A tale of three centre backs.
Celtic’s second most expensive Summer import was Carl Starfelt at a reported £4.5m from Rubin Kazan. After Furuhashi, this was a big investment by Postecoglou.
Many studied the data, including me, and thought, “he’s not an aerially dominant behemoth like Jullien”, but looks a good fit for what we thought at that time “Angeball” might entail.
Cameron Carter-Vickers arrived on the last day of the transfer window as a seemingly rushed and last-minute loan deal from Tottenham Hotspur. An itinerant loanee, he’d done well in the physically tough English Championship with numerous clubs, but he wasn’t very tall and his squat frame resembled a body builder.
Stephen Welsh has come through the Celtic academy and was pitched into many difficult situations last season. Another relatively short centre back, he made the errors and learnt some harsh lessons as one might expect a 20-year-old thrown into a dysfunctional side. This season he has again been called upon in tough European ties.
All are well passed the 900-minute mark this season.
Half term report time.
Defensive Actions
My benchmark for purely defensive action performance is to plot Defensive Action Success Rate (winning tackles, headers, intercepting the ball, blocking shots etc) expressed as a % of the actions “won”. That is, irrespective of which team then gets possession, what % of challenges and intercepts were won versus the opponent.
Secondly, I plot Possession Won from Defensive Actions. Bit of a mouthful but essentially consider all defensive actions as above and ask “what % resulted in Celtic winning / maintaining possession?”. Clearly, it is vitally important when defenders commit to defensive action they win possession otherwise it becomes a lottery in terms of the ball in your own defensive third.
We are looking at defensive actions only, not passing or attacking actions.
Over the years this has proved an excellent match up to the “eye” test. That is, does the data match what we think we see with our own eyes. So, for example, a young Van Dijk is always top right (“good”) and Efe Ambrose is always bottom left (not so “good”).

The current cadre of centre backs are highlighted and the “average” of all is in Green.
Carter-Vickers has established himself within the upper echelons of recent Celtic centre backs so far this season. Dominant in both aerial and ground challenges, he has the highest rate yet of securing possession from his defensive actions (95%).
Given context: he is young (24); he is on loan; he has been adapting to a new system; he has had a cavalcade of defensive partners around him, this is a highly impressive outcome.
Jullien’s data relates to previous seasons as he continues his recovery from long term injury. If he can get back to that level, and that is a big “IF”, then the partnership with Carter-Vickers is a tantalising one.
On the other hand, Starfelt is in the sad left bottom corner with Hendry and Ambrose. Not a happy place.
His DASR% is the lowest of all centre backs I have recorded (69%).
The main issue for Starfelt is aerial challenges. Whereas Carter-Vickers win / loss on those is 108/16, Starfelt’s record is 98/42.
There are a number of weaknesses to observe.
He simply isn’t that tall. Wikipedia has him at 6’1 but I’d not be surprised if that is an inch or two generous.
He doesn’t jump well/high.
His positioning is often out – he is often caught under the ball and doing what some defenders often do, overreaching to win challenges he cannot (AKA Doing a Duffy).
On Welsh what I’d point out is that he is improving. His numbers are better than last season, and as a very young defender, to already be near the “average” band is decent development. He too suffers from the “a bit small” despite the Celtic official site laughably having him as 6’3!
Defensive Errors
Most goals are due to a number of small things going wrong, but occasionally there are glaring individual errors.
This is a subjective count, but it is a consistently recorded metric (by me!) and, again, since 14/15, it fits well with the “eye” test.

Duffy’s calamitous and short Celtic career was not just about his individual contribution. His personal error rate was largely down to his nature of trying to plug gaps and defend situations outside his direct control – overreach.
He suffered from having a dysfunctional alignment of defenders around him; lack of consistent partner; an aging midfield defensive shield; and two full backs bombing on.
So, context is important.
Many of the same structural weaknesses can be observed this season. Two full backs learning new inverting patterns; a never-ending change in personnel; a new repurposed midfield shield; a very high line; an ultra-attacking philosophy prone to counter attack risk.
Given that context we have Carter-Vickers being the least personally error prone centre back I have recorded since 14/15, and Starfelt, whilst not being in the Ambrose / Duffy territory, is way above what I consider an acceptable benchmark of 0.5 per 90m.
Note that both Jullien and Welsh are on the “right” side of this average. Jullien’s errors tend to come against lower ranked opposition.
Ball Progression
Finally, I consider ball progression.
The ability to get the ball forward through passing or running with the ball.
The data for this has been from 19/20 on a consistent basis so a smaller sample.

