The Athletics Data Checklist: Wind, Altitude, Shoes and the Part That Cannot Be Explained

THE ATHLETICS DATA CHECKLIST: WIND, ALTITUDE, SHOES AND THE PART THAT CANNOT BE EXPLAINED

On the night of 1 August 2026, the Olympic Stadium in Tokyo was lit with not a single spectator in the stands. In the men's 100m semi-finals, Su Bingtian — 31 years old, Chinese — ran 9.83 seconds. It was the Asian record, and the first time a man born in Asia reached an Olympic 100m final in the electronic era. Two hours later he ran again, finished sixth, and clocked 9.98.

The scoreboard printed 9.83. The scoreboard did not print the wind.

The anemometer reading in his lane that night was +0.9 metres per second — legal, comfortably under the +2.0 ceiling, but it turns a bare time into a conditional fact. Without that +0.9 line, every later comparison shifts. And the 0.15-second gap between semi-final and final, which most reports called a mental collapse, sits mostly in the recovery window: two hours between two maximal efforts, at 31, after a season compressed by a pandemic.

That night carried two stories. One was told a thousand times, with 9.83 at the centre. The other was told only where someone bothered to read the conditions column. The second story is better.

1. WHY I READ A TRACK WITH A CHECKLIST

I was a track athlete before I became a sports data analyst. In 2026 I edited a running magazine. In 2026, a second-year student in Tokyo, I ran a blog analysing the World Cup with data. Before Germany played South Korea, I pointed out that Germany's expected goals stood at 2.1 against South Korea's 0.6, but South Korea had recorded 121 sprints and a second-half PPDA of 7.8 — extreme pressing intensity. I predicted Germany could go out. A male commentator mocked me online. South Korea won 2-0, Germany went home, and the blog was shared thousands of times overnight.

When the data speaks, the laughter is nothing but noise.

But that was the lesson of being 20. The lesson of being 28 is more complicated. In 2026 I joined a betting analytics firm in Tokyo. Before the Euro final between Italy and England I presented the case that Italy's average PPDA of 8.9 — the most aggressive pressing in the tournament — against England's 11.4 meant Italy would control the game. A colleague laughed that a Japanese woman only reads numbers and does not understand Wembley psychology. I put my hand on the table, showed thirty matches of data and said the numbers do not lie. Italy won on penalties. In the meeting room, emotion asks and data answers.

What I carried from football into athletics is not PPDA. It is a working habit: before reading any metric, take inventory of what you have and what you are missing. Football leaves hundreds of thousands of data points per match, and my job is to pick three that describe it truthfully. Athletics is the reverse: one race leaves exactly one number, and my job is to find everything that number does not say.

I call it the data checklist. A track and field mark, read properly, needs seven layers. First, conditions: wind, altitude, surface, temperature, shoe type, time of day, number of prior rounds. Second, athlete condition: the year-by-year personal best curve, current-season form, age and where the athlete sits on the age curve, injury history. Third, the qualification mechanism: entry standard or world ranking, the three-per-country cap, national selection rules. Fourth, the event landscape: who leads, how deep the challenger group is, the age structure of the leading group. Fifth, competition rules and anti-doping. Sixth, the training system and coaching staff. Seventh, risk — injury, withdrawal, sanction, technical error.

What the checklist taught me is an uncomfortable discipline: when the input data is empty, the correct answer is not to guess until the gap is filled. The correct answer is to state plainly that there is not enough information to assess. It sounds simple. In an industry that needs a piece every week, it is the hardest sentence to say.

1. WIND: THE FIRST VARIABLE AND THE MOST OFTEN OMITTED

World Athletics recognises a mark as a record in sprints and horizontal jumps only if the tailwind does not exceed +2.0 m/s. Above that, the mark still counts in the results, still decides the race, but never enters the record book. For the athlete, that is a suspended sentence: the fastest run of a life, and nothing to show for it.

For the reader, it is the single most important fact cut from the headline.

