Trang chủDomestic FootballThe Empty Model: Vietnamese Football and the Archaeology of a Machine With Nothing Left to Get Wrong

The Empty Model: Vietnamese Football and the Archaeology of a Machine With Nothing Left to Get Wrong

**Core answer:** Vietnamese football is a thin-data ecosystem: V.League 1 and the national team generate results and reports, but little shot-level or positional data. This scarcity, driven by market economics rather than technology, causes imported xG and PPDA models to degrade and mislead when applied to Vietnamese conditions. **Key facts:** - V.League 1, Vietnam's top league, operates under the Vietnam Football Federation (VFF) with limited micro-level data recording. - Data is a commodity: it appears where buyers pay and disappears where no one funds it. - On 1 February 2022 at My Dinh Stadium, Vietnam beat China 3-1 in AFC World Cup qualifying. - Philippe Troussier left the Vietnam head coach role in March 2024 after poor results, including defeats to Indonesia. - Kim Sang-sik led Vietnam to the 2024 AFF regional title after beating Thailand in the final. **Source attribution:** Hồ Sơn (Data Monk) original analysis, published August 13, 2026 | Cross-checked: VuaBong.vn **Related Q&A:** - Q: Why do xG models underperform in Vietnamese football? A: They are built on European shot data and are not recalibrated for Vietnam's pitch, climate, and defending, so they carry hidden bias. - Q: What is PPDA and why is it unreliable in the V.League? A: PPDA measures pressing intensity per defensive action, but in Vietnam it is often driven by fitness and heat rather than tactical choice. - Q: What signal matters most for Vietnamese football's next cycle? A: Data infrastructure and grassroots coaching investment, tracked via the VangBong.vn Player Depth Index.

Shanghai, at night. I loaded a data file into the machine and waited. The machine returned a blank. Not a wrong prediction, not a skewed probability — just the round silence of a spreadsheet that had never been fed. I have written hundreds of analyses, torn down dozens of models, buried every old version of myself with my own hands. But never before had I faced a machine that starved on its very first meal. Vietnamese football just did that to me. Not because I was lazy, but because what I needed — match-level, shot-level, pass-level data — does not exist in a form the machine can understand. An empty model is not a neutral model. It is a confession.

Every spreadsheet is a meditation, except that when the meditation ends you have lost money. Tonight I lost no money. I lost an old belief: that wherever there is football, there is data. Not true. Wherever there is football, there is memory — but memory is not data, and a match remembered by fifty thousand spectators can still be an invisible match to a machine.

This article is an archaeology of that blank. I will dig down into the sediment where Vietnamese football lies, not to find gold, but to find the first brick that was laid crooked. Because once you know why data disappears, you understand why so many models that shine on paper died silently on tropical grass.

Data disappearing is not the loss of data — it is a kind of data. And that kind of data, in Vietnam, is telling us more than any metric.

A thin land of data

Based on my experience following matches, there is a moment that any analyst who has worked with Southeast Asian football recognises: you are not short of matches, you are short of the depth of matches. A V.League match has 90 minutes, a scoreline, cards, spectators. But it does not have the second layer of data — the layer that in Europe is treated as default: shot location, shot angle, type of play, pressure before the shot, the expected-goal value of each attempt.

V.League 1, Vietnam's top league, operating under the structure of the Vietnam Football Federation (VFF) and a professional football joint-stock company, is a real league. It has a history from the early 2000s, with long-established clubs such as Hanoi FC, Hoang Anh Gia Lai, Song Lam Nghe An, Da Nang, Binh Duong, Thanh Hoa, Hai Phong. It has players, contracts, money. But it lacks the infrastructure of micro-recording. In England, every Premier League match generates thousands of positional data points per second. In Vietnam, most of what you have is the scoreline and the match report — things a model cannot chew.

I once thought this was a technical problem, one that would be solved when cameras got better, when technology got cheaper. I was wrong in the most naive way. The problem is not the camera. The problem is the market: a data-recording system only survives if someone pays for it to exist. And in Vietnam, the people who pay for detailed football data barely exist at sufficient scale. There is no large legal betting market, no high-paying professional analytics industry, no sports media rich enough to buy a data package. My machine starved not because of technology, but because of economics.

