The Haematocrit Curve and 12,000 Bot Accounts: Two Transfer Files I Have Chased for Nearly Two Decades

On 14 July 2026, in a hotel four kilometres from the Benito Villamarín stadium, I sat across from a 96-page file. Page 71 was an annex almost nobody reads: a biological monitoring table for a 22-year-old Brazilian winger, 1.79 metres tall, 71 kilograms, registered in the contract under the single name Márcio, shirt number 17. The table had twelve columns, and the ninth recorded haematocrit by quarter. March 2026: 43.1%. November 2026: 44.6%. June 2026: 46.8%. February 2026: 49.4%. June 2026: 51.9%. An upward curve, steady, almost drawn with a ruler.

When a biological marker rises in a straight line while match volume falls, the cause is not training. I stayed until three in the morning redrawing that curve on graph paper, cross-referencing it against the third-division São Paulo state calendar, and found one detail that kept me awake: the steepest increase occurred during a period in which the player featured for only 210 minutes across four matches, two of them as a substitute.

Betis hid the doping inside a contract annex; I read it backwards, page by page, to find it. But it took another two years. In May 2026, when the Spanish anti-doping agency announced a two-year ban for the player, with erythropoietin as the prohibited substance, my 96-page file was finally classified as "valuable". Before that, people called it the hobby of a retired swimmer.

I tell this story not to praise myself. I tell it because in the current season, as transfer reporting in Vietnam fills with numbers inflated by algorithms, supporters are being placed in exactly the position I occupied nearly twenty years ago: reading a figure written by the seller, and believing it is objective truth.

Context: a market priced by three kinds of people

To understand how an unknown third-division Brazilian player could be bought for twelve million euros in 2026, you have to look at the financial structure of Spanish football at the time.

Between 2026 and 2026, La Liga broadcast revenue nearly doubled after the major clubs began selling rights collectively rather than individually. The new money did not flow evenly. It flowed to the two largest clubs, and everyone else had to compete by buying players to hold a European qualification place. The result is what economists call competitive inflation: when the supply of high-quality players does not grow at the speed of the money, the gap is absorbed by substitute markets — lower divisions, under-scouted countries, players with no public data.

The Brazilian third division in 2026 was a perfect blind spot. No full broadcast coverage, no detailed event data, no independent valuation system. A player there is worth 200,000 euros when clubs trade with each other, and twelve million when the buyer needs a number for the accounts.

Three kinds of people price a footballer. The first is the talent assessor, working from video and data. The second is the intermediary, working from relationships and timing leverage. The third is the liquidity creator, using money flow to turn a name into an asset that can be resold. In the Márcio file, all three appeared, and two of the three did not care in the slightest whether the player was fast or slow.

The Brazilian ran fast; the Betis contract ran faster, until the doping caught up.

Core one: reading a biological curve like a price chart

Haematocrit is the percentage of red cell volume in whole blood. In adult men, the conventional reference range sits between 40% and 50%, and natural variation between individuals can reach four to five percentage points. A reading of 52% in an athlete does not automatically mean wrongdoing.

What does not fall inside natural variation is the rate of change.

In my file I built a seven-column comparison table. The three most important columns were haematocrit, minutes played per quarter, and reticulocyte count. Reticulocytes are immature red cells newly released by the bone marrow. When the body produces red cells naturally, reticulocytes rise first, then haematocrit follows. When red cells are introduced from outside, the marrow is suppressed by feedback, reticulocytes fall, and haematocrit rises. Two markers moving in opposite directions are more informative than any absolute threshold.

In the Márcio file, the period from June 2026 to February 2026 showed haematocrit rising 2.6 percentage points while the reticulocyte count remained flat at a low level. That was the data structure I needed. It was not enough to conclude, but enough to open a file.

I want to pause here and state a principle I have kept across 44 years of watching this industry: a data anomaly is not a verdict. It is a question that requires additional evidence. In this case, additional evidence came from two directions. The first was a former club doctor who agreed to meet me at a suburban café and confirmed he had once been asked not to record certain results. The second was a team medication log listing an item ordered in quantities far beyond ordinary treatment needs.

