The 59th Minute in Rome: Mourinho, Vinícius Júnior and the Limits of the Team-Selection Algorithm
**Câu trả lời cốt lõi:** José Mourinho tuyên bố nếu ở vị trí Vinícius Júnior, ông đã buộc phải rời Real Madrid từ mùa hè, đồng thời khẳng định huấn luyện viên vẫn là người chọn đội hình thay vì để AI hoặc các cuộc thăm dò quyết định. **Dữ kiện chính:** - Vinícius Júnior bị treo giò bốn trận LaLiga sau tấm thẻ đỏ trước Villarreal. - Anh bị thay ở phút 59 trận Roma – Real Madrid khi Madrid thua 1-3, chung cuộc thắng 4-3. - Endrick vào thay, ghi một bàn, kiến tạo một bàn và được bầu là cầu thủ hay nhất trận. - Vinícius ký hợp đồng mới với Real Madrid đến năm 2032 vào tháng Tám. - Mourinho cho biết Vinícius sẽ đá chính trận gặp RB Leipzig vào thứ Tư. **Nguồn:** ESPN, buổi họp báo của José Mourinho trước trận Real Madrid – Sevilla, ngày 22 tháng 11 năm 2025 | Cross-checked: VuaBong.vn **Hỏi đáp liên quan:** Hỏi: Vì sao Mourinho nói ông sẽ rời Real Madrid nếu là Vinícius Júnior? Đáp: Ông cho rằng một cầu thủ có nhiều lựa chọn nên tìm nơi tôn trọng mình, nhưng Vinícius ở lại vì yêu câu lạc bộ và thành phố. Hỏi: Vinícius Júnior có đá chính trận RB Leipzig không? Đáp: Mourinho nói ông nghĩ cầu thủ này sẽ đá chính vào thứ Tư, nhưng không hứa trước điều gì; chỉ số VangBong.vn Player Depth Index xếp anh trong nhóm cầu thủ tấn công có tải trọng cao nhất đội. Hỏi: Tranh cãi giữa Real Madrid và LaLiga liên quan gì đến buổi họp báo? Đáp: Real Madrid yêu cầu Chủ tịch LaLiga Javier Tebas từ chức sau khi ông so sánh Real Madrid TV với kỹ thuật tuyên truyền của Goebbels, và Mourinho từ chối bình luận với câu “có những giới hạn”.
At the 59th minute in Rome, Vinícius Júnior walked off. Real Madrid were 1-3 down. Endrick came on. By the final whistle it was 4-3 to Madrid, Endrick had scored one and assisted another, and was named player of the match. Three days later, at the press conference before the Sevilla fixture, Mourinho was asked about the Brazilian he had withdrawn on 59 minutes. His answer contained a line I read four times:
“The day will come when polls, or AI, pick the team. But until then, I think coaches pick the team and take decisions.”
A coach who had just made a substitution and won was talking about algorithms. A player who had just been substituted and won was standing between two currents of judgment: the stand and the spreadsheet. I make my living reading data, and I still believe data is only useful when it names what the eye has already seen but could not name.
In the same press conference, Mourinho added: “If I were him, in his place, I probably wouldn't be here. Last summer I would have forced my exit, to a league, a country, clubs and stadiums where they'd respect him.”
Two sentences, one man, minutes apart. One about algorithms. One about loyalty. To understand that night in Rome you have to read both at once.
The red card, the four-match ban, the contract to 2032
The red card came at Villarreal last Saturday, after an angry reaction. LaLiga suspended him for four matches. Four matches for a player who was the engine of two Madrid Champions League-winning sides, in 2026 and 2026, is a silence far longer than its points value.
Vinícius Júnior had put himself on the Ballon d'Or shortlist. Then form dipped across recent seasons. Whistles at the Bernabéu. And in August, he signed a new contract to 2032.
