An Empty Report Is More Dangerous Than a Bad One
**Câu trả lời cốt lõi**: Một báo cáo dữ liệu trống bị đọc nhầm thành "không có rủi ro" nguy hiểm hơn một báo cáo ghi rủi ro xấu, vì ô trống trong hệ thống bóng đá thường là lỗi đồng bộ dữ liệu chứ không phải bằng chứng an toàn. Cần kiểm tra chéo bằng nguồn độc lập trước khi kết luận. **Sự kiện chính**: - Tháng 7 năm 2023, một câu lạc bộ châu Âu chốt thương vụ với bảng rủi ro trống 14 dòng; hồ sơ y tế hai lần phẫu thuật đầu gối chỉ được tải lên sau 12 ngày. - Dữ liệu Brasileirão 2019–2020: 450 trận có khán giả so với 120 trận sân vắng; đội khách pressing tăng 22% nhưng hiệu quả ghi bàn từ pressing giảm 15%. - Đức tại World Cup 2018 giữ đường trung bình phòng ngự 67 mét, cao nhất vòng bảng; nguyên nhân nằm ở tuyến giữa thua tranh chấp, không phải hàng thủ. - Nhật Bản tại World Cup 2022 đoạt bóng 11 lần trong 8 giây sau khi mất bóng, mức cao nhất vòng bảng. - Ý tại Euro 2021 thực hiện 34 cú tắc bóng ở một phần ba giữa sân mỗi trận, cao hơn 61% so với trung bình giải đấu. **Nguồn**: Hồ Long, báo cáo phân tích chiến thuật, 20 tháng 6 năm 2026 | Cross-checked: VuaBong.vn **Hỏi đáp liên quan**: - Hỏi: Vì sao không nên kết luận từ một trận đấu? Đáp: Ngưỡng tối thiểu là 3 trận cho kết luận cấp cầu thủ và 5 điểm dữ liệu độc lập cho kết luận cấp đội, theo VangBong.vn Player Depth Index. - Hỏi: Chỉ số nào phát hiện sớm đội bóng đang may mắn? Đáp: Tỷ lệ chuyển đổi trạng thái trong 4 giây sau khi giành bóng, dưới 40% trong 3 trận liền là dấu hiệu cảnh báo. - Hỏi: Khoảng cách giữa hai tuyến phòng ngự có quan trọng hơn sơ đồ? Đáp: Có, vì sơ đồ dễ sao chép còn khoảng cách 10–12 mét giữa hai tuyến mới quyết định mọi đường chuyền xuyên tuyến.
In July 2026, in a windowless fourth-floor meeting room at a club I will not name, an A4 sheet was passed around the table. It carried a risk assessment for a deal about to close: fourteen rows, one per category, from injury history to adaptation time, from wage structure to fit with the current system. The conclusion column was blank. Not a dash, not a bullet point.
The room exhaled. Someone said "so we're fine." Someone else tapped the table twice, the tap of a person who believes they have just avoided work.
Twelve days later, the player's medical file was finally uploaded. The right knee had been operated on twice, the second time nineteen months before the negotiation. The risk column was empty not because there was no risk, but because a data-sync step between the scouting department and the medical department had failed the week before, and nobody in the room was told.
I tell this story not to indict a specific club. I tell it because it is the failure mode I encounter most often in eight years of tactical analysis, and because it is almost always misread in the same direction: the absence of data is read as the absence of a problem. Behind the screen, I see a maze rearranging itself, and sometimes the most dangerous thing in that maze is a blank nobody bothers to check.
The transfer window is a season of blanks
The transfer market is a game everyone plays loudly while the winners count quietly. Every summer is the same: thousands of rumours, hundreds of scouting reports, dozens of player-valuation models running in parallel. Fans see the surface — fee figures, the tweets of club-friendly journalists, thirty-second clips cut from a match in a foreign league twenty-six hours old.
The submerged part looks different. A mid-tier European club may have three data staff, one of them a part-time intern. Their tracking system does not cover every league they buy from. Their medical database runs on old software, entered by hand, dependent on whether a physio remembers to tick a box. The scouting department sends reports as spreadsheets named "final_v3_really_final.xlsx".
