Unpacking Armstrong, and Avoiding Cognitive Dissonance
The Stuart Armstrong “will be / won’t he” sign a new contract saga rumbles on. Rumours of interest from English Premier League sides persist, and why not? After scoring 17 goals last season in winning the treble, the 25-year-old is maturing into the complete central midfielder, and carving an international career.
I have no insight into the thinking of Armstrong, nor what is happening behind the scenes between player, agent(s), club and suitors. But it is a tale we’ve seen on Planet Football many times. Player enters final year of contract, knows he can leave for free in a year. Interested parties smell a bargain. Player seeks maximum recompense. And good luck to him. It is a short career and he is a blossoming talent whose next contract is potentially the most lucrative of his career. Assuming a 3-4 year deal he will be 28 or 29 when the next renewal comes up. He owes it to himself and his family to maximise his earnings and career aspirations in his next move. I am not sure all that needs saying, but football fans being what they are, the murmurs of treachery and “good riddance” are, sadly, inevitable.
He may leave, and we may be disappointed (I will, most definitely). But let’s be honest about the player he is and the value he provides to the team. The “good riddance” comments along with the “he’s easily replaced” remarks reflect the cognitive dissonance of the utterer. Don’t fall into that mentally torturous trap.
The Sacrificial Lamb
When I wrote on the renaissance of Armstrong last season (splendid read, highly recommended, and one click away!), I concluded he needed to perform consistently at the coal face to rival Brown as a first pick central midfielder. Also, that he tended to give the ball away. This was true, based on the data I collected. What was clear was that his performances as a central midfielder far outstripped those when wide left. Armstrong sacrificed himself for the team and did his best in a role where he was less effective than his preferred position for 18 months, under Deila. And we should perhaps remember that when criticising any decision he makes to move – there were no international call ups in his first 18 months at Celtic.
And so, based on the data captured for Armstrong, there were more attacking actions, but also a relatively low pass rate for a central midfielder. What I didn’t realise then is I lacked the language and methodology to describe a player like Armstrong.
The Pack Man
In the initial match summary of the season, the wondrous Celtic Play The Blues (you cannot miss it!), I introduced the use of a concept called “Packing”.
Let me copy the explanation here (edited – the videos in the link are excellent primers):
“Time to introduce “Packing” and “Impect”. I have no idea the genesis behind the use of those terms but they are explained very well at Bundesliga.com. My summary:
Packing A completed pass that, on receipt and control, has taken 1 or more opposition players out of the game. That is, they are now behind the ball relative to their own goal. Pack Rate The total number of players taken out of the play by a completed “pack”. Impect A value given to the players taken out of the play:
3pts for every defender (taking defenders out the game is really good!)
2pts for each midfielder (taking midfielders out the game exposes the defence)
1pt for a forward (Sviatchenko and Simunovic can pass around the lone striker all day)
The same principle applies to dribbling with the ball, except there is only the dribbler to credit as opposed to having a passer and receiver for a pass. All other passes that do not result in an opposition player being taken out the game get rewarded with – a pass completion of 1! So now all passes have a value AND we can see the relationships between players.”
I don’t believe there is, nor ever will be, a “magic bullet” to ascribe football performance. For balance, the use of packing has been questioned by the respected Statsbomb. In the field of football analytics, there are those seeking to sell services to professional clubs, and in that scenario, I would want to have robustly proven models too. For my use case, the keen amateur hoping to engage the Celtic support to explain performance, I believe “packing” is another very useful tool amongst many in the narrative forming tool box. So, I will persist.
If we consider the performance of the team regarding passing and creativity against Linfield, there are many measures available. Let’s consider pure passing statistics.
This highlights well the Griffiths phenomenon. Completing 13 passes in 69 minutes, only 9 from open play, yet leading the team in xA. That is, his 13 passes resulted in an expected goal return of 0.32 goals. He also created the assist for Rogics’ goal. Unsurprisingly, Rogic provided 5 key passes and Tierney 4. Armstrong created 2 chances, but is well down the xA table with a mere 0.06. This is despite completing 110 passes with 96% success rate. It is not surprising to see the two centre backs and the goalkeeper at the bottom of this analysis. An aside - the two full backs being near the top highlights the extent to which Linfield packed (sorry) the middle of the defence and midfield forcing Celtic wide.
This analysis focusses on the final pass, and does not consider what came before. Neither Armstrong nor Brown, despite 229 successful passes ay 96% accuracy fare well using this analysis.
Power Pack
Let’s now consider the Pack and Impect data.
Firstly, by the Swedish way, a shout out to Lustig. He does like a long speculative pass, no bad thing against packed defences. 29% of his passes took out at least one opponent.
But the Packing and Impect statistics were seemingly invented to showcase what Armstrong is all about. 43% of all his 110 successful passes (47 passes!) took out at least one opponent (he averaged 2.91 opponents taken out with each successful pack pass). No one else comes close.
I know this is only one game but (confirmation bias alert) I was convinced last season that Armstrong would be a star by this measure. Simply, he breaks the defensive lines. Yes, he can often give the ball away but that is the nature of risk/reward when attempting to beat opponents rather than play the safe option (let’s not call it the “Bitton pass”). Armstrong progresses the team up the park into areas closer to goal thus increasing the risk to the opposition of a shot or a chance being created. I will stick my neck out to say we’d see this pattern in most games he plays. Priceless.
A final statistical reveal is his connections – whom do his pack passes put in possession?
Armstrong connects effectively with Sinclair, arguably our most dangerous attacker (oh, Paddy!). He is also effective bringing Tierney and Rogic into the game, showing a left sided bias. It is interesting his relative lack of connection with Forrest given Forrest was the top pack receiver in the game. I can only attribute this to the fact Armstrong predominantly plays right to left as a right footer. His connections with Dembele also highlight a further limitation of our beloved Griffalo who was on the pitch for 69 minutes. But not for this article.
Conclusion
The Packing and Impect statistics powerfully illustrate the worth of Armstrong to the team. At 25, he is maturing nicely and we are all benefitting from the patience both parties have shown during his Celtic career. He would be a huge loss in my opinion. A 21-year-old French prospect is not a direct replacement just because he cost a lot of money. I passionately hope Armstrong stays. If he does not, he leaves with my best wishes and I suspect he will always be known as the Pack Man.
