Ajax at Groningen: Four Changes, a New Back Line and a 90-Minute Test

At 16:45, the team sheet appeared on the third monitor in my office in Barcelona. I am used to reading these sheets in a fixed order: goalkeeper, back line, midfield, attack, then the bench. This time, my hand stopped at the first line. Marc-André ter Stegen starts in goal for Ajax. Second line, Julian Brandt is deployed on the right wing. Last line of the bench, Yves Bissouma sits there, unused. Those three names, in any database I have built over more than fifty years, do not belong to Amsterdam.

In the summer of 2026, I saw the Opta ghost – and since then, my eyes no longer trust what they see. I learned a reflex: when a dataset contains two facts that cannot coexist, the error lies in the dataset, not in reality. Before saying a single word about Míchel Sánchez's tactics in the away trip to Groningen, I must talk about this team sheet itself.

But I cannot stand still either. The ball rolls at 16:45, and any analysis of a team sheet has a shelf life measured in minutes. So I do what a data writer at 68 still always does: I split the sheet into two layers. The first layer is tactical structure – what I can read, verify, and cross-reference against what Groningen will do. The second layer is source integrity – what I am obliged to hang at the top of the article, because ignoring it strips every downstream conclusion of its validity.

Context: three days after Hajduk, and a back line taken apart

Ajax arrive in Groningen in what I call "the second week of a reaction cycle". Three days earlier they lost 1-2 to Hajduk Split. Shortly before that, they were held 1-1 by NEC – a match that Dutch analysis describes as the kind of dropped points far more costly than a defeat. Those two results combined create pressure not on the season, but on individual positions. And a coach's reaction in that situation is usually not a change of system. It is a change of people.

Four positions changed compared with the Hajduk match. Ter Stegen returns in goal. Rosa comes in at right-back. Bouwman partners Janse at centre-back. Wijndal plays left-back. Kehrer drops out of the starting eleven to the bench. That is a back line dismantled and reassembled in a different order, with two of the four men being new elements within the same unit.

I spent three weeks in 2026 building a homemade xG model, only because a colleague in Barcelona mocked me for looking at a data table without watching the match. The lesson from that summer was not "trust the model". The lesson was: a model only means something when you know how many matches it was built on. A back four that has not played together long enough is a variable without a distribution. You cannot compute the variance of something that has never happened.

On the other side, Groningen are not a team for experiments. Van Bergen, Ondrejka and Willumsson form a trio sharing one trait: they prefer the space behind a defence to the space in front of it. They do not need possession to inflict damage. They need one straight pass and three seconds.

That is the context. And that is why I argue this team sheet is not a simple rotation.

Layer one: the new back line and the transition problem

Start where the risk concentrates. The Bouwman–Janse centre-back pairing is the biggest blind spot on this sheet, and not because they are poor. It is because of structure.

When a coach changes both centre-backs at once, he is not changing two individuals. He is changing the entire information system of the defence: who covers whom, who steps up when, who takes responsibility for the ball behind the full-back, who calls the offside line. None of that lives in a tactics book. It lives in the muscle memory of a pair that has played a few hundred minutes together. Bouwman and Janse, as far as this sheet shows, do not have that memory.

Rosa at right-back and Wijndal at left-back are both full-backs with a tendency to push high. In a possession structure that is a necessary condition: you need numbers in the upper third to pin the opponent back. But a necessary condition is not a sufficient one. When both full-backs push high, the space behind them becomes the territory of that new centre-back pair – and it is territory they have not been trained to defend together.

I do not have Groningen's PPDA for this season. I do not have xG either. And I will not invent them. But I have a principle that has held for me for fifty years: when a team changes two centre-backs and pushes both full-backs high, that team is betting it will score before it is punished. That is a wager that can be measured, not a belief.

Kehrer's demotion to the bench is the most notable and the most ambiguous detail. The sheet gives no reason. It could be form. It could be fatigue after Hajduk. It could be a specific tactical calculation: Kehrer is the type of centre-back who likes to step up and engage early, while Groningen play long balls behind the defence. If Míchel wants a pair that sits deeper, dropping Kehrer is logical. If he wants rotation because of the schedule, that is logical too. Two different explanations lead to two opposite conclusions, and I have no data to choose. I record both and let the match adjudicate.

