The 20-Millimetre War: How Elite Athletics Is Being Repriced from the Sole Up

Five Thousandths of a Second in Saint-Denis

On the night of 4 August 2026, on the purple track of the Stade de France, the eight fastest men on the planet fitted inside a frame barely two metres wide. The scoreboard hung in the air for almost thirty seconds before it answered. Noah Lyles: 9.784. Kishane Thompson: 9.789. Five thousandths of a second. Less than a breath, less than the time it takes a neon sign to blink once.

I was sitting less than forty metres from the track, headphones plugged into a small recorder, my left hand resting on a notebook. The notebook has seven columns: reaction time, splits at 0-30m, 30-60m and 60-100m, peak speed, stride frequency, and a note on feel. Those seven columns are my entire data bank for one final. Three hours later, with the stands empty, I sat alone under the last of the floodlights and wrote the first line: reaction 0.178 seconds, the slowest of the eight, and still the champion.

That is why I love this sport. Muscle speed has a ceiling. The information layer that decides the result does not.

The Empty Seat at Tokyo National Stadium

The empty summer taught me that an empty seat is also an athlete.

In 2026, when European football returned in stadiums without spectators, I sat in front of three screens and collected data from the first twenty-six Bundesliga matches. Home advantage fell from an average of 0.44 goals per match to 0.15. That collapse taught me something every sport with a crowd must hear: the competitive environment is a variable, not decorative background.

Athletics learned that lesson more painfully. The Tokyo 2026 Olympics took place in a hollow National Stadium. World records still fell, on the track, in the throws, in the jumps. But analysts like me had to peel a new layer off the historical data: the no-crowd layer. A fast run in front of ten thousand screaming people and a fast run in silence carry completely different psychological value, even though the stopwatch does not care.

In 2026 Japan returned to centre stage, with the World Athletics Championships at Tokyo National Stadium from 13 to 21 September. I came back to the city where I studied, rented a flat near Sendagaya station, and every morning walked twenty minutes to the track area. A championship summer compresses a nation's emotions into nine days. It also compresses the pressure on every measurement.

The Vocabulary of a Data Monk

Before anything else, I have to talk about the instrument I look through.

In football, xG answers the question of how many goals a chance is worth. In athletics, the equivalent of xG is a speed curve. I split a 100m race into four segments. Reaction time runs from the gun to the instant the foot leaves the block. The 0-30m split speaks to acceleration. The 30-60m split speaks to peak velocity. The 60-100m split speaks to speed endurance, the ability to hold velocity while the body begins to pay back its oxygen debt.

Three more variables rarely noticed outside the industry. Ground contact time, measured in milliseconds, tells you how long the foot stays on the surface per stride. Stride frequency and stride length are two quantities that always trade off against each other. And finally, vertical oscillation of the centre of mass: the less the body bounces up and down, the less energy is burned lifting the body into the air instead of driving it toward the line.

None of these appear on a scoreboard. Nobody in Sendagaya buys a ticket to watch vertical oscillation. But this is where most error is created and erased. I call them the non-scoring metrics: they never touch the finish line, yet they decide who touches it first.

Layer One: Placing a Result on a Coordinate System

An athletics result is meaningless on its own. 9.79 seconds is a bare number. Place it beside Usain Bolt's 9.58 world record from Berlin 2026, beside the Olympic record, beside the season's world lead, and it starts to speak.

My first principle: every performance must be anchored to at least three coordinates. The historical coordinate, meaning the world, Olympic and continental records. The seasonal coordinate, meaning the world lead and the athlete's own best. And the qualification coordinate, meaning the entry standard the federation publishes.

The entry standard for the men's 100m in World Championship qualifying is 10.00 seconds. It is an administrative line drawn by people, and it carries strange power: it turns a man who ran 10.01 into someone who must chase ranking points, and it turns a man who ran 9.99 into an invitee. Same afternoon, same track, two different fates separated by two hundredths of a second.

I check those three coordinates before writing a single word. Skipping that step means volunteering to work with blind data.

Layer Two: Adjusting Value — Wind, Altitude, Surface

This is the layer scoreboard readers skip, and the layer I spend the most time on.

Wind is the most clearly legislated variable. A performance is only ratified as a record if the tailwind does not exceed 2.0 metres per second. But a legal limit does not erase physical effect. A 1.9 metres-per-second tailwind in a 100m race can save an athlete roughly a tenth of a second. Over 200m, where athletes run through a bend and the wind shifts direction mid-race, the complexity is greater still.