Here is the point I want carved into stone: data is not a natural phenomenon. Data is a commodity. It appears where there is a buyer, it disappears where no one pays. When I watch a V.League match on screen and cannot look up the expected-goal value of an attempt, I am looking at an empty market, not an empty pitch.

There was a time I thought I understood Southeast Asia because I understood Europe. That was the arrogance of an apprentice. Vietnamese football is not a buggy version of European football. It is a different ecosystem, with different economic rules, and a model imported from Europe will break at exactly the point where it is most confident: at the point where it assumes data is default.

xG and the trap of imported metrics

xG does not score goals, but it makes people argue more than the real ball ever does. And in Vietnam, it does something more dangerous: it creates the illusion of precision.

I remember 2026. I was 35, a senior expert at a new sports platform. Before a match in the closing stretch of the season, I published an analysis using xG. The team I assessed had an xG of 2.8 against the opponent's 0.4; I predicted a 3-1 win, while the traditional experts all picked a draw. The result was 3-1. The article reached fifty thousand views in twenty-four hours. I thought I had just proven something about the power of data.

I had proven nothing. I had merely matched one roll of the dice.

That is the lesson it took me years to digest. xG is a model — meaning a set of assumptions about the probability that an average shot becomes a goal, based on the historical data of a specific set of matches, in a specific league. When you take an xG model built on tens of thousands of shots in Europe and apply it to a tropical league, you are doing what I call data migration. And data, when it migrates, degrades.

In Vietnam, a shot from 20 metres does not carry the same meaning as in England. Different pitch, different temperature, different humidity, different ball quality, different goalkeepers, and a different way of defending. A defender in the V.League may drop deeper, hold a different distance, leave a different gap. Apply the European model without recalibration and you are not analysing Vietnamese football — you are analysing your own assumption about Vietnamese football, then calling it data.

Every model is wrong, but a few are wrong usefully. An imported xG model applied to the V.League can still be useful if you know where it is wrong: it undervalues long shots in conditions where the goalkeeper is blinded by sun, it overvalues short passing combinations on bad pitches, and it ignores entirely the variable I consider the most important in Southeast Asia — the heat. Heat is not decoration. It is a tactical variable, and it is not in any European data package.

At high temperature and humidity, pressing intensity drops, the distances between lines stretch, and the number of decisive actions in the final 20 minutes falls sharply. A model built in England does not know this. It thinks minute 85 is the same as minute 15. But in a V.League match in May, minute 85 is another planet.

The PPDA metric — passes allowed per defensive action — is the same. It measures pressing intensity, and it assumes pressing is a purely tactical choice. But in Vietnam, sometimes it is a consequence of fitness, of a crowded schedule, of the pitch. You cannot read a low PPDA as "this team presses well" if you have not excluded the fitness variable. Every time I am about to write the word random, I ask myself: how many confounding variables have I excluded? Here, the answer is usually: not many.

The HAGL golden generation: a beautiful model that died on the grass

If there is one Vietnamese story worth excavating more than any other, it is the Hoang Anh Gia Lai football academy, known by a name tied to a famous European academy, opened in 2026. It was one of the largest youth-development projects private money had ever poured into Southeast Asia.

The Empty Model: Vietnamese Football and the Archaeology of a Machine With Nothing Left to Get Wrong

In theory, it was a beautiful model. You gather the country's most talented children from the age of ten or eleven. You feed, house, educate, and train them in a closed environment built to European standards. You teach them a technical, passing, ball-control philosophy. After seven or eight years, you have a generation of players raised uniformly, playing together since childhood, understanding each other in every run. On paper, you have built a La Masia of the highlands.

And indeed, that generation produced the names a whole decade of Vietnamese football had to mention: Nguyen Cong Phuong, Nguyen Tuan Anh, Luong Xuan Truong, Nguyen Van Toan, Vu Van Thanh, Tran Minh Vuong. These were players with individual technique above the norm, men who could dribble past three defenders in one action, men who made spectators stand up.

But that model died on the grass, and its death could have been predicted with data — if we had data. It died for three reasons no results table displays.

First, a closed academy produces players who are uniform in style but lacking in diversity of type. The HAGL generation was famous for small, skilful attacking players, but thin in physique, short of defensive, duelling, aerial profiles. A team cannot live on beautiful names. It lives on balance, and balance is not in the philosophy of an academy that knows only one way to play.