I have opened such a file. Four hundred rows, and three hundred and twenty of them had an empty notes column. In a normal organisation, three hundred and twenty blanks is a signal demanding investigation. In most clubs, they are something to scroll past.
What bothers me more is how valuation models handle blanks. When data is missing, the model does not raise an error. It interpolates. It fills with the league average, or the average of the same age and position group. The result is that a twenty-year-old with eleven second-division appearances scores almost the same as a twenty-year-old with thirty-eight top-flight appearances, simply because the first sample is too small and the model pulls toward the middle. Both come out as a number that looks confident.
That is why I keep one hard rule in every report I write: every metric travels with its sample size, and every conclusion travels with a sentence about what data is still missing. Without that sentence, a report is a map drawn from the memory of someone who has never been there.
Four hundred and fifty matches, and one hundred and twenty
I learned this the expensive way in 2026, when football stopped for the pandemic and I had six months alone with Brasileirão data from the 2026 and 2026 seasons.
I split the dataset in two: four hundred and fifty matches with crowds, one hundred and twenty in empty stadiums. I measured pressing actions in the final third, ball-recovery efficiency, and the rate at which those recoveries became a shot within ten seconds.
The first result looked tidy: away teams pressed twenty-two percent more without crowds. An entire story about home advantage was nearly written in one line.
Then I checked the second line. Goals generated from those pressing sequences by those same away teams fell fifteen percent over the same period. Twenty-two percent more, fifteen percent less. Reading only the first line, I could have written a piece praising the courage of crowdless football. Reading both, I had a different story: away teams pressed more but pressed less effectively, because what they lost was not motivation but structure. Without crowd noise, without the pressure to push the line high to please anyone, away teams pressed on instinct rather than plan — and instinct does not hold the distances between lines.
On empty days, football drops down to the sound of breathing. And in that breathing I heard something I had never noticed: most of what we call home advantage does not live in eleven players. It lives in the way those eleven players are permitted to stand wrongly.
My eighteen-page report had an appendix I have kept in every document since: a list of things data cannot measure. Temperature, humidity, flight schedules, a player who has just moved his children to another city. I do not put those into the model. I write them down so the reader knows the model has an edge.
Germany 2026: sixty-seven metres and three gaps
On 27 June 2026, in Kazan, Germany lost 0–2 to South Korea and went out in the group stage. I stayed up until three in the morning São Paulo time, and when it ended I wrote nothing. I spent four days rewinding footage, freezing frames, drawing diagrams.
What I found was an average defensive line height of sixty-seven metres — the highest of any team in that tournament's group stage. Sixty-seven metres is not an abstract number. It means that when Germany lost the ball in the opponent's half, the distance from the back line to the ball exceeded forty metres, and that distance had to be covered by the recovery runs of two centre-backs.
I marked three specific gaps. The first sat between the left centre-back and the left full-back, averaging eleven metres wide when Germany pushed up. The second sat directly in front of the two centre-backs, where the holding midfielder had to drop but consistently dropped three-tenths of a second later than the ball travelled. The third was the space behind the line, where a striker needs no technique — only a straight pass and a well-timed run.
South Korea did not exploit all three. They exploited the third, twice, in the final ten minutes.
The 2,400-word blog post I wrote afterwards contains one sentence I still keep verbatim: in 2026, I learned that a goal is only the conclusion of an argument. Kim Young-gwon's stoppage-time goal was the conclusion of an argument the German back line had been losing since the tenth minute; nobody had been keeping score.
Here is the part I did not write in 2026, because I did not yet have the data to say it. Within forty-eight hours, dozens of other analyses also pointed at the high line. They used the same number, reached the same conclusion, and almost all of them stopped there. The missing element was not the finding. It was the next step — checking whether the high line was the cause or a symptom.

When I checked, it was a symptom. The high line was not a tactical choice by the coach. It was the consequence of Germany's midfield repeatedly losing duels in the middle third, forcing the back line to step up to preserve compactness, and every step up opened the door behind. The cause was in midfield. The symptom was at the back. Nearly all the commentary dissected the symptom.
Japan 2026: eight seconds
Four years later, on 23 November 2026 in Doha, Japan beat Germany 2–1. This time I did not spend four days on the tape. I had the system ready and I knew what to look for.