Aerodynamic models typically estimate that over 100m each +1.0 m/s of tailwind is worth roughly 0.05 to 0.06 seconds. That is an estimate, not a law, because the effect depends on height, stride length, posture and baseline speed. But it is enough to show that the difference between a run in +1.9 and the same athlete in -1.0 is roughly a sixth of a second. Over 100m, that is the gap between a medal and an exit.

Three examples show where the wind column sits inside big marks. On 21 August 2026 in Eugene, Elaine Thompson-Herah ran 10.54 with +0.6. On 1 August 2026 in Tokyo, +0.9 sat behind Su Bingtian's 9.83. On 4 August 2026 in Paris, Noah Lyles won the 100m final in 9.79 with +1.0, in a race where second place was separated by thousandths of a second. Every thousandth was read that night. The wind line was not.

In long jump and triple jump the wind effect is far larger than in sprints, and every conversion table used in ranking systems is a constructed estimate. Which means: never compare two marks without comparing the wind column. If a report gives you one number, it has removed half the evidence.

1. ALTITUDE: THE GIFT AND THE PENALTY

On 14 October 2026 in Mexico City, Jim Hines ran 9.95 — the first electronically timed sub-10 — and the record stood for fifteen years.

Mexico City sits above 2,200 metres. At that altitude the air is thinner, drag is lower, and a sprinter or horizontal jumper receives a real bonus. Bogotá, at roughly 2,640 metres, offers more.

But altitude does not flatter every event. For middle and long distance it is a sentence. The partial pressure of oxygen falls, the body cannot supply enough oxygen for sustained power, and a 1,500m runner is slower at the same effort. A 3:30 at altitude is worth far more than the same time at sea level.

The direction of the correction flips by event group. That is the detail most reports skip, and the reason altitude belongs to the conditions layer rather than a footnote.

Some record tables annotate marks set at significant altitude. That annotation is an admission that the number does not stand alone. The reverse also holds: when a distance runner moves down to sea level and goes much faster, it does not automatically mean improvement. Read the conditions and the question answers itself.

1. SHOES, SURFACES AND A DIVIDEND THAT BELONGS TO NOBODY

Over the past five years the biggest gains in track and field have not come from lungs. They have come from midsoles.

Carbon plates and super-foams appeared on track and road, delivering a measurable advantage. World Athletics capped sole thickness, required competition shoes to be available to the public for a set period, and restricted prototypes. That is a governing body admitting a technology race exists and needs limits.

Surfaces belong to the same group. When Hayward Field reopened in 2026, national and area records fell across the 2026 World Championships held there. Part of that was the athletes. Part was the new track and new rubber. The two cannot be fully separated, but naming only one tells half the story.

The clearest example that a mark is a bundle of conditions is Eliud Kipchoge's 1:59:40 in Vienna on 12 October 2026. It was never ratified as a world record — not because it was not fast enough, but because it was not an open race under standard conditions: rotating pacers, a drinks car, an organised support team, a prototype shoe. World Athletics acknowledged Kipchoge ran 1:59:40. It also said it was not a record.

By the same logic, the pacing lights used in Berlin road races are a legal tool in some events, illegal in others, and always part of the mark. Kipchoge's 2:01:09 in Berlin on 25 September 2026 came in the best conditions a marathon can offer, including a laser line on the road.

When a mark improves, I split it into four parts: athlete, shoe, surface, natural conditions. Exact ratios cannot be assigned. All four must be named.

1. THE PERSONAL BEST CURVE AND THE PATIENCE TEST

One test in my framework only runs when year-by-year data exists: the personal best curve.

Take an athlete's best mark in each year, plot it. Most curves descend steadily and then flatten into a plateau. A single-year jump several times larger than that athlete's own average annual gain deserves a question.

To be clear: asking is not accusing. It is a screen, not a verdict.

Su Bingtian illustrates both sides of it. He ran 9.83 at 31, an age when almost every sprinter is past peak. On the curve, it is an anomalous point. In coaching context, it has an explanation: a technical rebuild of the start, a new training group, a rebuilt physical base. The screen asks. Context answers. Only when context fails do other tools come in — the biological passport, whereabouts failures, long-term sample storage, and the slow business of reallocating medals years later.