Second, their technical beauty was raised in a youth environment where opponents were also young and also small. Stepping into the V.League, with burly foreign defenders, hardened veteran midfielders, matches decided by duels — that generation lost the advantage that was its identity. Individual technique does not beat collective intensity. That is a law found in no brochure.

Third — and this is the part I want to stress most — the HAGL model was misjudged from the start because it was sold as a story, not as a system. An academy is not a beautiful building or a beautiful philosophy. It is a production system, and a production system is measured by the ratio of players who mature to standard against total intake. If you take in three hundred children and produce six stars, that ratio is two per cent — not bad, but not legendary either. The problem is no one measures. People only count the famous ones, because the famous ones sell tickets.

Here I touch a position I have carried for years: retired stars opening youth academies are largely a commercial stunt. I say this not to insult anyone, but because I have seen too many youth projects built around the image of a famous player. What gets invested is a brand, not a curriculum. What is missing is the dullest and most important thing: a system for training grassroots coaches. A country can produce one star from a flashy academy. But a country can only produce a football culture when it has thousands of grassroots coaches, properly trained, paid enough to live, working in places with no cameras.

The HAGL golden generation is a gem. But one gem does not make a mine. And if you build an entire football nation around one gem, when the gem rolls away, what are you left with? You are left with a cheering stand and an empty spreadsheet.

Park Hang-seo and the paradox of stability

In 2026, Vietnamese football appointed a South Korean coach, Park Hang-seo. It was one of the most important decisions in the modern history of this football nation, and it was also a fascinating data problem — because it ran against every model I had ever built.

My models, back then, liked possession teams, teams with high attacking metrics, teams with positive xG. Park Hang-seo built the opposite: a tight, disciplined, waiting team that exploited mistakes. This is what in my language I call a negative model — a team that wins by not losing, rather than by scoring many goals.

On paper, his team often had unremarkable attacking metrics. But the results were historic. In 2026, Vietnam won the Southeast Asian regional tournament. In 2026, the team reached the quarter-finals of the Asian Cup. In 2026, for the first time in history, the team reached the third round of Asian World Cup qualifying.

And on 1 February 2026, at My Dinh Stadium, Vietnam beat China 3-1. This is a match I remember vividly, not for the scoreline, but because it was a slap in the face of every model — mine and those of people better than me.

Let me tell you why. Before that match, my model, based on the balance of forces, squad value, and recent form, made China the favourite. That was a reasonable conclusion on paper. The Chinese squad had a higher transfer value, players abroad, a football nation with greater resources. If you look only at static data, China wins. But that match took place right around Lunar New Year, before a feverish My Dinh crowd, in an emotional context no model can encode.

That is what I want to say about Park Hang-seo's stability. He did not build a perfect attacking machine. He built an emotional machine. He understood that in Vietnamese football the decisive factor is not metrics but spirit — what I call the forgotten variable. European models have no cell for "spirit". They call it "noise" and throw it in the bin. But in Southeast Asia, what gets thrown in the bin is precisely the most important thing.

The paradox of stability is this: a team stable in results can have average process metrics. If you judge that team by process data, you will say it is lucky. If you judge it by results, you will say it is efficient. Both are true, and both are incomplete. The truth is Park Hang-seo optimised for something the model cannot measure: the ability to turn a match into a match his team can win. That is a skill, not a metric.

I once argued fiercely about this with colleagues. Some said Park's team won through luck. I said: if luck repeats for four years, it is no longer luck — it is a skill you have not yet encoded. But then I had to warn myself: careful, this is precisely the reverse trap. You can call everything an "unencoded skill" to avoid admitting a model failed. I have reminded myself of that many times.

Troussier: an imported model against a tropical wall

By 2026, Park Hang-seo left the seat. His replacement was Philippe Troussier, a French coach with international pedigree and a philosophy of attacking, possession-based, modern football. On paper, a step forward. An imported premium model.

And it broke. Not slowly — fast, clearly, so clearly that he had to leave the position in March 2026 after a run of poor results, peaking with defeats to Indonesia.

I spent a great deal of time excavating this break, because it is a perfect example of my argument about data migration and imported models. Troussier brought a philosophy built in Europe, where players have a good technical base, where pitches are perfect, where match tempo is governed by leagues of high uniform quality. He tried to apply that philosophy to a Vietnam national team with very different characteristics.

