The Mislabelled File in Mixcoac: The Cost of Contaminated Data in Football Analysis

On my desk in Incheon, the file arrived tagged “football.” Inside: twenty-five information points, not one team, not one player, not one coach. The entire content revolved around a violent traffic altercation on Avenida Revolución in the Mixcoac district of Benito Juárez borough, Mexico City, a video that spread fast across social media, and an internal file opened by the Mexico City Secretariat of Citizen Security targeting both the attackers and the conduct of traffic police officers at the scene.

In thirteen years covering this industry, I have grown used to data arriving late. Data arriving on the wrong subject is new to me.

Benito Juárez borough hosts a stadium that has staged professional fixtures. That is a geographic coincidence. The original report never mentions the stadium, never mentions a match, never mentions any football institution. But an automated classification model only needs to catch that place name once, and the whole file is pushed straight into the football analytics pipeline.

A wrong label does no damage where it appears; it does damage at every step downstream, where nobody checks the source again.

That is why I treat this incident as more serious than a street brawl.

The deep-analysis framework I work with has nine dimensions: tactics and technique, club finance and the transfer market, results and public-opinion cycles, league landscape and team positioning, rules and governance, management and dressing-room dynamics, risk profile, media narrative and expectations, and industry transmission. All nine exist to answer one question: what is actually happening to a football team.

When the input is a traffic altercation, all nine return the same verdict: insufficient football-relevant information.

Newcomers read that verdict as failure. I read it the other way round. A system that returns “not enough data” is a system that is still honest. The frightening system is the one that still returns a complete football answer — with diagrams, with heat maps, with conclusions — purely because it was asked to answer.

In 2026, while writing for local radio stations, I filed a match report on a game I had never watched. I wrote from someone else's description and added numbers from a statistics site of unclear origin. It went to air and nobody caught it. Three weeks later that site revised its data, and my report became a false memory. Since then my rule has been simple: no source, no numbers.

Data only means something when we ask at the right moment; ask at the wrong moment, and every figure is noise. Ask about the wrong subject, and every figure is organised fabrication.

In 2026, when K League 1 stadiums closed during the pandemic, I sat at SportsData Korea collecting data from 142 matches played without crowds and set it against 142 pre-pandemic matches. The home win rate fell from 47% to 41.5%, and average goals per match rose by 0.7. That dataset was clean: same league, same recording method. I still took until December to finish the report because I wanted every variable to be perfect. A colleague said it plainly — good data published late is no different from a prediction made after the match.

The lesson was not about speed. It was about defining what your dataset measures before you measure anything.

In 2026, when Morocco reached the World Cup semi-finals, I spent five days on their six matches. Losing the ball, they completed a 5-4-1 defensive shape in an average of 2.3 seconds. Full-back Achraf Hakimi advanced an average of 58 metres per match, and when he dropped, the flank was covered by midfielder Azzedine Ounahi. Twelve heat maps, 3,500 words, 1.2 million views.

But to get those numbers, the first job was deletion. Deleting matches from a different fitness phase, deleting sequences that did not begin from a controlled loss of possession, deleting clips mislabelled on aggregation platforms.

The deletion work never appears in the final article. It sits in the submerged part.

Gaps do not disappear on their own; they simply change their name to failure. The Mixcoac case shows that the submerged part is being undervalued across the entire football content chain, from raw data to analysis to the news report.

Every tactic is a hypothesis until an opponent forces you to answer. Every data model is a hypothesis until reality forces it to say: I don't know.

The second notable element is the public reaction. After the video spread, opinion questioned the violence of the group on camera, and also the way traffic officers intervened at the scene. The Internal Affairs unit of the Mexico City Secretariat of Citizen Security opened a file and summoned the officers involved to establish whether protocol was followed. That is a civic accountability cycle.

In football we have an identical cycle under a different name: VAR.

Both turn on a vague concept. In football it is “clear and obvious error.” Here it is “football-related content.” Both are written as though they draw a hard line, when in practice they depend on who is reading and when.

This season I have rewatched a great many VAR incidents. Same contact, same camera angle, two referees, two different conclusions. Neither is technically wrong. They are simply answering two different questions while believing it is one.

Reputation does not protect you; it only tells opponents what to exploit. That holds for players, and it holds for data. A famous, heavily cited data source gets trusted beyond its merits. The big football data aggregators sit exactly there: treated as truth, when most of their value lies in speed rather than accuracy.

For readers following the annual season, this has a concrete meaning. When a title-race pressure table appears after a matchday, what matters is when the underlying dataset was labelled, by whom, and how many matches were quietly removed.

Your team's next match will produce a new number. Whether it deserves belief depends on whether the person who produced it dares to publish the submerged part.