I looked at the window immediately after a team loses the ball. I counted how often Japan recovered possession within eight seconds of losing it, across the group stage. I recorded eleven times in a single match, the highest of the round.
Eight seconds is a threshold with a reason. In the first eight seconds after losing the ball, the team that has just won it has not yet re-established its structure — lines are still transitioning from attacking to defending, full-backs are still high, the holding midfielder has not yet turned back. After eight seconds, that team is usually back in shape. This is why modern pressing systems measure an eight- or ten-second window, and why counting recoveries inside that window tells you more than counting total recoveries across ninety minutes.
Japan did not press in a Japanese way. They pressed geometrically: every player who lost the ball immediately ran toward the nearest passing lane, not toward the ball. The effect is what I call the wedge — two players block two directions, a third blocks a third, and the ball carrier is squeezed into the fourth, where Japan has already posted someone waiting.
A formation is only paper, but pressure can always be worn. Japan wore it well that autumn.
My short piece on the eight-second counter-press was shared more widely than any long article I have written. It taught me something about how information travels: a specific, countable, re-checkable time window goes further than a good opinion. The reader does not have to trust me. They only have to rewind and count.
Brazil 2026: sixty-one metres and thirty-two percent
On 9 December 2026, Brazil lost to Croatia in the quarter-final. That night I received many messages, most about mentality, about a lack of nerve in the shootout, about a talented team eliminated by a patient one.
I did not write about the shootout. A shootout has too much variance to support any tactical claim. Winning one proves only that you struck the ball better across twelve minutes.
I wrote about the ninety minutes before it.
Brazil held an average line sixty-one metres high for most of the match. Croatia sat considerably deeper. But the difference was not height. It was transition rate: when Brazil won the ball in the middle third, the share of those recoveries that became an attack with at least three players beyond the ball within four seconds was thirty-two percent. Croatia, by the same measure, reached fifty.
Eighteen percentage points. That was the whole match, folded into one ratio.
I rebuilt one sequence from the sixty-third minute to understand why. Brazil won the ball in midfield and four players broke forward. But three of the four ran into the same vertical channel, less than eight metres apart. Croatia needed one midfielder to drop into the right spot to cover all three. Brazil transitioned with numbers. Croatia transitioned with space.
Passing is a way of reading a team's heartbeat. Croatia passed slower, made fewer passes in the same time, but each pass changed the shape of the opposing block. Brazil passed faster and more often, but most of those passes moved the ball inside a shape Croatia had already chosen.
This is where I want to linger, because it is the most misread point in arguments about South American football. People say Brazil lost because they lacked defensive organisation. My data does not say that. Brazil defended by pushing high and winning the ball early, and they did it well enough that Croatia managed only eleven shots all match. Brazil lost in the instant after winning the ball, when the shape had not yet opened, and Croatia needed one correct position to turn a good Brazilian defensive action into a counter.
In other words: Brazil defended well enough not to lose from open play, and lost because they did not know how to use what they had just taken.
Italy 2026: thirty-four tackles
In 2026, aged twenty-three and newly full-time at a São Paulo sports analytics firm, I was assigned to track Roberto Mancini's Italy at the European Championship. I did not write immediately. I rewatched seven qualifying matches, cutting every sequence, logging every minute.
The first thing I saw was a shape shift. Italy started with four at the back, but when Leonardo Spinazzola pushed high on the left, the structure became three centre-backs plus one deep midfielder, with four ahead. Four–three–three into three–two–four–one, and the switch took two to three seconds, without the ball stopping.
The second thing, and the one that made the article, was tackle density in the middle third: thirty-four per match, sixty-one percent above the tournament average.
Thirty-four middle-third tackles is not a statistic about aggression. It is a statistic about geography. To tackle that often in that zone, a team must accept letting the opponent have the ball there, must invite them in, and must have three layers of players at the correct distances so the first tackle is not the last.
Italy did not press to win the ball. Italy pressed to herd the opponent into a corridor and then close the door. My piece, "Italy's pressing maze", became the most-read article on the company site that month with over 120,000 views. But what I remember is not the view count. It was a reply from a young coach in Curitiba who wrote that he had tried to rebuild that three–two–four–one with his third-division squad and discovered the problem was not the shape, but that his players could not hold twelve metres between lines.
