Nine Layers of Athletics Analysis: When an Analyst Must Say 'Insufficient Data'

In the small office in Tokyo where I spend most of my time each season, there is a data column I have never deleted. It is empty. Not empty because I forgot to fill it, but empty because I deliberately left it as a reminder. That column sits in the tracking sheet of an athletics meet I was once assigned to analyse, and at some point the entire input source vanished. The headline would not load. No athlete name appeared. Marks, dates, venues, wind readings - all blank.

People assume the job of a sports data analyst is to sit in front of thousands of numbers and turn them into predictions. There is a harder skill, rarely mentioned: recognising when you do not have enough data to say anything at all. That empty column taught me more than any packed spreadsheet. When the data speaks, laughter is only noise - but when the data falls silent, that silence itself is the most important signal.

Athletics is the sport where data is almost the entire story. No goals, no scorelines, no ambiguous passages of play. Only time, distance and height - quantities measurable to the hundredth of a second and the centimetre. That is why athletics was the first sport predictive algorithms reached, and also the sport that most clearly exposes their limits. A model can compute the probability of a record falling, but no model can compute what happens when the input feed is cut.

I did not arrive at this work through academia. I was a track athlete, used to having my body measured every day. When I moved into analysis, I carried that habit with me: trust nothing but what can be measured. But the track also taught me that a figure, detached from context, can deceive more dangerously than plain ignorance.

In the summer of 2026, as a second-year student in Tokyo, I wrote a World Cup blog based on data. Before Germany met South Korea in the group stage, I pointed out that Germany's expected goals stood at 2.1 against South Korea's 0.6, yet South Korea logged 121 sprints and a second-half PPDA of 7.8 - an enormous pressing load. I predicted Germany could go out. A male commentator online sneered: 'What does a girl know about football, talking about pressing?' South Korea won 2-0 and Germany went home. My blog was shared thousands of times overnight. Every sneer is an unlabelled data point - I logged it and moved on.

Two years later, when the pandemic halted global football, I collected data from the first 26 Bundesliga matches played behind closed doors. Home advantage, normally worth 0.44 goals per match, fell to 0.15. I built a 'no-crowd' betting model, backed under-priced away teams, and won 17 of 20 bets that month. Home advantage is a hypothesis; COVID was an accidental experiment - and that experiment taught me that context decides the meaning of a number, not the number itself.

That empty summer also taught me that an empty seat is also a player. An empty stand is a tactical variable, not scenery. And when I sat in a meeting room at a Tokyo betting-analytics firm arguing that Italy's average PPDA of 8.9 - the most aggressive pressing at Euro 2026 - dwarfed England's 11.4, a male colleague laughed: 'Japanese women only see numbers, they don't understand the psychology of Wembley.' I slammed the table, put up a chart of the last 30 matches and said: 'The numbers don't lie. You lose if you keep sitting deep.' Italy won on penalties. PPDA does not shoot, but it carried the Italians to the night they lifted the trophy.

In the meeting room, emotion asks and data answers. But outside it, when the data runs dry, I learned that the most honest answer is sometimes simply: unknown.

To analyse an athletics story seriously, I always move through nine layers. They are not a ritual but a way of avoiding self-deception. Each layer answers a question, and if any layer lacks data, I must stop and mark 'insufficient information' rather than plug the gap with guesswork.

Layer one: Performance and event. Every athletics analysis starts from a concrete mark - a time, a distance, a height. No mark, no analysis. The first step is to place that mark on a coordinate system: how far from the world record, how far from the qualifying standard, where it sits in the season ranking. But those three comparisons only mean something once you know the conditions that produced the mark.

A sprint mark is ratified as a record only when the tailwind does not exceed 2.0 metres per second. Above 1,000 metres of altitude, thin air lifts speed artificially, enough to turn a decent athlete into a record-breaker for an afternoon. And since shoes with carbon plates and supercritical foam arrived, part of a performance belongs not to the athlete but to the technology. A poor analyst reads the mark and concludes. A decent analyst reads the mark, then asks under what conditions it was produced.

Layer two: Athlete condition. This is where data is most easily abused. An athlete who runs three seconds faster than last year over 400 metres may be a natural step forward after a year of training, or a signal of an abnormal leap that warrants scrutiny. I use a rough threshold: if the one-year improvement exceeds roughly three times the career annual average, it deserves a close look. But that threshold only runs when I have a year-by-year series. No series, no conclusion. There is also the injury map: sprints and jumps risk hamstrings and Achilles tendons; throws risk shoulders and elbows; distance running risks stress fractures. Know the discipline, narrow the risk zone. Not knowing the discipline, I can say nothing. And peaking is a question of timing: an athlete who finds peak form in championship week usually planned it months in advance, not by luck.