I did not publish the doctor's name. He lost his job two months after our meeting, and that is a price I do not want anyone else to pay.

It is worth noting that the club was never found in breach in this matter. The player was banned; that is an individual event. My reporting of the existence of a contract annex containing biological data is a procedural event. The two are different, and I keep the distance between them.

Core two: reading a contract from the last page backwards

The technique I use on a transfer contract is to read it in reverse. The first page is the presentation, written to be published. The last page is the annex, written so nobody reads it. If you want to know how a deal really operates, start where the light is weakest.

In my 96-page file, the annexes fell into five groups: staged payment terms, image rights, signing fee, resale terms, and termination.

The third group caught my attention. The signing fee was recorded at 2.1 million euros, equal to 17.5% of the total deal value, payable to the player and to an unnamed third entity described only as a "cooperation representative". In a club's accounting structure, signing fees are allocated to personnel costs, not to transfer value. That means when a regulator reads the report, they see a smaller figure than the number published in the media.

This is the root of a problem still unresolved years later: signing fees for free agents do more damage than transfer fees, because they sit outside the core scope of financial fair play supervision. A club can sign a free agent on a modest salary and an enormous signing fee routed through intermediary entities, and technically no clause is breached.

I do not trust transfer fees; I trust numbers that have been crossed out. In this file, the crossed-out number sits in a handwritten marginal note on page 84, where an accountant wrote two abbreviations that took me three weeks to decode. They indicated that 400,000 euros of the signing fee went to an overseas account, not to the player.

Sending money to an overseas entity is not illegal. Not describing that payment clearly in the published contract is a transparency problem. The two are different, and I keep that distinction in every piece I write.

Core three: the Girona bot network and the server logs

In 2026 I was 51 and in what newsrooms call the veteran phase. In this industry, veteran is a polite word for no longer being assigned the stories that need speed.

Even so, I followed a case that felt familiar. A Catalan club newly promoted to the Spanish top flight sold a 22-year-old defender to an English club for a reported 25 million euros. Independent analytics sites valued the player between 2.2 and 2.8 million. A multiple of roughly ten.

Girona inflated a player's value with a bot network; the real value was in the server logs.

I began with the player's Instagram account. Between March and June 2026, it grew from 8,400 followers to 412,000. A 49-fold increase in four months, for a 22-year-old defender who had never played for the national team.

I downloaded the public interaction data for the most recent 40,000 engagements. With help from a data engineer I knew through a betting investigation, we analysed three indicators: the share of accounts with default avatars, the share with fewer than five posts, and account creation timing.

The result: 12,000 of the 40,000 interacting accounts shared technical fingerprints so similar it was hard to credit. They were created in the same window, used the same API password pattern, and interacted on an identical rhythm — a new cluster of engagement every 47 minutes. Humans do not interact on a 47-minute rhythm. Machines do.

I traced that API password pattern back to a digital media company registered in Barcelona. According to public company records, its legal representative shared a surname with the club president. I have no evidence of a direct instruction, and I did not write that a direct instruction existed. I wrote that a relationship between two entities existed, and that the relationship was not disclosed in any document of the deal.

My 3,500-word investigation was ignored by the national federation. No sanction followed. In 2026, European football's governing body began requiring clubs to justify transfer values against real indicators, and part of the impetus came from files like this one.

From the 2026 press room to the 2026 Girona bots: power only changes shirts.

What I learned from this case was not a conclusion but a method. Before 2026 I investigated with video. After 2026 I investigate with server logs, API traces, and interaction rhythm models. The same logic: find where the data is generated, not where it is published.

Core four: valuation structure and the art of writing a number into an asset

There is a question I always ask before any financial file: who benefits if this number is believed?

In the 25 million euro deal, four groups benefited. The selling club, because transfer revenue is recognised immediately in the financial year, balancing the accounts and improving compliance with spending limits. The intermediary, because commission is calculated as a percentage of deal value. The buying club, because the investment is recorded as an asset that can be amortised over time, creating a flexible accounting instrument. And the player, who can use the figure as the basis for his next contract.

