AIMS, Mangrove Forest and the 42 km Distance: The Data Problem Behind HDBank Green Marathon 2026

23:40 Osaka time. A four-page PDF sits in my inbox: the route map of HDBank Green Marathon 2026 running through Can Gio district, Ho Chi Minh City. I zoom into the stretch from km 28 to km 34. Red earth path along mangrove forest, canals on both sides, almost no shade on the road surface. In the bottom corner of the map, a small line of text: the route has been measured and certified to AIMS standards.

That line carries more weight than the entire rest of the press release. Vietnam hosts hundreds of road races every year. Far fewer hold international measurement certification. And even fewer dare place a 42.195 km distance on the same road system as a 5 km children's category. That is the point I want to dissect.

Context: a race designed around a distribution curve

HDBank Green Marathon 2026 is organised by HDBank in partnership with Unique Company and the Ho Chi Minh City Athletics Federation, under the technical guidance of the city's Department of Culture and Sports. This is the fifth season. The distance system covers 5 km, 10 km, 21.097 km and 42.195 km, plus children's and team categories. The expected date is September 2026, at Can Gio — a mangrove biosphere reserve recognised by UNESCO.

The race structure says a great deal before any result exists. A race with four distances spanning 5 km to marathon is not aimed at elite athletes. It targets the widest possible distribution: beginners, families, recreational running clubs, and a layer of semi-competitive runners who use the marathon as a personal yardstick. Reading that allocation, I can already predict the shape of the data the organisers will collect on race day: plenty of volume data, very little peak-speed data.

Based on my experience tracking and logging community races in both Vietnam and Japan, the four-distance model closely resembles the structure of Japanese local races, where the marathon serves as the flagship distance and shorter events act as entry points. The difference lies in density. Japan holds thousands of races a year; Vietnam holds hundreds. Lower density means each race has to carry more functions, and that puts pressure on measurement quality.

In 2026, as a journalism student in Osaka unable to reach Yodoko Sakura Stadium, I built my own dataset from old match footage: 1,240 pressing sequences by Cerezo Osaka, logged to derive PPDA. I predicted the team would decline once the league resumed, since losing home advantage would hurt their pressing. They finished fourth, below my second-place projection. I had to add a crowd-effect variable to the model. That lesson followed me into writing about running: a model without environmental variables is an incomplete model.

AIMS certification: the most valuable data point in the release

To recreational runners, AIMS certification often looks like administrative detail. To serious runners, it is the whole story. AIMS is the association of international marathons and distance races, setting measurement standards for road courses. When a route is certified, the distance has been measured by standard method, with checkpoints and a documented margin of error. A personal best set on an uncertified course has sentimental value only. Set on a certified course, it becomes data that can be cross-referenced.

This point is rarely discussed. A GPS watch typically reports a distance one to three percent longer than reality, depending on tree cover, buildings and satellite signal quality. On a route running through Can Gio's mangrove forest, that error could be larger. Without certified kilometre markers, every runner finishes with a different number, and the results table becomes noise. AIMS certification is how that gap is closed before the race even starts.

Why 42 km is the flagship distance, not 5 km

In marketing terms, 5 km is the easiest distance to sell: low barrier, fast registration, good photos. But in a race's data structure, 42.195 km is where information begins to exist. A 5 km runner generates a single data point. A marathon runner generates 42, plus data on how speed decays over time — what analysts call second-half pace drop.

Second-half pace drop is the most accurate indicator of whether a runner has prepared enough. A well-prepared marathoner usually keeps the gap between first and second half within a few percent. An underprepared one collapses somewhere between km 30 and km 35. With thousands of participants, the organisers hold a dataset on the endurance thresholds of Vietnam's running community — something never published anywhere.

Seen as a distribution, the 5 km, 10 km, 21 km, 42 km structure resembles an exponential decay curve. Registration numbers fall as distance rises, but the amount of data generated per person rises in the opposite direction. This is why I always read a distance breakdown before reading any other part of a release.

Terrain variables: what never appears in the promotional deck

Can Gio is a mangrove biosphere reserve about 50 km southeast of central Ho Chi Minh City. The route passes through forest, along rivers and along the coast. This terrain is beautiful in photos and complicated in calculation. Three variables need logging: sandy or soft ground on certain sections, exposed tree roots on forest stretches, and the high humidity characteristic of mangrove areas.