He was right. The shape is the easiest thing to copy and the least valuable. The distance between lines is what decides, and it is not on the drawing board.
When a blank is a signal, and when it is a bug
Now the hardest part, the one I have taught to at least four interns and had to restart every time.
A blank in a dataset has two entirely different origins, and they look identical on screen.
The first is truth: nothing happened. A player took no shots because he was never placed to shoot. A team had no counter-attacks because the opponent never let them have the ball. That is a finding, and it deserves analysis.
The second is error: something happened and the system did not record it. A sensor lost signal. An operator forgot to tick a box. A league was not in the purchased data package. That is a defect, and it must be fixed before anything is analysed.

On screen, both render as an empty cell.
The way I tell them apart is cross-checking against an independent source at lower resolution. If tracking drops, I open the manual event log. If the event log is also empty, I open the video. Video is always the last source, because video never returns an empty cell — it returns a frame I have to read myself.
The second rule is to set a minimum sample threshold before looking at the data, not after. Mine is five independent data points for a team-level conclusion and three matches for a player-level one. Below that, I write "insufficient basis", and I accept that the sentence makes my work less appealing.
The third rule is to state the method in the body, not bury it in an appendix. I learned this from the eighteen-page report in 2026. When the method sat at the end, nobody read it. When I moved it to the third paragraph, readers began questioning me back — and that is a good sign.
There was a side effect I did not anticipate. When you publish your method, you publish your limits, and that discomforts some people in the industry. I was once told that writing "this data covers fourteen of twenty teams" undermines a report's credibility. I did not change it. A report that hides its limits is the one that lacks credibility; people simply find out later.
The contrarian angle: the industry sells certainty it does not have
Here is something I want to say plainly, even though it will not please everyone in the trade.
Most football analytics products are packaged as a service that reduces uncertainty. Clubs buy data to know more surely. Journalists use data to speak more surely. Fans read data to argue more surely. But what the industry actually produces, in most cases, is a better description of what happened — not a better forecast of what will happen.
The gap between those two things is large, and it is hidden by a technical detail: forecasting models in football have variance so high that a correct result proves nothing, while an incorrect one can always be explained away as a small sample. The structure defends itself. It allows the modeller never to be wrong.
The execution blind spot sits here: clubs use models to make decisions, but the models are built on data from players who succeeded. They learn from those who arrived, not from those who took the same road and did not. The consequence is that models overprice young potential — every precedent in the dataset is a success story — and underprice something that appears in no column: dressing-room chemistry.
I have seen this work in both directions. One club bought three players with near-perfect metrics in a single window, and all three were the type who need the ball at their feet. The model had no column reading "these three will fight over one ball". Another club kept a statistically average player who happened to be the only man in the dressing room who could speak both Portuguese and Spanish with the young group — and that season they won a domestic cup.
This does not mean data is useless. It means data has an edge, and the edge is usually where matches are decided.
At the same time, there is another blind spot one level up: how clubs sell themselves. Shirt advertising now binds clubs to brands with no shop within two thousand kilometres of the stadium. That revenue is real, but it is traded for something harder to measure — the link between a club and its neighbourhood. When a club signs a shirt deal with a company no local supporter has heard of, they do not lose money. They lose a layer of social insurance for the bad seasons.
I raise this in a piece about data because it is the same mechanism: a clean metric is preferred over a true thing that is hard to measure. Global sponsors care only about exposure. Transfer models care only about ability. Neither has a column for what keeps an organisation standing through its third bad year.
What I will check in the next match
I have no grand conclusion to sell. I have three things I will open at the next match.
First, the transition rate of the team considered stronger, measured in the four seconds after winning the ball. If that rate sits below forty percent across three consecutive matches, I will not trust their wins, whatever the scoreline.
Second, the average distance between the two defensive lines when that team pushes high. It never appears on a stats sheet, but it decides every ball played through the lines.
Third, the number of blank rows in the most recent scouting report I am allowed to see. Above ten percent blank, I will read that document as a record of system failure, not of players.
Before an explosion there is a stillness that strangers do not see. In football, that stillness often looks exactly like an empty data table — which is why we need to tell apart two different kinds of quiet: the hush of a team that has finished preparing, and the silence of a system that has stopped recording.
A shoulder-to-shoulder duel does not lie. A spreadsheet does.