Altitude is the second variable. Above 1,500 metres, air is thinner, drag falls, and both sprint and middle-distance events benefit. Running in Bogotá, in Nairobi, in Mexico City means running inside a physics experiment nobody writes into the minutes. Federations still ratify altitude records, which is administratively reasonable, yet the analyst must hold an invisible asterisk.

The track surface is the third variable. The Stade de France and Tokyo National Stadium both use synthetic surface systems developed by the same group of manufacturers. The stiffness, elasticity and energy return of that layer have become part of the performance itself, to the point where the industry now notes it the way it notes weather.

Temperature and humidity are the fourth, routinely underestimated. A humid Tokyo evening affects the body's thermoregulation very differently from a dry European evening, and in events from 400m upward that difference converts into seconds.

When I read a result, I read it with four maps: wind, altitude, surface and climate. Those four maps together are often worth more than the result itself.

Layer Three: Twenty Millimetres and Four Percent

Here we have to talk about what sits under the foot.

In January 2026, World Athletics published the first rulebook limiting the construction of elite racing shoes. From 30 April 2026, sole thickness in track events could not exceed 20 millimetres, and in road events could not exceed 40 millimetres. Each shoe could contain only one rigid plate embedded in the sole, usually carbon fibre. Initially the federation also required shoes to be available for retail purchase, a clause designed to block athlete-only prototypes. That clause was removed in 2026, while the thickness limit and the single-plate limit remained.

Why legislate at all? Because of data. In 2026, Wouter Hoogkamer's research group published findings showing that a shoe with a carbon plate and a high-performance elastic foam improved running economy by roughly 4 percent compared with traditional shoes. Four percent. Over a marathon, four percent equals minutes. Over a 100m, it can be the entire gap between gold and fourth place.

In October 2026, in Vienna, Eliud Kipchoge ran a marathon under two hours inside the INEOS 1:59 project, in 1 hour 59 minutes 40 seconds. That mark was not ratified as a world record, because it was run under staged conditions: a rotating team of pacers, a wind-shielding car, a chosen surface. The athletics community split in two. One half called it the breaking of a human limit. The other called it an equipment experiment. Both halves are right, and the coexistence of those halves is the subject of this article.

From 2026 onward, every distance record must be read with a question attached: was this made by muscle, by the sole, or by both? I have never answered that question fully. But one thing I know for certain: whoever does not ask it is reading half the data.

Layer Four: The Door to the Championships Is Not on the Track

There is a large blind spot in how the public follows athletics: people believe places at championships are decided on the track. Mostly true, but the most decisive part happens on paper.

The current World Championship qualification mechanism runs through two doors. The first is the entry standard: run faster than the threshold inside the designated window and you qualify. The second is the world ranking: a points system built from performance and placing at graded competitions, counting a set of best results inside a twelve-month window.

These two doors produce a kind of tactical behaviour the audience never sees. An athlete who runs 10.05 at a low-tier meet can bank more points than one who runs 10.02 at a high-tier meet but finishes fifth. Which means the competition calendar becomes a tactical variable: which meet, at what moment, against whom.

The physical cost of that tactic is real. Every elite 100m is an exposure of the body to the largest impact load the tendon-muscle system can tolerate. Racing often enough to accumulate points raises the probability of hamstring and Achilles injury. Racing too little to save the legs for the main event means losing the place. This is an optimisation problem with no general solution, only a solution specific to each athlete, each coach, each sporting nation.

In Japan, that problem is solved in a distinctive way, which I will address later.

Layer Five: The Sprint Power Map

When I redraw the men's 100m map over the past decade, I see three blocks.

The first is the United States, with a vast collegiate system producing a continuous stream of athletes. The second is the Caribbean, Jamaica in particular, where high-school championships carry competitiveness on the level of an international meet. The third is Africa, shifting from middle and long distance into sprint events as well: Botswana and Namibia have produced 200m and 100m runners who reach major finals.

Asia sits elsewhere. For years the region was described with a cliché: a lack of innate speed. That explanation is scientifically lazy. Asia's sprint problem lies mainly in grassroots participation numbers, talent identification systems and coaching philosophy, not in muscle fibre structure as many assume.