Of those four, only the player carries real risk. If the 25 million does not match ability, the pressure falls on the player, not on the people who signed the contract.

I followed the player's career over the next five seasons. His average minutes per season in the Spanish top flight were 1,180. His defensive event metrics sat at the average for full-backs in his position. There was no sign of a 25 million euro player, and no sign of a 2.5 million euro player either. He was somewhere in the middle, exactly as the independent data predicted.

That is the whole problem. Transfer value does not measure ability. It measures the strength of the storytelling machine behind the ability.

Core five: when the substitution rule meets the financial rule

One thing rarely discussed is the direct link between the five-substitution rule and clubs' spending structure.

The five-substitution right advantages deep squads, but it also turns the final 20 minutes into a war of attrition. Previously, with three changes, a coach had to hold resources back for the worst case. With five, tactics become a resource-allocation problem across 15-minute blocks.

The transfer consequence is clear. A club that wants to compete in the final 20 minutes needs at least seven players good enough to come off the bench. That demand pushes up the price of the near-good-enough group — precisely the group mid-tier clubs usually sell. The result is a spiral: big clubs buy more, small clubs must revalue their assets higher, and the middle of the transfer market is dragged upward.

Based on my experience watching matches between 2026 and 2026, average substitutions per match in Europe's top leagues rose from roughly 4.1 to 8.6. This is a structural change, not a small trend. And every structural change in football ends up in a balance sheet.

Vietnamese clubs watching European football often copy the tactics and ignore the finances. That is a mistake. If you want to press for 90 minutes, you need 16 players of sufficient quality. If you have 11, you are buying a tactic you cannot resource.

The contrarian angle: the legitimate side of the investigated party

I have spent most of this piece laying out anomalies. Now I must present the other side, because a file without the other side is a dishonest file.

First, a club valuing its own asset highly is not fraud. It is profit maximisation, and it is lawful. The duty to check lies with the buyer and the regulator, not the seller. When I wrote about the Girona bot network, I did not say the club broke the law. I said the governance system had no tool to inspect a new form of asset.

Second, using social media to build a player's profile is standard marketing in the entertainment industry. Music has done it since the 1990s. Film much longer. Football following suit is not a moral failing. The problem is that when an asset is valued on marketing indicators generated by the seller, the market loses its cross-check function.

Third, and this is the point I want to stress most: regulators are not weak because they lack competence. They are weak because they are designed to check what can be checked. A system built in 2026 to audit an organisation's financial statements has no tool to inspect 12,000 social accounts created in four months. That is a design gap, not a conspiracy.

Before deepfakes there were transfer rumours; both are tricks that need exposing. But they need exposing differently. A rumour needs data to be refuted. A deepfake needs technical infrastructure to be identified. Using one tool for both is methodological laziness.

I must apply that standard to myself. In the 2026 investigation I nearly made an error. In the first draft I wrote that the media company had been "established to serve the deal". That is an inference about intent, not a fact. I deleted the sentence. The company was founded 14 months before the transfer, and I had to say exactly that, even though it made my piece weaker.

An investigator who writes what he wants to believe becomes a propagandist with a spreadsheet.

On refereeing errors and the limits of video evidence

In 2026, in Russia, during the semi-final between France and Belgium, security stopped me at the gate to the commentators' area because a woman's name appeared on my accreditation. They said the area was for men.

I did not raise my voice. I bought a ticket in the stands, carried a small camera in my coat pocket, and recorded the match. In the 67th minute I captured an attacker pushing the ball with his hand inside the penalty area. The referee waved play on. I sent the 15-second clip to a refereeing body I had contacts with. The result did not change, but the clip was used in refereeing training material for the 2026 European Championship.

In 2026 they closed the press room; three decades later I opened the files wide.

But I must be clear about one thing regarding video evidence: it is limited by camera angle. A frame shows a hand touching the ball. It does not show force, intent, or whether the referee had line of sight. In my writing I always note the minute, the player's position on the pitch, and a "footage dated..." annotation so readers know exactly what I am relying on. I publicly state that the footage copyright belongs to the author, to protect the independence of the source.