Midfield: Amrabat, Klaassen, Mokio and a question about the line-breaking pass

The three names in midfield tell a far clearer story than the defence. Amrabat, Klaassen and Mokio are a trio designed for two jobs: controlling tempo and protecting the space behind when both full-backs push on.

Amrabat is the anchor. His role is not to create chances but to guarantee that when Ajax lose the ball in the final third, someone is standing in the right place to stop the first counter. In a match where the defence has already been shuffled, the existence of an anchor is a survival condition. That is why I argue Amrabat starting is not an attacking choice. It is an insurance choice.

Klaassen is the runner into the box. He is the midfielder whose value lies not in passes completed but in arrivals inside the area at the right moment. In a match where Ajax may have to attack a packed block, a midfielder who knows where to stand inside the box is worth more than one who knows how to pass.

Mokio is the young variable. And this is where I want to pause, because it concerns how I read a team sheet.

There is a pattern I have seen repeat many times over more than a decade, across every league: when a big club hits a short-term results crisis, the coach tends to hand a young player a start. The press calls it "giving him a chance". I call it "transferring risk". A young player does not have enough minutes to be judged fairly, so he becomes the person responsible for a structure he does not control. If Ajax play well, it is the coach's courage. If Ajax play badly, it is the youngster's immaturity.

I am not saying Míchel is doing that. I am saying the sheet is insufficient to distinguish the two possibilities. And a data writer has to say so.

What stands out is that this midfield lacks one specific profile: the line-breaking passer. Amrabat passes sideways and backwards. Klaassen runs. Mokio has not yet proven he can unlock a low block. If Groningen sit deep, Ajax will need a pass threaded between the opposing centre-backs – and I do not see a specialist for that pass in this trio.

That is a structural gap, and it is not any player's fault. It is the fault of a sheet assembled to protect a defence before that defence has had time to form.

Attack: Brandt right, Gloukh left, Leonardo central

The front three is the brightest part of the sheet, and the part that convinces me Míchel prepared for a specific match rather than a generic one.

Julian Brandt on the right. This is a decision worth analysing, because Brandt is a left-footed player operating on the right – meaning he tends to drift inside and create shooting situations on his stronger foot. But when deployed on the right in a structure with high full-backs, Brandt has a second option: wait for the full-back to overlap and receive in the inside channel, where he can turn and face the centre-back.

That is how you isolate a specific full-back. And if Groningen have a weaker full-back on that side, this is an explicit calculation.

Gloukh on the left is the 1v1 weapon. He is the type of player who can beat a man in tight space, which is what Ajax need when Groningen stand as a block. And Marcos Leonardo centrally is the receiver of cut-backs from both flanks – the kind of number nine who lives on passes nobody records, because they only become assists after he has already shot.

A beautiful number is like a perfect pass: it needs no explanation, only to be seen.

This structure tells me three things. First, Ajax will attack mainly through the two flanks, not through the middle. Second, they will need full-backs to join in to create 2v1 situations – which raises transition risk. Third, if Groningen block both flanks, Ajax will have to switch to plan B, and I do not yet see plan B on this sheet.

The bench holds plan B: Bissouma, Baas and Arokodare. Bissouma offers added midfield intensity. Baas offers a replacement centre-back. Arokodare offers a target man in case Ajax must go long. That is a bench built on a contingency logic, not a rotation logic. A rotating team puts youngsters on the bench. A contingency team puts game-changers on the bench.

Layer two: source integrity and three names that do not belong to Amsterdam

This is the part I have to write, even though it is uncomfortable.

The report I am analysing describes Míchel Sánchez as Ajax's coach and lists Marc-André ter Stegen, Julian Brandt and Yves Bissouma as Ajax personnel. In the database I have built and cross-checked over many years, those three names are tied to other clubs. Míchel Sánchez is tied to another club in Spain.

There are three possibilities, and I rank them by probability.

First: this is a hypothetical scenario, an alternate dataset, or a simulation. In that case the entire tactical analysis above holds within that frame, and I have no right to impose outside reality upon it.

Second: there is an editorial error, where player and coach names were mixed from different sources. This is the most common error in modern sports journalism, and it often goes undetected because nobody cross-checks a team sheet before kick-off.

Third: this is real information and I am missing data. Low probability, but I do not rule it out, because I have been wrong many times in my life and I remember every one.