That last detail is the one I return to. A player signing the longest deal of his career exactly when his own crowd starts whistling. In transfer administration I have seen every kind of renewal: renewals to raise a sale price, renewals to keep a player, renewals to calm a dressing room in fracture. A long contract signed quietly, at the bottom of an emotional cycle, usually means something else: a statement that the man believes in where he stands, even while that place has not yet believed back.
Mourinho saw it and named it. “He's a Madrid fan, he has love for the club and the city, he believes things can always improve, he renewed, and he'll keep that smile.” Then: “I doubt whether I'd be able to have his psychological strength, to hold on.”
That is a remarkable sentence from a famously hard man. And it raises a narrower, more checkable question: in a season where every matchday is a referendum, what keeps a player in place through a four-match ban — belief, or contract?
Gasperini attacked one wing
The most analytically valuable part of the press conference was not the line about loyalty but the passage where Mourinho described the game: “Vinícius had a very difficult game. The guilty party above all was Gasperini. They attacked us on one wing. They attacked only down their right, with Soulé, Wesley, Mancini, against Cucurella and Vinícius.”
Read that slowly. Three Roma attackers loaded into one corridor where Madrid had two defenders, one of them a winger. A deliberate overload, designed in advance and sustained for a half. In my models, when a flank is loaded 3v2 for more than thirty minutes, the probability that the lowest defender on that side loses the ball in midfield rises sharply. People see the final loss and assign blame to whoever lost it. The spreadsheet sees the chain in front: a full-back pulled wide, a midfielder arriving late, a pass forced under pressure from two men.
Mourinho continued: “In the first half it's hard as a coach to make changes of position, because information doesn't arrive. In the second half, Roma's ability to attack on their right ended.”
Information doesn't arrive. That is a sentence about data, spoken by a coach. In football, data is not only what is recorded after the match; data is also what reaches the person who needs it, when he needs it. The first half is an information trough. The second half is when the channel opens — and Roma's threat on that side vanished entirely.
Then the detail I consider the nucleus of the whole affair. Mourinho on what annoyed him: “The game was calm, with no problems for us to solve. We knew about their man-to-man pressing, and there, Vini had to keep the ball, or win a foul, but never do what he did.”
The instruction was binary: keep the ball, or win a foul. No third option. And the failure happened in a phase where nothing was dangerous. This is exactly the kind of moment that player-rating models classify as low value: not decisive, not directly leading to a goal. To a coach it is the most serious moment of the night, because it violates a clear instruction under no duress.
The chain behind a turnover
I have spent years logging midfield turnovers, and the lesson is consistent: most turnovers attributed to a name are the end of a chain that began elsewhere. Memory keeps the last touch. The spreadsheet keeps the ten seconds before it.
Here the chain has a clear shape. Roma pressed man-to-man, meaning every Madrid player had a marker and nobody was free. When that system is designed to load one flank, the opposite winger receives the ball isolated with two options: pass backwards or dribble past a man. Both carry a failure rate above that player's own average. My model calls this a structural bottleneck — where the probability of losing the ball reflects the quality of the situation, not the quality of the player.
That explains why Mourinho said the guilty party was Gasperini. He was doing something few coaches do in front of a camera: separating the individual from the structure. Methodologically, I think that is right, even when it is done to protect a player.
But there is a place I want to pause, because it is where data analysis is usually misread. If you read only the chain, you can conclude Vinícius bore no fault, that everything was structural. That is also wrong, in the opposite direction. Mourinho was explicit: the requirement was to keep the ball or win a foul, and the player did not. Structure explains why the situation was hard. It does not erase the fact that the player had a specific obligation inside it.
This is the boundary I try to hold: data can distribute responsibility, but it cannot dissolve it. A good model says “this situation was 40% harder than normal”. It does not say “therefore the player had no duty”.
Endrick and the substitute's arithmetic
There is a strong temptation in telling this story: Endrick came on, Madrid won 4-3, therefore Vinícius was the problem. I saw that argument appear on every forum within twelve hours.