Refereeing errors are part of football and should be recorded as data, not as personal accusations. When I write about an error, I am not writing about a referee. I am writing about a training system and an appointment process.

Investigation method: how I build a file

I include this section at the end of every investigation for three reasons: so readers can check me, so colleagues can replicate, and so I do not fool myself.

Data collection: contract documents from internal sources, match event data from commercial providers, social media data collected through public interfaces, and company registration records from public authorities.

Analytical tools: self-built spreadsheets for biological data and transfer values, rhythm-pattern analysis for social interaction data, and cross-verification against at least three independent sources for every claim.

Source reliability: I classify sources into four tiers. Tier one is verifiable primary documents. Tier two is testimony from a direct participant supported by documents. Tier three is testimony without documentary support. Tier four is industry rumour. I never build a claim on tier three or tier four alone.

Limits of this file: I have no access to banking data, no power to subpoena witnesses, and no legal authority. My conclusions should be read as hypotheses with supporting evidence of varying strength, not as verdicts.

What 44 years of watching taught me

The industry has changed enormously on the surface. In 2026 I started at local radio stations, recording on cassette and writing on a typewriter. In 2026 I was pushed out of a press room. In 2026 I read contract annexes by hand. In 2026 I learned to read APIs. In 2026 I filmed with a small camera hidden in a coat pocket.

But the power structure has not changed. Those who control the money flow still write the number. Those who write the number still control the story. And those who control the story still decide who sits in the press room.

The only thing that changed is the tool for checking. In 2026 I borrowed a colleague's tape and watched it 30 times in one night. In 2026 I downloaded 40,000 interactions and found 12,000 machine accounts. The same act: patiently re-reading what others skipped.

In Vietnamese football I see a familiar paradox. Supporters have a great deal of information and very little capacity to verify it. Transfer reports are shared faster than they can be checked. A rumour can circle social media in two hours, while verifying it takes two days.

That speed gap is the habitat of every kind of manipulation.

A progressive thought: responsibility belongs to whoever reads it a second time

I will not end with a call for reform. I will end with an observation about behaviour.

In every file I have opened, the decisive factor was never a secret document. It was whether someone was willing to read it a second time. The annex was on page 71 of a 96-page file. The haematocrit curve was in column nine. Twelve thousand bot accounts sat inside a 40,000-row dataset. None of it was hidden somewhere secret. All of it sat in plain sight, where nobody bothered to look.

For Vietnamese supporters following a transfer season full of figures, I suggest one small habit: every time you read a transfer fee, ask three questions. Who published this number. Who benefits if I believe it. And is there a source independent of the publisher.

Those three questions require no degree. They require only patience.

And patience is the one thing football has never been able to buy with a transfer fee.

Glossary of professional terms

Haematocrit: the percentage of red blood cell volume in whole blood. The conventional reference range for adult men is 40% to 50%.

Reticulocyte: an immature red blood cell newly released by the bone marrow. It rises when the body produces red cells naturally and falls when red cells are introduced externally.

Erythropoietin: a hormone stimulating red blood cell production, used as an oxygen-transport enhancer and banned in sport.

Signing fee: a payment to a player or representative on signing, especially common for free agents, typically booked as personnel cost rather than transfer value.

Third-party ownership: a model in which part of a player's economic rights belongs to an entity other than a club, widespread in South America and Europe before being restricted.

PPDA: passes allowed per defensive action, used to measure a team's pressing intensity.

API trace: technical information left when an account or application connects to a platform, useful for identifying automatically created accounts.

Interaction rhythm pattern: the repeating timing pattern of engagements, used to separate human from automated behaviour.

Disclaimer

This article is based on publicly available information and the author's own long-term monitoring files. Some individual names have been omitted to protect sources. The conclusions presented here are assessments based on evidence of varying strength, not legal findings. The content is intended for sports information purposes only and does not constitute any betting advice. Sporting outcomes are highly uncertain, and readers should approach the analysis rationally.