I am 68, but data is younger than I have ever been – every season it grows another layer of teeth.

What I want to tell the reader is not "doubt this article". What I want to say is: pay attention to the structure of doubt. A team sheet has two kinds of error. The first is factual – who plays, who does not. The second is intentional – why the choices were made. The first can be verified by eye. The second never can, even with all of Opta's data.

And this is what I consider most important in this entire piece: most of what is written about team sheets is inference about intent, presented as though it were fact. When someone says "Míchel is rotating because Ajax have a European fixture coming", they are telling a story, not reporting a datum. The story may be true. But it has not been proven, and it never will be, because coaches do not publish their reasons.

When the stadiums fell silent in 2026, I understood: football never died, it simply took off its clothes and revealed its skeleton. And the skeleton of a team sheet is the numbers on minutes played together, not stories about motivation.

The contrarian angle: correlation is not causation, and a team sheet is not a tactic

This is where I want to break a reading habit.

There is a pattern in sports journalism I have seen for five decades: once the lineup is published, analysts immediately build a causal story. The coach changed four positions, therefore he is sending a message. Player X was dropped, therefore he has lost trust. Player Y starts, therefore freshness is the priority.

Every one of those sentences is a causal claim inferred from a single correlation: the presence or absence of a name on a piece of paper.

A single correlation is not enough to establish causation. It is not even enough to establish a trend. To say Kehrer was dropped for form, you need data on Kehrer's form across many matches, compared with data on Bouwman and Janse, plus a model of how the coach makes decisions. You have none of the three. You have one piece of paper.

This leads to a counterintuitive conclusion: a team sheet is data about people, not data about tactics. It tells you who will be on the pitch. It does not tell you where they will stand, which way they will run, or who will cover for whom. Those things only appear once the ball rolls, and they often differ sharply from what the paper suggests.

I have seen this many times. A lineup published with four defenders can operate as a back three in possession and a back five out of it. A midfielder listed as "left wing" can play as a free attacking midfielder. The paper is not wrong. It is simply incomplete.

There is a second contrarian layer, and it concerns this club specifically. When a big team changes four positions after two poor results, the media usually reads it as panic. I read it the other way, at least here: four changes is a sign of deliberate preparation. A panicking coach changes one or two positions to "wake the team up". A coach with a plan changes an entire unit.

That does not mean the plan is right. It only means it is a plan, and it can be judged.

The third contrarian layer, and the one I want readers to carry away: Ajax's biggest risk in this match is not the new defence. It is the transition window between attack and defence. A new back line that is misjudged rarely collapses through lack of skill. It collapses because the ball is lost in the wrong place, at the wrong moment, when the midfield has pushed up and both full-backs are already high. That is a systemic error, and it will not appear in any statistical table until it has already happened.

What to watch: the first fifteen minutes and a signal for the next round

I am not predicting the result. I do not have enough data, and I stopped predicting after the summer of 2026, when I was right about the champion but a Spanish editor told me nobody read the way I wrote.

What I offer is a set of signals.

The first fifteen minutes will answer the biggest question: whether Bouwman and Janse hold the correct distance from each other. If that distance swings too wide when Ajax have the ball, Groningen will find the gap within the next ten minutes. If they hold a stable distance with one always sitting deeper than the other, the structure is working.

The second signal is Amrabat's position when Ajax lose the ball in the final third. If he is there within three seconds, the anchor is working. If he is absent, the rest of the defence will carry responsibility for a gap they were never trained to fill.

The third signal is how often Brandt receives in the inside channel, facing a centre-back rather than a full-back. If that number is high, the isolation plan is succeeding. If it is zero, Groningen read the intent before kick-off.

And the fourth signal, the one I consider most important for the next round: whether Ajax keep the same back line in the following match. If Míchel changes centre-backs again, then today's sheet is not a step in a building process. It is an experiment, and experiments do not produce data – they only produce new hypotheses.

I will be in front of the screen at 16:45 with three windows open: the match, the live data feed, and a blank window for what I cannot predict. The third window is always the most important one. It reminds me that every model is a confession about what its author does not know.

And if Ajax win 3-0 tonight with a back line that has never played together, I will not say I was right. I will reopen the team sheet and look for where I misread it. That is the only way a data writer at 68 still has a reason to write tomorrow.