The result was a team out of rhythm. The defence pushed high to press, but the press was ineffective because the players were not used to that intensity in hot, humid conditions. The midfield tried to control the ball, but ball control requires pass quality under pressure, and under pressure those passes went missing. Gaps appeared behind the high defensive line, and Southeast Asian opponents — teams not strong in possession but very good at counter-attacking — exploited exactly those gaps.

This is the lesson I call the asymmetry law of the imported model. When you import a model from a stronger ecosystem into a weaker one, you cannot import the whole ecosystem. You import only the shell of the model. In Europe, high pressing works because the whole squad has fitness, technique, habit. In Vietnam, if you bring only the idea of pressing without the underlying fitness and habit, you bring a trap for your own team.

What I find most regrettable here is not the failure — failure is normal. What is regrettable is how people draw the lesson. Most of the public concluded that "modern football does not suit Vietnam". That is a wrong and dangerous conclusion, because it turns a problem of execution into a problem of identity. The truth is not that modern football does not suit. The truth is that the transition to modern football needs a buffer period, needs investment in fitness, needs time, and needs a coach who understands you cannot jump from Park's negative model to Troussier's positive model in one season.

Here again is where I must be careful with myself. It is very easy for me to stand outside and say "I knew it". I did not know it. I only knew that every model needs an adaptation period, and that period is never granted enough to a foreign coach in Southeast Asia.

Kim Sang-sik and the return of the simple

By 2026, Vietnamese football turned back to a South Korean coach, Kim Sang-sik. And the national team won again, this time the regional championship at the end of 2026, after beating Thailand in the final.

I do not want to turn this into a story of "Korea good, France bad". That is the kind of thinking I hate. What I want to point out is a rule about models: when an ecosystem is not ready for a complex model, the simple model wins. Not because simple is inherently better, but because simple fits the existing infrastructure better.

Kim Sang-sik returned to basics: defensive organisation, discipline, quick transitions, exploiting the strengths of existing players rather than forcing them into a new mould. It is a less ambitious model in theory, but one that fits reality better. And in football, the model that fits reality always beats the model that is beautiful on paper.

But — and here I must be blunt — a regional title does not prove Vietnam has solved its problem. It proves Vietnam is better than regional opponents at a given moment. Those are two different things. At regional level, a well-organised team with good spirit and a few technically superior players can win. At Asian and world level, those things are not enough. And I say this not to diminish the achievement, but to keep a cool head — because Vietnamese football has a worrying tradition: after every success, people forget where they are.

A personal shock and a mandatory warning line

I must tell you a story, because it explains why my articles always carry a warning line.

After the success of 2026, a betting company invited me to be chief analyst. At the 2026 World Cup, my model — based on PPDA and the height of the defensive line — correctly predicted a shock result: South Korea beating Germany 2-0. I posted a status urging everyone to bet with the model. I was right, and I became arrogant.

In the knockout round, my model believed Brazil would beat Belgium, because Brazil had better defensive metrics. I asserted it live on air. Brazil lost 1-2. Many clients lost money because they listened to me. I argued fiercely on social media with a colleague, then spent three weeks rewriting the entire codebase, adding tournament variables and a randomness factor.

Since then, every article of mine carries a line: the model is only probability, not prophecy. And I tell this story not to self-flagellate as a performance. I tell it because it is true, and because that truth is why I never tell you that Vietnamese football will win, or lose, or be champions. I can only tell you what I have seen, and what what I have seen is hinting at.

Thirty per cent that is not in the report

Here I want to open a perspective I call the thirty per cent that is not in the report.

Every football match has a visible part and a hidden part. The visible part is the scoreline, the cards, the basic statistics — what you find in any bulletin. The hidden part is the remaining thirty per cent: the goalkeeper's fear before a penalty, a young defender's hesitation for an instant, the silence of a dressing room after a conceded goal, the heat blurring eyes in the eightieth minute.

If the monk knows only xG and not the goalkeeper's fear before the goal, he will degrade from an explorer into a librarian. I nearly degraded like that. I once watched a match and saw only dancing numbers, forgetting that behind every metric is a human being who is afraid, hoping, tired.

In Vietnamese football, that hidden thirty per cent carries more weight than usual, because the visible part is so thin. When you lack detailed data, the hidden part takes a larger share of the whole story. That is why a purely data-driven model performs worse in Vietnam than in Europe: not because Vietnamese players are harder to understand, but because the signal-to-noise ratio is lower. You have less data, so you must rely more on judgement. And judgement does not get written into the report.