Medal reallocation is the coldest part of this sport. Olympic medals have changed hands more than a decade after the race. The runner-up who finally receives gold is already in another phase of life.

The screen has a mirror. When an athlete makes no progress and suddenly runs faster, the question is not the body. It is the conditions: wind, altitude, surface, shoe. Bad analysis almost always fails on the second direction.

There is an ethical risk here. Someone holding data and asking questions about an athlete can harm that athlete with a correct question asked in the wrong place. I learned that at 20, when a comment reduced me to my gender.

1. QUALIFICATION: TWO DOORS AND THE PENALTY OF CROWDING

A place at the Olympics or World Championships comes through two doors: the entry standard, and the world ranking built on points from designated meets weighted by competition tier and placing. Most places flow through these two channels, with the rest reserved for universality places and relay teams.

The first consequence is the calendar. Ranking points require racing often, in the right meets, at the right time. That is an invisible tax: a denser schedule, thinner peaks, and an athlete who may arrive at the main event already spent. Federations with travel and medical budgets profit from this system. Those without do not.

The second consequence is heavier: a maximum of three athletes per country per event.

Where a country has depth, the national trial is harder than the global final. At the 2026 US trials for Paris, Christian Coleman ran 9.93 in the 100m final and did not earn an individual place. In most countries that time is a final ticket. In the United States it is a relay seat.

I call it the fourth-place risk. It lives in the structure of the meet, and it exists before anyone gets into the blocks.

The US model adds another layer: one race decides everything. A reigning world champion can miss the Games after one bad evening. That is structure, not failure. In Japan, where I live and work, selection leans on designated meets plus in-season marks — which turns selection into a longer, more public, more psychologically costly argument.

Inside all of it sits a calculation no broadcast prints: the cost of the relay. A sprinter doing four relay legs and three individual rounds in a ten-day meet is signing a badly skewed physical contract. By the individual final, those legs may have already paid the bill.

Based on my experience watching meets in Tokyo and Osaka across many seasons, I record three things before I record the mark: the recovery window between rounds, the number of starts in that week, and who set the pace. In most cases those three explain the gap between heat and final better than any talk of nerve.

1. THE POWER MAP AND THE SIGNS OF A GENERATIONAL HANDOVER

Reading an event without knowing who leads is like reading a scoreboard without knowing the rules. In the sprints, the map changed hands after Usain Bolt. An era of one ruler gave way to a cluster: Jamaica and the United States producing a group capable of 9.7x and 9.8x, none far enough ahead to be called dominant. In a cluster event, the value of consistency falls and the value of a peak rises, because the title is decided in one race.

In the middle distances the balance shifted too. Kenya and Ethiopia still produce most distance athletes, but at 1,500m and 5,000m Europe and the United States reclaimed the centre. In the Paris 2026 men's 1,500m final on 6 August, Cole Hocker won in 3:27.65, an Olympic record, while Jakob Ingebrigtsen finished fourth. When the man with the season's best time misses the podium, analysis built only on personal bests has already failed.

In the throws, the United States retains the greatest depth. Ryan Crouser put the shot 23.37 metres in Eugene on 18 June 2026 and kept pushing the event's limit. In the pole vault, Armand Duplantis built a ladder of world records almost annually: 6.20 indoors in 2026, 6.26 outdoors in August 2026. He does not dominate because rivals weakened; he is redefining the standard.

In the women's 400m hurdles, Sydney McLaughlin-Levrone turned a technical event into a sprint. She ran 50.68 in Eugene on 22 July 2026 and 50.37 in Paris on 8 August 2026. The 50-second barrier, deemed impossible a decade ago, is now a matter of timing.

In race walking and women's throws, China and Japan are two strong systems with different organisational logic. In the marathon, after Kipchoge's 2:01:09 in Berlin and Kelvin Kiptum's 2:00:35 in Chicago on 8 October 2026, the two-hour barrier became an engineering question. Kiptum's death on 11 February 2026 removed its best candidate. That is the clearest reminder that data does not capture everything.