Signs of change have arrived. Abdul Hakim Sani Brown, a Japanese athlete of Ghanaian descent, lowered the Japanese national record to 9.95 seconds in 2026, a mark considered impossible for an Asian sprinter twenty years ago. A national record does not redraw the world map. But it redraws the map of aspiration for a generation of seventeen-year-olds watching television in Osaka.

What interests me more is the structure behind that number: an athlete developed inside the Japanese school system, moved to a US collegiate training environment, then returning to compete in domestic meets. That is a hybrid model, and I believe hybrid models will define the next decade of Asian athletics.

Layer Six: Teams, Training Systems and the Collective Machine

Nobody runs a 100m alone, even when only one person is on the track.

Noah Lyles trains inside a group based in Clermont, Florida, under Lance Brauman. What matters there is not the name but the architecture: a squad of athletes across different events training together, racing each other every morning, turning a training session into a miniature competition. Analysts call this controlled internal competitive pressure, and it generates measurable output without an official meet.

Japan has taken a nearly opposite path, and done so very effectively.

The corporate team system, jitsugyodan, turns athletics into a salaried profession with dormitories, doctors and rehabilitation specialists. An athlete runs for a company; the company pays a salary and receives brand value. The Hakone Ekiden, the long-distance relay between universities held every January, is the media summit of this system: an event with a television share comparable to a major football match, capable of turning a twenty-year-old student into a national face in a single morning.

This kind of system has one strength and one fatal weakness. The strength is stable resourcing, which keeps Japanese distance events formidably deep. The weakness is that it places the competition calendar above the athlete's physiological cycle: a runner can be asked to empty the tank for a corporate team and then report to the national team weeks later. In my data, this is the single most common cause of missed peak seasons.

In the meeting room, emotion asks and data answers.

Layer Seven: Biological Passports and Returned Medals

There is a data layer I am never allowed to skip, even though it moves slowly and carries little drama.

The athlete biological passport is a longitudinal monitoring tool. Instead of hunting for a specific prohibited substance in a urine sample, the system records an athlete's blood and steroid markers over years and compares each new sample against that person's own baseline. A marker jumping outside its expected range does not automatically create a sanction, but it opens an investigation.

Alongside it sits the whereabouts obligation. Athletes in the testing pool must file a precise location for one hour each day so that out-of-competition testers can find them. Three missed tests in twelve months constitutes a violation, even with no positive sample. It is one of the few rules where an administrative failure carries consequences comparable to a biological one.

For the analyst, the most important consequence of both mechanisms is the concept of reallocation: results can be rewritten years after the track has cooled. A world title can belong to the runner-up, in a ceremony held in another city, before a different crowd. In my historical data, that means every medal table is a living document, not a closed record.

I once had to rewrite four analytical pieces because a single reallocation decision was published. The lesson was simple: when I declare an athlete a champion, I must attach a timestamp. Accuracy in this sport has no present perfect tense.

Risk Is a Category, Not a Number

There is a bad habit I deliberately avoid: assigning each athlete a single risk rating.

The risk profile of a 100m runner contains at least six separate lines. The first is hamstring injury risk, tied directly to the number of maximum-velocity exposures in a season. The second is technical risk: one false start, one lane infringement, and an entire season disappears in a tenth of a second. The third is peaking risk, when the peak arrives too early at a domestic meet and nothing is left for the main event. The fourth is age risk, with career curves taking different shapes in different events. The fifth is administrative risk, involving the biological passport, whereabouts filings and eligibility. The sixth is market risk, when an athlete's commercial value collapses faster than their performances.

The most common mistake in my profession is collapsing those six lines into a single impression, usually an optimistic one. A weak analyst always writes that risk is low. An honest analyst writes that there are six risk lines, that three of them lack sufficient data for assessment, and that one of them can detonate on any given morning.

The Twelve-Month Window and Hidden Value

The transfer market has no rumours, only prices finding their way back to themselves.

Athletics has no transfers in the football sense, but it has a real market: the market for appearance fees. Major meets pay leading athletes to show up. Those fees are negotiated across a bundle of variables: personal bests, recent form, media profile, the athlete's home market, and most importantly the ability to sell tickets.

Hidden value lives here. An athlete who runs a steady 9.90 across three seasons can carry higher commercial value than one who once ran 9.83 and then spent two years injured. The market does not reward peaks; it rewards reliability. It is a rule young athletes often learn too late, after trading stability for a single explosion.

