Celtic's Performance Data: The Case for Change
The points raised in this blog were discussed on Episode 4 of The Huddle Breakdown. Please follow, share and watch/listen.
We are now in a place where surrendering a 2-goal lead and conceding another 4 goals is considered “excellent at times” and “a lot to be positive about” by Lennon. Results wise, Celtic’s current run is unprecedented in modern times with 2 wins in 11 and 27 conceded with 1 clean sheet.
But we are about performances! So, what does the key performance indicators say about Celtic this season?
Performance Indicators
It is convention for media to summarise a match into a stat box. For example, the BBC give you shots on target, total shots, possession, corners, fouls.
When I summarise a match for myself into a virtual “Stat Box” I use the following:
KPI Rationale Possession Indicative of game and ball control and provides in sight into style (possession-based; counter attacking etc) PPDA Pass Per Defensive Action indicates pressing effectiveness both for and against Total Impect Indicates the extend teams are able to play through opponents – a score of forwards passes and how many opponents you take out the game. Total Shots You won’t score if you don’t shoot! Simply shot volume In Box Possession The number of possessions or touches in the opponent’s box as this indicates you are getting close to goal! xNPG Expected Goals excluding penalties. An indicator of the quality of the shots taken and how many goals you would expect given a “normal” outcome. xA The quality of the chances created. Irrespective of the shot outcome, what is the quality of the chance the passes provided.
I find reducing a game to those 7 KPIs tends to give an accurate summary. Others may disagree or add / subtract from this list. That’s fine.
Celtic’s Performance
Let’s get to it. Simply, I’ve compared last season (treble and progress in Europe) with this.
First the good news. Celtic are still largely possession dominant, having 63% of the ball across all competitions (66% in the SPFL).
But as the other stats show, this tells you something, but not a lot! PPDA is up 6%. This means Celtic are making a few more passes per opposition defensive action. This would tally with Lennon’s desire for the team to be more “front foot” and forward orientated. Similarly, Impect up 3% means the side are getting the ball forward and taking opponents out the game to a slightly higher degree. And in isolation those would be highly encouraging numbers.
But that is where the good news ends.
Shot volume is down 11%. Possession in the box down 21% and consequently the quality of shots taken is down 20%. Not surprisingly the quality of chances created is also down, 18%. Remember: football like many sports is a game of small margins. Movements in these KPIs of a few % points either way would pique my interest in normal times. These aren’t normal times and Celtic’s attacking performance in general, has collapsed.
Opposition Performance
Let’s look at the other side of the equation. How are the opposition performing against Celtic? Again, this covers all matches in all competitions.
Essentially, we see the opposite trend.
Celtic are easier to play against. How do we know this? The opposition are completing 7% more passes per Celtic Defensive Action, and their Impect average is up 12%. The latter means that the opposition are playing through Celtic’s lines, taking players out the game, to a greater degree (up 12%).
There is a consequential uptick in opposition shot volume (up 8%) and they are getting touches in Celtic’s box a whopping 20% more than last season.
Most worrying is that opposition xNPG (i.e. WITHOUT all these penalties Celtic have been conceding) is up 23%.
Finally, chance creation quality is 19% higher than last season’s bench mark.
What Does This Mean?
As mentioned, small changes to any of these KPIs would be noteworthy. These are huge shifts in performance and consider: this is the whole season not just the current bad run (in fact for full transparency this DOES NOT include Sparta Prague away (1-4), Ross County (0-2) or AC Milan away (2-4)). In reality, today, these numbers are probably worse.
We are seeing a 40% downward swing for Celtic in some key metrics including xNPG.
If this were to be translated into points (and performance impacts results – oh yes!) instead of challenging for the title in the SPFL, Celtic would be scrapping with Aberdeen and Hibernian for 2nd / 3rd place. Last season, Celtic might just have pipped Motherwell and Aberdeen to 2nd by these performances.
As @jucojames said on the Huddle Breakdown pod, once you get a game xG differential lower than 1 (i.e. Celtics xG is < 1 greater that the oppositions) then, given the nature of a low scoring sport and the multitude of things that can happen by chance in a football match, each game becomes a bit of a lottery. We are now in that place.
Given the disparity in wages and squad value (despite what some learned folk may tell you, this does impact outcomes quite considerably!), that is a calamitous fall off.
Conclusion
Like Celtic, there is no defence against the performance data. The need for change is unanswerable.
What that change is will often depend on your cognitive biases (pro/anti Lennon, pro/anti board, pro/anti the coaches) but muddle along and do nothing cannot be the answer.