Humidity is the most underrated variable in tropical races. At the same temperature, high humidity reduces the body's heat dissipation efficiency, pushing heart rate higher at the same running speed. For marathoners, that deviation accumulates hour by hour and shows up most clearly in the second half. Without hourly temperature and humidity data, any post-race analysis lacks a denominator.

If I were designing the data model for this race, I would request three things before race day: temperature and humidity logged every 30 minutes at three points on the course, certified kilometre markers per AIMS, and speed data split into 5 km segments. Those three are enough to turn a community race into a usable dataset for the following season.

The fifth season and the value of repeated data

This is the race's fifth season. In data analysis, a fifth season carries its own meaning: an event repeated five times begins to generate a comparable series. A first season gives a point. A fifth season gives a line. For community races, that series answers questions a single season cannot: how many participants return in year two, in which direction the average 42 km finishing time moves, how the dropout rate shifts.

According to information from previous seasons, the race has planted more than 2,000 trees and organised beach clean-up activities. This is a specific, verifiable fact, and it matters more than a slogan. Planted trees are a countable number. Beach clean-ups are a measurable weight. A race that commits to the environment and publishes a concrete figure gives that commitment a denominator.

The digital component: extending the race beyond race day

A notable element of this year's design is the Green Marathon online component, alongside integration into HDBank's app ecosystem. In data terms, this is a move to widen the observation window. A traditional race exists for roughly six hours on race day. A race with a digital component exists all year, generating data on training habits rather than only on race results.

HDBank has more than 36 years of involvement in sport, from football to running. A bank integrating a race into its own app creates a stream of data on customers' physical behaviour. This is a point to monitor, not a point to praise. In Japan, the model has been common for years: running apps linked to bank cards, points accrued per kilometre, and runners paying with personal data for the rewards they receive.

The contrarian angle: the biggest risk is not the absence of elite athletes

The release names no elite athletes. The conventional reaction is to classify the race as low value. I think that reading points in the wrong direction. The absence of a few elite athletes reduces media value for a few days. The absence of measurement data damages the race's value for the entire running community behind it.

If forced to argue the opposite case, I would put it this way: a community race does not need elite athletes because its goal is not performance but access. True. But access without measurement is just event management, not sport building. This is the blind spot of Vietnam's community race model: organisational capacity is rising fast, data-logging capacity is rising far more slowly.

Data does not create stories; it strips the stories off other people. A race that publishes full split data will reveal what no press release dares to say: what percentage of 42 km entrants actually finish, and where they collapse.

A comparison from the transfer market

There is a structure in football that I have always seen mirrored in community races: the loan deal with an obligation to buy. A small club develops a player, pays wages, absorbs injury risk, then hands the semi-finished product to a big club at a pre-agreed price. Vietnamese football has imported this model wholesale. In recreational running, the same structure appears in a different form: runners pay for their own training, their own entry fees, their own injury risk, and what they generate is transferred to brands as behavioural data and media imagery.

There is nothing wrong with that. It just needs to be named accurately. When a race talks about community, the accompanying data question is: what does that community get back beyond a medal. The best possible answer is open data — detailed results, split data, annual environmental impact reports. That is what Vietnam's community races lack far more than funding.

A note on cultural difference is also needed. Southeast Asian runners register for races with different motives from Japanese runners. In Japan, entering a marathon is an act of personal record accumulation, logged meticulously. In Vietnam, entering a race is often tied to photos, to friend groups, to experience. If organisers apply Japanese logging standards directly to a Vietnamese community, the completion rate for data fields will be very low. The right approach is to design automatic data capture that does not depend on whether a runner wants to share.

I collect errors, classify them, and then I know where the team is heading. With a race course the method is the same: collect pace collapses, classify them by segment, and you know where the water stations need to be.

Signals to track for the next cycle

Three indicators I will log after September 2026. First, whether the 42.195 km results come with split data for every 5 km. Second, the finisher-to-entrant ratio in the longest distance — that number reflects the community's actual preparation level more honestly than any leaderboard. Third, whether the environmental impact report includes tree-planting figures and collected waste tonnage.

An empty stadium, yet the numbers still carry noise. A community race does not need a record to have value. It needs to be measured. And every probability conceals a shock — I only make sure it does not repeat in the sixth season.