The Scottish Premiership Team of the Year (so far) using a data-driven approach
A look at some of the outstanding players as we approach the final third of the league season
This article was written for Scottishfootball.info by Data Bhoys, a Celtic fan with a keen interest in data analysis. They have started their own Substack account you can follow, as well as the Data Bhoys account on X.com.
This article is a breakdown of the Premiership team of the year so far, with two players given for each position to give an overall first and second XI .
So, we are just about two thirds of the way through the Scottish Premiership. The January transfer window has just closed, with teams trying to strengthen in a bid to win the title, qualify for Europe, or avoid relegation.
It has been a weird season in Scottish football. When is the last time a team not named Celtic or Rangers were top in February? When’s the last time Celtic were 3rd in February?? Motherwell playing liquid football??? Steven Pressley as Dundee manager choosing tactics based on a dream????
The overall quality of the league feels miles better this season than it has for a long time. It could be argued that this is largely due to Celtic and Rangers massively under-performing at different parts of the season, but regardless I think it has undoubtedly made for a more exciting league season overall.
Today, we will try to assess what the ‘Team of the Year’ might look like so far – but crucially, doing so through a purely statistical lens in an attempt to remove narrative, bias, and hyperbole from the equation.
Naturally, sometimes playing for Celtic or Rangers can inflate certain statistics through sheer volume of possession and dynamics of the games. However, I have tried my best to create a framework which adequately reflects ability regardless of whatever team a player plays for.
We would be here all day if I tried to go through every single metric for every single player - so for the sake of time and clarity, for each position I have created indexes which group certain metrics together based on which skill they best represent.
The metrics are percentile ranked and then weighted holistically in each index to try and give a comprehensive assessment of that particular skill. An overall score for the position is then created through a weighted composite of each individual index.
To run you through an example, I’ll go through how goalkeepers are assessed. I measured goalkeepers through 4 indexes: Shot-Stopping, Command of Area, Build-up Involvement, and Passing.
Shot Stopping is comprised of Goals Prevented per 90 and Save Rate %. Goals Prevented is worth 70% of the Shot Stopping score, and Save Rate % is worth 30%. It’s the same methodology for the other indexes.
The overall score for Goalkeepers is calculated as 50% Shot-Stopping, 25% Command of Area, 20% Passing, and 5% Build-Up Involvement.
The overall score and the scores in each index are then percentile ranked for each player to show who is excelling compared to their peers.
If you are unfamiliar, the way percentile rankings work is that if a player is in the 94th percentile, that means they are ranked higher than 94% of the dataset. It gets a bit redundant if there is only say, 13 players in the dataset as someone in the 75th percentile is essentially the 4th best player – just keep in mind the sample size for each position in this regard.
I will attach a cover image for each position that shows the exact weighting and metrics used - and if you are unsure of any metrics, this link here will take you to the Wyscout glossary where you can find definitions for all the different metrics used.
And all the data used in this exercise was collected on the 7th of February so, to be more precise this is the Team of the Year up to Matchday 25.
I’ve waffled enough so, let’s get into it!
GOALKEEPER
Sample size – 13 Goalkeepers with a minimum of 900 minutes played
Top ranked Player – Calum Ward (Motherwell)
Calum Ward is the 2nd highest ranked Shot-stopper in Scotland this season, and is by far the highest ranked in Build-up Involvement while holding above average rankings in Command of Area and Passing.
Ward is a solid keeper across all areas, and deserves his flowers for being an excellent shot-stopper who is playing his part in the football that Motherwell have entertained us with this season. The only keeper who comes close to the build-up involvement of Ward is Kasper Schmeichel, and even at that Ward is still averaging nine more received passes and eight more short/medium passes per 90 than the Dane.
The runners up award for this position goes to Raphael Sallinger of Hibernian. 75th percentile in both Shot-Stopping and Command of Area, he is just a little bit below average when it comes to footwork ranking in the 42nd percentile for both Build-up Involvement and Passing. The latter two are probably more influenced by tactical direction, to be kind to Sallinger.
RIGHT BACK
Sample size – 15 RB’s with a minimum of 900 minutes played
Top ranked player – James Tavernier (Rangers)
James Tavernier is perhaps approaching the end of his career now at 34 years of age, but he is still a very effective creative and attacking force and was the highest scoring right-back. He was the highest ranked in Progressive Passing and Final Third Impact while also giving admirable showings for Possession & Press Resistance and Aerial Defending ranking in the 86th and 71st percentile respectively.