Layer three: Competition structure and qualification. At elite level there are two routes to a major meet: hitting the qualifying standard, or accumulating world-ranking points. Each carries its own risk. The standard demands a perfect moment at the right time; the ranking demands consistency across many meets. Some nations run 'one race decides everything' trials, where a world champion can be eliminated on a single bad afternoon. Some cap three entries per nation per event, sending the world number four home. Understanding that mechanism explains an absence without any speculation about form or injury.

Layer four: Event landscape and national strength. No event exists in a vacuum. Each has its own landscape: absolute dominance, a two-horse race, a wide-open melee, or a generational transition. Looking at squad depth - qualified athletes, average age, the flow of young talent - I build a power map. But that map can only be drawn with names. Without names, I hold a blank map. Trends do not run. People run.

Layer five: Rules and anti-doping. This is the most sensitive layer, and the one demanding the most ethical care. The athlete biological passport tracks blood and steroid markers over the long term, catching anomalies a single test misses. The whereabouts system requires elite athletes to update their location daily; three missed tests in twelve months constitute a violation. There are testosterone-limit rules affecting certain women's events, and an authorised-neutral mechanism allowing athletes from suspended federations to compete as individuals. The lesson: the absence of a warning flag does not mean 'clean'. It means 'unknown'. Confusing the two is the gravest error in the profession.

Layer six: Team and training systems. No athlete peaks alone. Behind every mark sits a system: a fitting coach, a training group, a periodisation programme, a level of technology and recovery. Models differ - state professional systems, the collegiate route, school-based systems, altitude pipelines, overseas training groups. One of the most informative signals is a coaching change just before a major championship; it usually leaves a clear trace in results.

Layer seven: Risk landscape. After six layers I assemble a risk matrix: competitive, anti-doping, financial and career, eligibility, public-opinion and brand, systemic. But there is one risk I always place separately at the end: process risk. The danger that my own analysis is misread - taken as a 'safety clearance' when it is in fact a record that no information exists. A table full of 'not applicable' cells looks identical to a table full of 'no problem' cells.

Layer eight: Public narrative and expectation. Every athlete and every meet carries a story: a record chase, a prodigy's emergence, national pride, a comeback, a farewell, a scandal. The analyst's job is to test whether that story is supported by fundamentals, after correcting for wind, altitude and shoes. A story's position in the heat cycle - germinating, spreading, or at the peak needing to be cooled - decides how I must write.

Layer nine: Industry transmission. Finally I trace the flow from upstream to downstream: from youth development and equipment R&D, through athletes and meets, to broadcast, commerce and derivative markets. A new record can lift a shoe line's sales or open a sponsorship. But I can only draw that transmission path with at least one identified node. And in this layer I set my own limit: no odds or betting-market content, because the line between analysis and incitement is one I do not cross.

After those nine layers, I draw the lesson I consider most important, and it runs against my own instinct. The data analyst is driven to have an answer. When an empty cell appears, instinct says fill it with a reasonable guess. That is the moment analysis becomes fabrication. A complete model with every cell filled looks far more convincing than one that admits missing data. That is the profession's lethal paradox: what looks most professional can be most wrong. I call it the temptation of completeness. It is dangerous because it comes not from malice but from diligence misplaced. Correlation is not causation - an old mantra, but in athletics it runs deeper. A better mark may come from shoes, wind, altitude, a stronger rival pulling you along, a carefully planned training cycle, or a recording error. Assigning a single cause to a complex outcome is how humans comfort themselves in a world full of randomness. And humility before randomness is the antidote.

Humility is not evasion. Its correct definition is opening the door to the possibility that you are wrong. An analyst who refuses to conclude out of fear is useless. An analyst who concludes on evidence while stating that evidence's limits is the one doing the job decently.

The empty column remains. Every time I look at it, I remember that my job is not to provide answers but to ask the right questions and be honest about what I know. The transfer market has no rumours, only prices rediscovering themselves - and sports data is the same: no gap needs filling with guesswork, only gaps waiting to be filled with evidence. The next major season is coming. There will be nights a record falls and millions cheer, and nights a favourite collapses and millions are stunned. My task, between the cheering and the shock, is to keep the line between expectation and outcome unbroken. I do not predict sport; I measure the distance between expectation and result. And when that distance cannot yet be measured, the only honest thing is to say so.