Ajer is probably elite at European level, if not World level, for progressing the ball whilst running with it. No one comes near him in terms of Ball Carries.
Higher volume, is Pack Passes – these are forward passes that take at least one opponent out the game.
Here, Carter-Vickers is well ahead of the field in completing over 13 Pack Passes per 90m. The fact Starfelt and Welsh are not far behind suggests that team style and manager instruction might be an important factor. If players are told to put a lot of verticality into their passing, as I am sure Postecoglou would encourage, then that might say more than individual ability.
But, of the current cadre, Carter-Vickers has adapted to the required style better than most.
The number of Pack Passes is only part of the story.
The Packing Score gives an indication of how many opponents have been taken out and therefore the risk and impact of the passes as opposed to a simple count.

The Packing Score for Carter-Vickers is well above all other peers. His forward passing is not just high in volume but takes out more opponents. A reoccurring example is what has become a trademark right sided pass to the inside right half space to Rogic – a pass completed 54 times taking out 194 opponents.
Again, the fact Welsh and Starfelt are above 40 Pack Score per 90m suggests this is as much about style of play as individual ability.
The Verdict
Carter-Vickers
The late loanee, whom some of us shrugged at, has become mission critical. He is the dominant defender at Celtic.
He is already posting numbers commensurate with the best centre halves Celtic have had in the last 7 seasons.
Not only is he defensively excellent, he is relatively error free, and, crucially, excels in ball progression from deep.
Given his age profile (24) and the improvement left in him, not to mention the benefits he will gain If he gets a secure partner make this a MUST BUY.
I understood the initially agreed fee with Tottenham Hotspur was in the region of £2-3m but there have been press rumblings it is more than that.
Let’s hope his permanent signing is secured soonest.
Starfelt
It appears Starfelt has bene brought in as a mobile defender able to play in a high line and be safe with his passing.
What Postecoglou has perhaps underestimated is that in a league where most clubs average 75+ long balls per game, aerial security is a must for most defenders.
Starfelt struggles defending high balls, but also seems to struggle against good forward movement as his positioning is sub optimal and he gets into some very odd body shapes. Also, he is very one footed for a left sided centre back.
Although his passing is super secure, his proneness to error is way beyond the threshold I am comfortable with. His performance in the League Cup Final was calamitous. Many tend to forget this as Celtic won. But that performance was up there with Blackett at home to Molde, Toure at home to Borussia Moenchengladbach and Bitton in Astana as the most personally awful I can remember from a centre back. A career limiting performance.
On the latest Huddle Breakdown podcast (please subscribe so you never miss an episode) my colleague @jucojames called time on Starfelt’s Celtic career.
The current evidence supports that.
And consider this: Starfelt has had a safe and secure partner in Carter-Vickers; behind a competent goalkeeper (Hart); and in front of a coherently set up midfield. Yes, he is new, but he is not a rookie (26 with 4 Swedish caps).
Celtic need another senior centre back even assuming Jullien will return and be back to somewhere near his previous form.
Welsh
Welsh will shortly turn 22 and is learning his trade. Last season was his breakthrough and this has seen him pitched into many difficult situations.
Despite a lack of real height again limiting his aerial duel effectiveness, Welsh’s defensive action numbers have improved last season over this. That is hugely encouraging.
His passing is particularly encouraging and an area of his game he has massively improved upon since breaking into the team as he has stated in interviews.
At 22 he looks like he will be at least at the “average” level of acceptable Celtic centre halves with an upward trajectory and real signs he is able to play to Postecoglou’s instructions.
I’d be happy to have Welsh as a back up option for the senior centre backs.