The runners up spot goes to Keelan Adams of Falkirk. Adams is an outstanding progressor of the ball through carries and ranked very highly in Ground Defending and Defensive Screening.
LEFT BACK
Sample size – 14 LB’s with a minimum of 900 minutes played
Top Ranked Player – Kieran Tierney (Celtic)
Kieran Tierney has perhaps split opinion among the Celtic support at times this season, but he is still an immensely effective player.
Tierney was the highest ranked left-back in Defensive Screening, Progressive Passing, Possession Efficiency, and Final Third Impact. He also ranked in the 69th percentile for Aerial Defending, and the 77th percentile for Progressive Carrying.
He was down in the 15th percentile for Progressive Intent (likely an effect of Celtic’s playing style) and was a little bit below average in Ground Defending at the 46th percentile
The runners up spot will go to Harry Milne of Hearts but it was a close battle between Milne, Filip Lissah and Cristian Montano. Milne is very good in duels, coming in the 92nd percentile for Ground Defending and 100th percentile for Aerial Defending – he leaves a little to be desired when moving up the pitch but did post a very good 85th percentile ranking in Progressive Passing.
CENTRE BACKS
Sample size – 35 CB’s with a minimum of 900 minutes played
Top Ranked Players – Liam Scales (Celtic) and Miguel Freckleton (St. Mirren)
Liam Scales divides opinion among the Celtic support, but he has been one of Celtic’s most consistent performers during this turbulent season.
Scales was the highest ranked defender for Possession Efficiency, and also posted a 94th percentile ranking in both Aerial Defending and Progressive Passing.
Ground Defending was the big weakness for Scales only ranking in the 29th percentile, and Progressive Intent was also down at the 32nd percentile – similar reasons to Tierney I would hypothesise.
Miguel Freckleton may sound like an odd choice on paper, given how St Mirren have struggled in this league season.
Freckleton is crazy good when it comes to progressing the ball. He was the highest ranked player for both Progressive Carrying and Progressive Intent, and was in the 97th percentile for Progressive Passing. In defence he was in the 85th and 82nd percentile for Ground Defending and Defensive Screening while being slap bang in the middle for Aerial Defending.
He was rumoured to be the subject of transfer interest during the January window and it is not a surprise to me. He was by far the highest scoring centre-back in the overall score index.
We’ll select two runners up in this position to fit in our second XI, and it will go to Robbie Deas and Mats Knoester.
Deas was a huge surprise to me, especially after I searched his name on Twitter and read what Kilmarnock fans think of him… but he was in the 94th percentile for Ground Defending and the 82nd percentile for Progressive Passing. Go figure.
Mats Knoester was in the 91st percentile for Ground Defending and the 79th percentile for Aerial Defending, while excelling in Progressive Carrying at the 97th percentile.
CENTRE MIDFIELDERS
Sample Size – 43 CM’s with a minimum of 900 minutes played
Top Ranked Players – Nicolas Raskin (Rangers) and Elijah Just (Motherwell)
Nicolas Raskin was by quite some distance the highest ranked central midfielder in this exercise. Midfielders come in all different forms, – some attack-minded, some defensive minded – but here I was trying to assess midfielders based on a balanced view of both abilities.
Raskin was the highest ranked central midfielder in Progressive Passing and Possession Efficiency, while being in the 90th percentile for Progressive Carrying. He was a little bit below average in the 45th percentile for Ground Defending, but ranked above average in the other defensive indexes.
The other central midfield spot will go to someone who is wholly attack minded and that is Elijah Just of Motherwell.
Just did not rank highly in any of the defensive indexes, but he was the highest ranked player in Progressive Carrying, in the 95th percentile for Goal Threat and the 90th percentile for Final Third Creativity – while being in the 71st percentile for Progressive Passing, Progressive Intent, and Possession Efficiency.
For our two runners up, we’ll go with Elliot Watt and Brad Spencer. The model quite liked Luke McCowan and Connor Barron, but given the two are on the lower end of minutes played and in McCowan’s case being split at RW and LWB, I decided to go with Watt and Spencer.
Watt was in the 98th percentile for both Defensive Screening and Progressive Passing, while ranking in the 88th percentile for Possession Efficiency.