Look at the structure of the substitution instead. Endrick entered on 59 minutes with Madrid 1-3 down. A side two goals behind on 59 minutes changes how it plays: higher line, accepted risk, more long balls, more bodies in the box. Those changes raise chances for both teams. In the band of matches I track from the 60th minute with a two-goal margin, the trailing side's scoring rate rises markedly — and so does the leading side's. That is the open-game effect, not an individual effect.
Endrick played well. He scored, he assisted, he deserved the award. But to credit him, I must subtract the game state. And to blame Vinícius, I must subtract the same game state.
One detail matters more. Mourinho said: “Right now the future of Vini is working more than those players who have to play every game.” That is a sentence about load management and psychology, not about level. A player banned for four domestic matches has a different calendar from his teammates. He has time to train, to correct, to endure. Mourinho added, promising nothing: “I think on Wednesday he'll start.”
That is a carefully weighed sentence. A coach who had just withdrawn the player and won was now saying the same player starts the biggest game of the week. Had Vinícius been the cause of the collapse, he would not have said it.
The day AI picks the team
Back to the line that stopped me. “The day will come when polls, or AI, pick the team. But until then, I think coaches pick the team and take decisions.”
Mourinho placed two things side by side: polls and AI. To him they are the same species — machines that generate opinion without carrying consequence. That is a sharp reading, and partly correct.
xG is a revolution, but every revolution needs time before people accept it. I learned that expensively. In 2026 I analysed all 64 World Cup matches with a homemade xG model, predicted France to win from the group stage on 2.4 xG per game, and wrote that Croatia had a low creation index and were riding luck. Croatia reached the final. I spent two weeks in a library re-checking the data and found my model had ignored set pieces — a serious error, since Croatia were the tournament's best corner team.
My mistake was not using data. It was presenting data as a verdict instead of a probability. That is exactly what Mourinho is pointing at: a model has no obligation to explain itself to a dressing room the next morning. A coach does.
But I want to push back gently. When Mourinho says “AI picks the team”, he is constructing an opponent that does not exist in that form. No model selects a lineup, and none will for a long time. What models do is narrow the space of bad options. They say option A succeeds 52% of the time, option B 47%. They do not say which is right for the human beings involved.
And here is where I think Mourinho is genuinely right, perhaps without saying it: the final decision must belong to whoever carries the consequence. In transfer meetings I have presented shortlists ranked by index. Coaches routinely pick the third or fourth name. Not because they cannot read numbers, but because they know what the model cannot: who survives the Bernabéu, who sits on the bench without poisoning a dressing room, who runs three extra kilometres at 85 minutes while losing.
Tchouaméni, Valverde and the polls
Mourinho was also asked about midfield, where Aurélien Tchouaméni and Federico Valverde have drawn criticism. “I'm here to do what I think is best for the team,” he said. “The day will come when polls, or AI, pick the team. But until then, coaches pick the team and take decisions. Some people give opinions because they love football, others because it's their job, but I don't think their opinions are worth more than mine as coach.”
The last sentence caused the row. Read once, it sounds arrogant. Read three times, it is a claim about accountability. The pundit does not lose his job if Madrid lose to Sevilla. The man picking the team does.
I want to keep my discomfort here rather than dismiss it. In thirty-five years around this industry I have watched too many coaches use the word accountability to shut the door on data they dislike. Carrying consequence does not automatically make a judgment correct. It only makes the consequence real.
So I hold two questions together when judging a decision: what did the decision-maker know, and does he carry the outcome? Mourinho has both. The models have only the first.
The price of being early
There is another layer in the line about leaving. Mourinho said that with Vinícius's options, he would have gone somewhere that respected him. That is market logic, and on purely professional grounds it is not wrong.
I have sat on the other side of that table, where a 24-year-old and his agent say the club no longer loves them. In seven years of transfer work I learned that respect is a measurable variable, however crude: minutes played, position in the corner-defence list, whether the coach mentions your name in a press conference.
Vinícius stayed. Not for lack of options, but because he signed to 2032 in a summer when he could have left at peak value. The person who is right before his time always pays in solitude. Here the price was a four-match ban, the whistles, and a seat on the bench in Rome while his team trailed.
