, a solid foundation.

I have no doubt that advanced metrics now permeate most clubs and even Celtic. The average xG per shot under Rodgers in 2016-17 was ludicrously low by modern standards. This value has risen steadily and peaked at over 0.15 under Postecoglou in his second season. It dropped back this term, and this is one to watch for the coming season.

I discussed the (miss) use of Kyogo throughout the season and the inability to provide him with the type of low, fast, six-yard box service he needs. Small changes can have transformative impacts, however. If Celtic are back nudging 0.15 xG per shot and maintain shot volumes, we can hope to see goal returns like those the Australian achieved.

The difference between goals scored and xG is fertile ground for a whole lotta mischief including “finishing” (whatever that is), goalkeeping performances, refereeing, and general variance/luck. Attribution is a lottery!

Rodgers hit the lucky tree hard in season one, with 17 goals from outside the box over the season and a massive 0.53 overperformance on goals scored relative to xG. He inevitably suffered the mean reversion fairy's revenge in the next two seasons. Lennon’s 2019-20 campaign was thrilling and fortunate as his side conjured 0.19 overperformance relative to xG per game. Postecoglou was fighting the elements in year one and made serene progress in year two with a relatively steeled side.

Rodgers experienced a 0.24 goals per 90m shortfall relative to xG created. It is funny (not funny) that in seasons of personnel disruption “luck” seems to desert you, but when you have a settled and fit quality side, it shines brightly. In other words, not surprising at all. We can hope Rodgers gets in players he trusts with an uptick in quality and fortune with injury. The xG fairy thrives on positive vibes.

Attacking threat

I covered creativity in the previous article. Let’s consider whether Celtic’s attackers are giving themselves the best chances to score.

Are Celtic players getting into the optimal positions to score? Below we plot the volume of possession or touches within the opposition penalty box per 90 minutes with the number of big chances (a subjective assessment of each chance – think “he must score”!).

This season saw the highest volume of touches in the box and of big chances created. Great news! This had risen steadily under Postecoglou, but Rodgers’ side eclipsed even those values. Celtic are creating double the big chances compared to Rodgers’ first term and possession in the opponent’s box is up more than 30 per cent .

With those trends, we’d expect to see higher xG and therefore higher overall goal returns. The following illustrates the trends in shots taken inside and outside the opposition penalty box.

It is interesting that in the ongoing Euro 24 competition, there has been an uptick in goals from shots outside the box. I am not sure that translates to a club football trend where managers have more time with players to work on attacking patterns. But it might be a small reaction to the growing trend for teams to work the ball into a better position before shooting. Inevitably that leads to a certain same-ness and even predictability.

Those trends are clear at Celtic where the number of efforts within the box is trending upwards and those outside the box trending downwards. Shots inside the box volume was the highest I have records for whilst those outside the box rose very slightly on last season.

The trends are that Celtic are giving themselves better shooting chances by taking a higher proportion and volume of shots inside the box.

Celtic are taking shots much closer to goal than any time over the last six seasons. Each shot is roughly two metres closer to the opponent’s goal than in 2019-20. This is, again, likely a data-driven trend – simply you have a greater chance of scoring the closer you are to the goal. A slight caveat is that it generally gets more crowded the nearer the goal you are, and blocked shots are up to 5.12 per game, the highest since the 2017-18 season.

So, what’s the problem?

All the above illustrates that Celtic are once more in a good place at the foundation level. Doing many of the right things in terms of shot decision-making. However, they need a better quality of execution.

This maps the volume of ball losses without creating anything in the opposition defensive third with the volume of opposition goalkeeper saves per 90 minutes.

This season saw the second-highest volume of the opposition's keeper saves and final third losses. (Lack of) saves by the opposition goalies was something the 2022-23 team benefitted massively from. In general, the league saw an improvement in goalkeeper quality and for Celtic they experienced some mean reversion in ‘keeper play.

Final third loses is reflective of wing play and is a topic I returned to frequently over the season. Luis Palma, Yang Hyun-jun and Nicolas Kuhn were very wasteful in the final third compared to players such as Jota and peak James Forrest. Celtic rely on their wingers as goalscoring forwards, and this wastefulness was a key feature of the campaign from that part of the team.

Finally, StatsBomb’s on-ball value metric attributes value to shots taken in terms of the extent to which they increase or decrease the chances of a side scoring. Think of it like an xG value for on-ball actions. As regards shot OBV, StatsBomb’s model has a very dim view of the quality of the Celtic shooting. Albeit the overall average per game has improved from -0.22 to -0.1 this term.

My honest assessment on this is slight scepticism given the trends are in the inverse direction to the other shooting trends I have laid out above. Meaning, that perhaps their model is being refined season on season.

Summary

The trend across my season review articles is that the team performance data generally points to a strong foundation of the fundamentals. Shooting is no different. Like Postecoglou in season one, it seems likely Rodgers does not yet have the squad he wants, nor is the team functioning precisely to his instruction.

If (big IF) the recruitment is successful in the summer, I’d expect cohesion and familiarity to build on the foundations leading to a season like Postecoglou’s second.