Brad Spencer was really strong in the defensive indexes, 88th percentile for Ground Defending, 83rd for Aerial Defending, and 95th for Defensive Screening.
WINGERS
Sample size - 17 Wingers with a minimum of 900 minutes played
Top Ranked Players – Alexandros Kyziridis (Hearts) and Tony Yogane (Dundee)
The main issue with selecting wingers was that when you split it into right and left wingers, the dataset is quite small for each – so I compiled every winger, left or right into one bigger dataset and the highest ranked wingers take the spots.
Alexandros Kyziridis was the highest ranked winger for Progressive Passing and Final Third Creation, and was in the 88th percentile for Progressive Carrying, Progressive Intent, and Possession Efficiency. All that is missing from his game is Goal Threat.
Tony Yogane is obviously playing in a weaker side at Dundee but he is having a really impressive season. He was the highest ranked winger for both Progressive Carrying and Progressive Intent, while ranking in the 94th percentile for Final Third Creation. Like Kyziridis he is not much of a goal threat. We’re playing two strikers here so that’s where our goals will come from hopefully…
I can already hear you saying “What about Tawanda Maswanhise?”.
Maswanhise is obviously having a ridiculously good season in front of goal and he is the highest ranked winger for Goal Threat. But in almost all other categories, he is ranked in the bottom three. He obviously has insane value in front of goal this season, but he does not look like he is an effective player when asked to progress the ball or create for his teammates.
So - controversy out of the way, our two runners up here will be Hyun-Jun Yang and Calvin Miller.
Yang was the highest ranked winger in Possession Efficiency, while being in the 94th percentile for Goal Threat and the 81st percentile for Progressive Passing.
Miller was reasonably well rounded, with the highlights being a 94th percentile ranking in Progressive Intent and an 88th percentile ranking in Final Third Creation.
CENTRE FORWARDS
Sample size – 22 Centre-Forwards with a minimum of 900 minutes played
Top Ranked Players – Claudio Braga (Hearts) and Tyreece John-Jules (Kilmarnock)
Claudio Braga was the clear frontrunner of the centre-forwards.
The highest ranked striker in Goal Threat and Progressive Carrying, while being in the 90th percentile for Box Presence and the 86th percentile for Chance Creation. There were zero standout weaknesses in his game.
Tyreece John-Jules was the big surprise here and at time of writing he has just scored a hat-trick against St Mirren, so maybe I am onto something here.
John-Jules was the highest ranked in Box Presence and Link-Up Play while being in the 90th percentile for Goal Threat – and like Braga, there were no big red flag weaknesses.
Our runners up are Daizen Maeda and the recently departed Kieron Bowie.
Maeda was the highest ranked forward in Chance Creation (likely influenced through his minutes out wide) and in the 95th percentile for Goal Threat.
Bowie was in the 86th percentile for both Goal Threat and Box Presence while being in the 90th percentile for Chance Creation.
THE XI’s
So without further ado and with the help of Fotmob’s line-up builder, here is our Team of the Season (so far) first XI:
And the second XI.
A reminder that this exercise was done purely through a statistical lens – there are of course multiple different factors that go into player assessment, and everything is up for debate! You may disagree with the way I have weighted certain skills for certain positions - there are many analysts out there who will assess players in a different way than I have done so here. All methods are equally valid so long as they try to be fair and sensible with the use of data. It’s all about the interpretation of it, and I have tried my best to make sure this was an honest, robust assessment.
Oh, and if you want to know who the highest scoring player was across all positions for a “Player of the Year (so far)” - it was Claudio Braga. Closely followed by Alexandros Kyzridis. That damn Jamestown.
I hope you found this interesting and thank you for your time reading!
Not a single Celtic midfielder there isn’t surprising but it is shocking.
This is a cracking article. I always enjoy this sort of thing because it provides an entertaining and interesting framework for discussion. I think it’s really good the way your analysis has identified the outstanding performances by unheralded players.
Btw, isn’t the scouting and acquisition of Claudio Braga by Hearts the wake-up call for Scottish football at the dawn of the age of data and analytics?
I’m sure everyone who reads it will have an opinion on the methodology. Fwiw, my tuppence worth is that it might be useful to split the midfielders into DMs and AMs. Firstly, they have different skill sets, and secondly, the resulting team selection would probably better represent the actual systems used by the clubs.