The generational handover shows in one easily checked place: the age structure of the season's top ten. When seven of ten are past thirty, the event is about to change hands. When seven of ten are under twenty-five, it is entering a boom, and old records should be re-read.

1. HOME ADVANTAGE IS A HYPOTHESIS: THE TOKYO EXPERIMENT

In 2026, when football returned to empty stadiums, I collected the first twenty-six matches and found home advantage falling from an average of 0.44 goals per match to 0.15. I built a no-crowd betting model, backed underpriced away teams, and won seventeen of twenty. An empty summer taught me that an empty seat is itself a variable.

Athletics gave me a better experiment still. The Tokyo 2026 Olympics were held almost entirely without spectators. If home advantage came from noise, the host nation should have lost most of it. The opposite happened: Japan's gold medals rose from twelve in Rio 2026 to twenty-seven in Tokyo 2026, mostly in scored and combat sports.

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

If noise is only one channel, others remained live: host quotas, four-year investment, medical and recovery systems, familiarity with venue, time zone, food, referee language, and specific knowledge of how scoring works in each sport. Those channels do not vanish when the stands empty. They strengthen, because they are structure rather than sound.

For anyone reading a mark, the lesson is direct: when a home athlete runs faster than expected, ask which channel of advantage is switched on. When a home athlete runs slower at a crowdless event, do not call it weakness. Perhaps the only channel left had simply been muted.

1. THE CONTRARIAN ANGLE: CORRELATION IS NOT CAUSATION

Here is where data analysis lies to itself.

A fast track does not mean the athlete improved. A tailwind does not mean a new level. One final does not mean a season's baseline. A new shoe is not a physiological leap. And every wind or altitude conversion model is an assumption stacked on an assumption: useful for comparison, not objective truth.

This is why I never use the word prediction. I do not guess athletics; I measure the distance between expectation and the track. That distance always contains an unexplained remainder, and the remainder is where the next model is written.

A classic example of how a one-race model has a fat tail: at the 2026 World Championships in Daegu, the zero-false-start rule was in force. Usain Bolt, the surest thing at the meet, false-started in the 100m final and was gone in an instant. Every probability model built before that night became waste paper. A competition rule can erase years of data in one second — which is why the rules layer sits in my checklist rather than an appendix.

The reverse caution matters too. A jump in performance does not automatically mean doping, and a good season from a young athlete proves only that the athlete is young. The screen always returns a question. The answer must come from biological data, testing records and coaching context — not from the analyst's feeling.

And here I correct myself. Every laugh is an unlabelled data column. When the data speaks, the laughter is nothing but noise — but only after it has been labelled. A meeting-room laugh, a sprinter's tension before the gun, the weight a nation puts on a 22-year-old: that is data, merely un-encoded. The bad analyst calls emotion noise and discards it. The good one turns it into a variable and finds a way to measure it.

1. SIGNALS FOR THE NEXT CYCLE

Four lines go into my notebook.

First, the 9.7x cluster in the men's 100m. An event without a ruler produces races decided by reaction time and recovery windows. There, the valuable metric is not the season's best mark but the standard deviation of the runs. Consistency will be worth more than a peak.

Second, the 50-second barrier in the women's 400m hurdles. A barrier about to fall makes every historical comparison obsolete, and models built on history will err in the same direction at once.

Third, the shoe debate. As rules tighten on sole thickness and require public availability, the advantage shifts from prototype technology to access to technology. Wealthy federations keep benefiting, but differently.

Fourth, the qualification system. As world ranking decides more places, athlete calendars will be shaped by the need to accumulate points, and major meets will become destinations for the already exhausted. This is a variable no scoreboard shows, and it decides who stands on the start line.

One question stays open, and I think it will shape how we read athletics for years: when will broadcasters finally put wind, altitude and time of day in the same frame as the time?

Until then, every mark will appear on screen as a complete truth. And the reader, as always, will have to go looking for the missing part.

I choose to go looking. Not because it is more exciting than the number. Because it is the number, in its complete form.