Money, contracts and the gaps in the balance sheet

Now let us talk about money, because money is what my model never ignores.

Vietnamese football operates on a financial structure very different from Europe's. In Europe, club revenue comes from three main sources: broadcasting rights, commercial, and matchday. In Vietnam, that weighting is reversed. Most V.League clubs depend on a single source: sponsorship from a parent company or a wealthy individual. This is what I call the one-legged model — it can stand, but it cannot walk.

When a club lives on sponsorship money, it has no incentive to build a business system. It has an incentive to please the payer. And when the payer loses interest, the club disappears. I have seen too many Vietnamese clubs dissolve not because they lost on the pitch, but because the chairman ran out of money. It is a form of systemic risk no tactical model can measure.

The transfer market is the same. In Vietnam, most transfers have no publicly disclosed fee in the European style. There are transfers whose price is known only once completed, and sometimes never. Players move from club to club under arrangements the press can only guess at. For an analyst, this is a nightmare: you cannot build a player-valuation model if you have no prices.

I once tried to build a player-value index for the V.League, based on age, position, minutes played, and goals. I abandoned it after three weeks. The reason is simple: there was no dependent variable. You cannot teach a model to predict price if you have no data on price. And I tell this as a confession of my own limits, not a boast of my patience.

What is notable is that a few Vietnamese players have gone abroad to play. Nguyen Quang Hai once moved to France to play for a lower-division club. Doan Van Hau once went to the Netherlands on loan. These moves matter more than their scorelines. Every Vietnamese player going abroad is a precious data sample, because it allows a direct comparison of a player trained in Vietnam with the international baseline. The problem is the sample is too small. A few samples are not enough to conclude, only enough to suggest.

Academies and the trap of famous masters

I spoke of academies above, but I want to return, because this is where my position is clearest.

In Vietnam, as in China — where I live and work — there is a repeated pattern: a retired star opens a youth football academy, holds a grand opening, invites the press, takes photos with children, and a few years later the academy either closes or becomes a shopping centre in sports clothing.

I do not say this to mock. I say it because it is a systemic problem. The problem is not that those retired stars have bad intentions. The problem is that they are placed in a role they were not trained for: managing an educational organisation. Being a good player and being a builder of a development system are two entirely different skills. People confuse the two because both relate to football.

What is forgotten is the grassroots coach. A country can have a hundred flashy academies, but without thousands of properly trained grassroots coaches, that football nation will still lack players. Because most players do not come from flashy academies. They come from small pitches in the provinces, from schools, from youth leagues no one streams.

This is something I learned from living between two football nations. In China, I see large academies with European facilities, and I see modest results. In Vietnam, I see small academies with little money, and occasionally a star. Both are solving the same problem: how to make a system produce players at a stable rate. And both are losing that problem, to different degrees.

I do not want to stand above and lecture anyone about their own football. I am only an apprentice in both places. But an apprentice has one advantage: he is allowed to ask silly questions. And the silliest question I want to ask is: if we can build a beautiful academy, why can we not build a beautiful coach-training school?

When the god of randomness laughs at the model

This is the part where I must be most careful, because it is the part where I am most tempted to dodge responsibility.

After 2026, the football world went through a shock no model predicted. Leagues stopped, schedules were scrambled, spectators vanished from the stands, and when football returned, it returned in a different body. I have written many times that football stopped rolling in 2026. That is not a poetic line. It is an observation about data: your time series is broken, and every model built on the previous series loses part of its foundation.

Football stopped rolling in 2026, but randomness has never taken a lunch break.

In Vietnam, that shock resonated with internal shocks. A league already thin on data, when scrambled, became almost impossible to model. Teams strong in fitness were hit by crowded schedules. Teams dependent on a few individuals were hit when those individuals caught the virus or lost form. And results became harder to predict — not because football became more random by nature, but because the confounding variables the model does not control increased.

Here I must critique myself. It is very easy for me to write "everything is random" and go to sleep. But that is an evasion. Every time I write the word random, I ask myself: how many confounding variables have I excluded? If I have excluded none, I am not yet allowed to use that word. And in Vietnamese football, I have usually excluded nothing. That means I am not yet allowed to call it random. I am only allowed to call it not yet understood.