In Japan, this market carries an extra layer: nationality. An athlete with a place on the Japanese national team simultaneously becomes a domestic media asset, with advertising contracts an equally talented athlete from elsewhere cannot reach. It is one reason many mixed-heritage athletes in Japan show more patience with the domestic system, even when that system is not always optimal for their sporting development.

I like this part of the job: where value cannot be measured by a stopwatch but must be measured by the relationship between expectation and actual payment. I do not predict football; I measure the distance between expectation and goals. The same logic applies to an athlete stepping onto the track: I do not predict the time, I measure the distance between what the stands expect and the number that finally appears on the board.

Correlation Is Not Causation

Here I have to step outside my comfort zone.

Everything I have presented is correlation. Records rose at the same time soles thickened. Performances improved at the same time tracks were redesigned. Hamstring injuries increased at the same time competition density increased. None of those items proves causation, and I will not pretend they do.

The problem is that every variable I listed evolved together. Shoes, surfaces, nutrition, recovery, measurement tools, the importance of championships, prize money — all changed within the same decade. Isolating the effect of any single factor inside such a system is impossible to do cleanly. The analyst must accept living with that ambiguity rather than pretending the cause has been isolated.

Three kinds of noise require constant self-warning.

The first is selection noise. Record breakers are drawn from an enormous population while also enjoying the best possible conditions. Their high performance does not prove that good conditions produce high performance, because we never see the thousands of other athletes in similar conditions who failed.

The second is description noise. We measure what is easy to measure and ignore what is hard. Reaction time is easy. Confidence before a crowd of ten thousand is not. Both affect the final result.

The third is time noise. A mark run at twenty-one and an equivalent mark run at thirty do not carry the same predictive value. Every event has its own career curve shape and its own peak age. Ignoring the age axis means taking half your data and burning it.

Criticism belongs to the same category of data. Every laugh is an unlabelled column.

The Accidental Laboratory

Home advantage is a hypothesis, and the pandemic was an accidental experiment.

I return to this because it matters more than a personal anecdote. With empty stands, a whole set of variables vanishes from the equation: the screaming, the pressure of a home crowd, the psychological inertia of officials, and the travel burden on the visiting side. Athletics benefited from the absence of crowds in one very specific way: spectators could not disturb the session, could not shout while an athlete was recovering breath, could not shake a timing system.

But athletics also lost the most important thing a stadium provides: emotional acceleration. A relay runner inside a packed stadium can run the final leg a tenth of a second faster than the same athlete in training. That speed does not come from muscle; it comes from the central nervous system.

When I write about results from the no-crowd period, I always attach a note: this sample does not represent normal football, nor normal athletics. It represents itself. Anyone applying a no-crowd model to a season with crowds is performing an extrapolation, and extrapolation is the most dangerous work in my profession.

Even there, the data speaks. When data speaks, laughter is only noise.

Signals for the Next Cycle

If I had to give a short list of what I will track next season, I would not start with times.

First, squad depth. A nation with three athletes consistently under 10 seconds has a better foundation than a nation with one 9.85 star and everyone else above 10.30. Depth is a systemic indicator, and it predicts far better than individual peaks.

Second, competition density among leading athletes during the points-gathering phase. If a sprinter races more than ten times inside a twelve-month window, their hamstring injury probability rises substantially. This is one of the few regularities I am willing to call a rule.

Third, national records in Asia matter to me more than world records. A world record belongs to an exceptional individual. A national record in a country opening up its sprinting belongs to a system opening up.

Fourth, administrative decisions. Every time World Athletics amends shoe rules, entry standards or the calendar, it rewrites the coordinate system all of us use to read performances. A 5-millimetre change in sole thickness can rewrite hundreds of pages of historical data faster than any athlete alive.

In the end I remain a person who works with data. But I have been in this trade long enough to know that the most beautiful data is the data that forces me to say I do not know. The unexplained part is not a hole in my model. It is the part that keeps me sitting under stadium lights at three in the morning, notebook open with its seven columns, waiting for the scoreboard to answer.

The fastest runner is not necessarily the one who best understands why they are fast. The most accurate writer is not necessarily the one with the most complete data. Both share exactly one thing: a gap between what is known and what must still be waited for.

This analysis draws on public World Athletics data, official competition results, and the author's own field notes. It is provided for sports-information purposes and does not constitute betting advice.