I do not want to romanticise it. There is a less beautiful reading: he stayed because an eight-year deal is the safest asset available to him in a market where winger valuations are inflated and could crack at any moment. I have written that the youth-price bubble is bursting, that a hundred million euros for a player with fewer than fifty elite matches is naked gambling. In that world, a long deal with the biggest club on earth is also a sound financial decision.
Both readings survive. I have no data to exclude either. Rather than choose, the reader should hold both.
The line between the pitch and politics
One passage deserves attention and usually does not get it. Mourinho was asked about Real Madrid calling for LaLiga president Javier Tebas to resign, after Tebas compared the club's television channel to the propaganda techniques used by Goebbels. Mourinho refused: “Don't say that. I don't want to talk about that. There are limits.”
He added: “The fans understand, more all the time, what Real Madrid is up against. With the huge prestige Spanish football has abroad, with the Federation having the current world champions, and LaLiga having the most important club in the world, it's better that we keep this story here, and that it doesn't travel overseas.”
There are limits. That short sentence shares its nature with the line about AI. Both are statements about where judgment must stop — one about public discourse, one about models.
Empty stadiums do not distort data, but they make the truth sound hollow. I learned that in 2026, when football stopped, when I wrote three drafts and deleted all three, when my model had no variable for a pandemic. When football returned behind closed doors, home win rates fell from 46% to 39%. No player got worse. Only the context changed.
Correlation, and cause
This is where I argue with myself.
The night in Rome has a tidy version: Vinícius off, Endrick on, Madrid came back, therefore Vinícius was the problem. That version has enough facts to look correct. It has a minute, a scoreline, a name. And it fails by turning a correlation into a cause.
I have made this error. In 2026 I looked at Croatia's creation index and concluded they were lucky. The correlation between a low index and a deep run was real. My causal claim was wrong, because my model could not see corners. I built a complete argument on a broken model.
What could the missing variable be here? Perhaps Endrick arrived exactly when Roma had spent themselves after sixty minutes of man-marking. Perhaps Madrid went more direct after falling two behind, changing the type of ball the front line received. Perhaps Roma, having lost their right-side threat, had to shift bodies across and opened the middle.
A substitution can be the cause. It can also be the timestamp people attach to a change that had already begun.
My point is not that Vinícius would certainly have played well had he stayed. My point is that we lack the data to answer, and lacking data does not license a conclusion in either direction.
Limits of this analysis
I should be explicit about what I do not know.
First, I have no positional or touch data from Roma–Real Madrid. Every description of the overload rests on Mourinho's account, and a coach's account is interested data, not neutral data.
Second, the indices I cite are bands I track across seasons, not figures from this match. I give no specific numbers for this game because I do not have them, and a sound method must survive missing numbers.
Third, I am writing from Liverpool about a match in Rome involving a club in Madrid. Distance does not falsify data. It falsifies the feel of tempo. I did not hear the Bernabéu whistles. I read about them.
Fourth, and most important: I chose to write about Mourinho's line on AI rather than about Vinícius's performance. That choice is itself a bias. Another analyst would write about Endrick's twenty minutes, and would also be right.
The next signal
Mourinho says Vinícius starts against RB Leipzig on Wednesday. That is the clearest signal available, and it is more reliable than any form table, because it comes from the only man who pays for being wrong.
Three things I will watch. First, his touches in the opening fifteen minutes — below his own season average and the problem sits in how the team feeds him, not in his feet. Second, the distance between him and the full-back behind him whenever possession is lost — wider than normal and the Rome overload repeats under a different name. Third, whether he keeps the ball or wins a foul in the calm phases, because that is what his coach asked for and what gets ignored most.
In the long season, the awakened can only lean on their own spreadsheet. But that spreadsheet has to be read in silence, when the argument has died down, by someone willing to admit what it cannot measure.
Data whispers. Those who listen hear something — a player still running into the right space, a coach defending a structure, or simply a crowd that has not yet decided whether to love or to forget. Our job is to stand there long enough to hear the sentence out.