That is a big difference. Random is a conclusion. Not yet understood is a state. And an honest analyst must distinguish the two, even when the distinction makes him look less clever.

Correlation is not causation, and the trap of the data storyteller

I want to devote a passage to the mistake I see most in this profession, including in myself: confusing correlation with causation.

An example. You notice that teams that win a lot often have high possession. You conclude: possession leads to victory. But it may be the reverse: strong teams take the lead, and when leading they hold the ball more. Possession is a consequence of leading, not the cause. If you build a team around possession without understanding this, you will build a beautiful team that loses.

In Vietnamese football, this trap appears everywhere, because data is scarce so people readily assign causation to what they see. Team X changes coach and wins three matches. People say the new coach is good. Maybe true. Maybe that team just met three weak opponents, just got players back from injury, just changed home stadium. You cannot separate those variables if you have only three matches.

This is why I always stress sample size. A sample of three matches says nothing. A sample of thirty matches begins to say something. A sample of three hundred matches gives you the right to conclude. And in Vietnamese football, with a season of far fewer matches than European leagues, you rarely have enough sample to conclude anything with certainty.

I say this to warn myself more than to warn you. Because I am the one who tells stories with data, and the storytelling profession always tempts us to bend data toward the prettier story. A nice story about a team reviving is always more attractive than the truth that the team just met three weak opponents. My job is to choose the truth, even when it is boring.

The cross-border view and the trap of standing above

I live in Shanghai, I was born in Vietnam, and I write about football for the Chinese market. That position gives me something I must be careful not to abuse: the distance between two frames of reference.

When you stand between two football nations, you see things someone inside only one cannot. You see how data migrates, how it degrades, and where it is wrongly worshipped. You see that a metric treated as holy in one place is a joke in another. You see that the same figure can tell two different stories depending on which frame of reference it is placed in.

But that position also has a trap: it easily turns you into someone standing above, lecturing Vietnamese about China and Chinese about Vietnam. No audience likes being lectured. I must constantly remind myself: tell it as an apprentice in both places, not as a conference chair.

There is one thing I have observed that I consider important, and I tell it as an observation, not a verdict. Both Vietnamese and Chinese football are wrestling with the same question: how to move from depending on a few outstanding individuals to building a system that produces players stably. Both have tried many ways. Neither has found the answer. And both tend to look for the answer far away, while the problem lies close: in the grassroots coaches, in the youth leagues, in the things no one wants to sponsor because they do not give pretty photos.

What an empty model taught me

Back to the night in Shanghai. The machine returned a blank, and I sat looking at it for a long while.

I realised I had learned more from one empty model than from many full ones. A full model teaches me about the world. An empty model teaches me about myself — about the assumptions I carry without knowing, about the things I treat as default that are in fact privileges.

I treated data as default. That is the privilege of someone who works with European football. When I came to Vietnamese football, I carried that privilege like luggage, and the luggage became a burden. The empty model reminded me that not everywhere has data, and that having data does not mean having truth.

People say I am good at predicting. Wrong. I am only good at saying "I do not know" at the right moment. And in Vietnamese football, "I do not know" is a more frequent answer than I would like to admit.

Signals for the next cycle

So what do I watch in the next cycle?

I watch the data infrastructure first, because that is the root. If the V.League gains more detailed data providers, if clubs begin to hire analysts, if the media begins to pay for data, then everything else will change. Without that infrastructure, every analysis is a decorated guess.

I watch the grassroots coaching layer, because that is where a football nation is built or abandoned. A country can buy a star. It cannot buy a generation. And a generation comes only from people teaching children in places with no audience.

I watch how clubs manage money, because the one-legged model will always break. A club living on a single revenue source is a club waiting for its end. I want to see clubs begin to diversify revenue, even when that is less glamorous than a blockbuster signing.

And I watch patience. This is the hardest signal to measure and the most important. Vietnamese football has changed national-team coaches many times over the past decade. Every change is a wiped board and a fresh start. A model cannot mature if it is erased after every season. And patience, like spirit, has no cell in any of my spreadsheets.

I will return to this topic — to Vietnamese football's data infrastructure, and to the price of not being able to measure. Perhaps next time I will have enough data for my model to be wrong usefully, instead of merely returning a blank. Until then, I sit here, in Shanghai, looking at an empty spreadsheet, reminding myself that a blank is also an answer. Just one I do not yet want to hear.