Vietnam's Track: Seven Years of Data and the Structure Behind the Medals

May 2026, Morodok Techo Stadium, Phnom Penh. I sat in the technical area, notebook in my left hand, a spreadsheet open on my laptop. A young Vietnamese athlete had just finished the women's 800m heats. She did not make the final.

No one in the stands remembered her name that night. But I wrote it down: 2:09, four seconds slower than the personal best she had set three weeks earlier at the national championships.

Four seconds over 800m is not a form slump. A twenty-year-old does not lose four seconds in three weeks because her legs went flat. Four seconds is the fingerprint of something a results sheet never prints: competition load.

On the day football stopped, I started counting strides again. In 2026, when the pandemic wiped out the global calendar, I spent four months with five seasons of V.League data and three major European leagues. I assumed I would go back to football afterwards. Instead I realised I was holding a toolkit that could be applied to any sport with a stopwatch. Athletics is the most honest sport for data, because it does not allow arguments about outcomes. Only the explanation leaves room for error.

And in the explanation, we are getting quite a lot wrong.

Part 2 — Methodology context

Vietnamese athletics has a data problem that football does not have: it is barely recorded systematically.

In the V.League I can access event-level data by minute, PPDA, final-third recoveries. In domestic athletics, what I have is mostly printed result sheets, photos from organisers, low-bitrate livestreams, and my own notebook. No wind column. No reaction times. No 200m splits.

Missing data does not mean analysis is impossible. It means you build the framework first and pour the numbers in afterwards.

Since 2026 I have built two indices for Vietnamese track.

The first is the Competition Load Index (CLI). I count official appearances over twelve months and divide by the number of genuine recovery weeks between them. My warning threshold is 1.4. Above that for two consecutive cycles, the probability of injury or a performance drop at the next outing rises clearly in my dataset.

The second is the Medal Concentration Index (MCI). I take a delegation's gold medals at a Games and divide by the number of athletes who actually won them. An MCI of 1 means every athlete won exactly one gold. An MCI of 3 means the average winner took three.

My sample covers five SEA Games from 2026 to 2026, four national championships per year, plus national youth data. In total I cross-referenced more than 2,300 individual competitive appearances.

I do not publish this model to predict who wins. I publish it to answer a different question: what structure is producing our results, and is that structure durable.

Part 3 — The evidence chain

First block: our medals are more concentrated than we think.

Start with a number that is easy to verify. At the 2026 SEA Games in Phnom Penh, Nguyen Thi Oanh won three individual golds in the 1500m, 3000m steeplechase and 5000m. She ran the 1500m and the steeplechase on the same day, roughly two hours apart.

That is one of the most impressive physical performances Southeast Asian athletics has seen. But if you look at it through the Medal Concentration Index, you see something else.

Vietnam's MCI in women's middle and long distance at that Games sat at a very high level. When one athlete delivers three golds in four days, the system is concentrating risk at a single point.

I checked this across five Games. Vietnam's women's middle and long distance group consistently posted an MCI of 2.5 or above. The men's and women's sprint groups sat close to 1.0 — each gold attached to a distinct face, with very little repetition across Games.

Those two numbers tell opposite stories. The distance group has depth but is thin at the top: a handful of athletes carry almost the entire output. The sprint group has more faces but lacks anyone who survives two Games cycles.

Both are risks. One is simply louder than the other.

Second block: competition load is the most neglected variable.

Back to the 800m athlete I noted in Phnom Penh. I traced her schedule across the fourteen months before the Games. She made 21 official appearances, six of them across two rounds of the same meet within 72 hours. Her CLI in that window was around 1.8.

Across my 2,300 appearances, athletes with a CLI above 1.4 for two consecutive cycles performed 1.9 per cent below their personal best at the next outing on average. Over 800m and 1500m, that equals two to three seconds. Exactly the gap between a heat and a final.

This is where I want to slow down, because it sits at the centre of everything.

Vietnamese athletics has an organising paradox. Because there are few official domestic meets, each one is forced to serve several goals at once: national team selection, form assessment, and the host province's media requirements. The result is that an athlete may have to run three events across two days, in three different rounds, to satisfy all of them.

In football we call that a congested calendar and we have data to manage it. In athletics we call it competitive spirit.

Injuries are random. Schedules are not.

Third block: the talent pipeline is narrowing exactly where we look least.

I tracked the number of under-20 athletes meeting the standard for the national championships across three groups: sprints, middle distance, and jumps. Over roughly seven years, the sprint figure fell slightly but held steady. Jumps stayed flat. Middle and long distance grew in headcount but skewed sharply towards three training centres.

In other words, we have more middle-distance runners, and they come from fewer places.

That is a two-sided signal. Professionally, concentrating resources into three centres raises the floor of coaching quality. Structurally, it means that if one centre loses coaching staff, we lose roughly a third of the pipeline for a cycle.

Hai Phong taught me this: the star is not on the shirt, it is in the index.

In 2026, while working as a data consultant for the football club in Hai Phong, I found a young midfielder named Vu Minh Hieu with an average PPDA of 6.8 — the best in the club's academy. He pressed exceptionally well but drew no attention from the coaching staff because of his modest frame. I brought the numbers into the meeting room and asked for him to be given a chance. In round 17 that season, Minh Hieu made 14 recoveries, provided one assist, and the team won 2-1.

The lesson was not that data is always right. The lesson was that undervalued athletes are usually undervalued because people read the wrong index, not because they lack ability.

On the track, the most misread index is absolute performance. A nineteen-year-old running 4:05 for 1500m can be overlooked because she is slower than someone running 3:58. But if the 3:58 athlete has trained professionally for four years and the 4:05 athlete for eight months, what you need to measure is the slope of the progression curve, not its endpoint.

I do not have a long enough series to draw firm conclusions about that slope at national level. I have just enough to say we have never tried to measure it.

Part 4 — The contrarian angle

By now my hypothesis sounds clear: Vietnamese athletics results are over-concentrated, competition loads are too high, and the talent pipeline is too narrow. Three problems, one shared cause: a lack of planning.

The counter-hypothesis deserves serious consideration: perhaps that concentration is a rational choice, not a flaw.

The argument is strong. For a sport system with a limited budget, pouring resources into three centres and a targeted group of athletes is the way to maximise medals per unit spent. That is the logic of most small sporting nations. Singapore does it in swimming. Jamaica does it in sprinting. Concentration is not a disease; it is a strategy.

I tested this with a simple comparison. If concentration were the effective strategy, delegations with high MCI should show correspondingly higher total gold counts. In my five-Games sample that correlation exists but is very weak, and it disappears once I remove the case of a single delegation.

In other words: concentration correlates with results, but I do not have enough evidence to say it causes them.

And this is where I have to correct myself.

Over the past four years I have repeatedly presented the concentration index as a warning. Looking back, I presented it too close to a causal claim. My data shows a distribution pattern. It does not show what produced that pattern.

There are at least three gaps in this dataset that I must state plainly.

First, I lack official injury data. I infer risk from performance drops, but a drop can come from a training cycle, from competitive psychology, or from an athlete deliberately running within herself in a heat. I cannot separate those three causes with the numbers I have.

Second, five Games is a small sample. Five data points cannot separate signal from noise, particularly when each Games has different weather, scheduling and opposition.

Third, I have no data on athletes who quit. My dataset only records those still competing. This is a classic selection bias, and it biases me towards rating the system better than it is.

A season is a confession of tactics. Seven years of data is a longer confession, and most of it is about what I have not measured.

Part 5 — What to watch in the next cycle

The ball only rolls one way, but data can look every way.

If I had to pick one signal to track in the coming cycle, I would not pick gold medals. I would pick two others.

The first is the average Competition Load Index of the targeted athlete group in the twelve months before a Games. If that figure falls below 1.3, it signals that somebody has started managing the calendar with data rather than with the organiser's schedule. If it stays at 1.7 or above, we will see those four-second drops in heats again, and again nobody will record them.

The second is the number of under-20 athletes from outside the three major training centres who meet the national championship standard. This measures the health of the pipeline, not performance. It never appears on a medal table. But it decides the medal table four years out.

People call me a data monk. A monk does not need a cathedral, only the truth.

The truth I hold is an incomplete one. I have enough data to say that the way we record our own track is hiding the most important questions: who is racing too much, who is improving faster than their results suggest, and who quit before we wrote the name down.

The 800m athlete in Phnom Penh that night did not make the final. She did not appear in any delegation summary either. She exists only in my notebook and in a spreadsheet nobody reads.

My job is to get that spreadsheet read.

Our job is to decide whether we want to count medals, or to count the strides that lead to them.