A Blank Analysis Report in Nagoya and the Data Standards Vietnamese Volleyball Needs

It was three forty in the morning in Nagoya. I opened notebook forty-one, checked it a third time, and the result was the same as the two previous checks: empty. No match name, no date, not a single metric qualified for the table. One month, 64 matches, and every dead-ball situation charted, yet the dataset for the second-stage analysis had nothing to open.

I shut the laptop and did not publish. In twenty-eight years in this trade, this is the fourth time I have accepted losing a week of output to hold one principle: empty input, empty output.

What made me write this down is not a technical incident. The pressure to deliver a verdict after every round of fixtures is what erodes the standards of volleyball analysis, in both markets where I work.

Since the 2026-25 season, professional volleyball in Japan has entered a new cycle under the SV.League, raising the bar for data infrastructure. Statistics are published rally by rally: scorer, attack type, starting position, first-pass quality, number of block touches, ball direction after contact. Each rally is broken into five to seven data fields, each with its own code table. That code table is the invisible part when you read a summary printed on a news page.

In Vietnam, public data from the national championship mostly stops at set scores and a hitter leaderboard. To get a perfect-pass rate, a dig rate or blocks per set, an analyst has to re-tag the video rally by rally. Based on my experience tracking matches, two people charting the same match can differ by 10 to 15 percent on block-touch metrics, simply because each draws the line at a different instant.

I do not impose the reading habits of the Japanese league on Vietnamese volleyball, nor the reverse. The professional standards of each market differ at the level of definition, well before the level of data quality. The job is to state which market you are standing in before quoting any percentage.

One paradox consumes most of my time: dead-ball situations account for under a third of the points in a match, yet take nearly two thirds of my charting hours. I watched 17 matches only to find the gap a player leaves behind his own back every time the opponent serves. That habit is useful, and it is also a blind spot unless I keep a separate notebook for open-play sequences.

My data pipeline has four layers: raw scores, match records, rally coding, and only then the model. They connect through one condition: every row must attach to a match, a date, a set and a situation. When the first layer loses match names and dates, the three layers above collapse at once, and they collapse quietly.

That is the state the second-stage analysis returned to me: a blank result. Nine analytical categories were all flagged as insufficient information: tactics, data, competition system, competitive landscape, rules, team building, risk, public narrative and industry transmission. The reason lies in having nothing to load, not in the difficulty of those nine categories.

Technical and tactical metrics are where the gap shows most plainly. My table has six columns: attack efficiency, blocks per set, ace-to-error ratio, perfect-pass rate, dig rate and point distribution by zone. Without match names and dates, those six columns still read smoothly, but they say nothing, because nobody can verify which match they belong to.

Blocks per set is the clearest example of the standardisation trap. A 3-0 win gives a three-set sample; a 3-2 win gives a five-set sample. Mix both into one ratio without stating the set count and you produce a metric that looks objective while only reflecting match length. The ace-to-error ratio works the same way: seen alone, the ace count makes an aggressive server look outstanding; seen with the error count, he may be trading point for point. I once read a summary that listed aces only, and it told the opposite story to the original scoresheet.

Dig rate depends almost entirely on definition. Is a rally counted as a successful dig when the libero keeps the ball for the setter, or merely when the ball is touched and stays in court? Two definitions produce two different results on the same video set, and the distance between them is larger than any difference in ability I have ever measured.

First-pass quality is where cross-league comparison goes wrong most often. The Japanese coding system splits the first pass into three grades, and a pass counts as perfect only when the setter has all three attacking options available. Many places grade by the impression of someone sitting outside the court, which means the yardstick differed from the start. Placing two percentages side by side without stating the code table is comparing a ruler with a piece of string.

Point distribution by zone requires the coordinates of the landing spot, something a paper scoresheet almost never carries. Without coordinates, I cannot separate a team that attacks the wings efficiently from one that attacks the wings and happened to meet a weak block.

Small samples are the next trap. A hitter who plays four sets in one round can swing 15 percent in efficiency from his own previous round, and most of that gap sits inside the noise band. I doubted the 30 percent figure I had measured, so I watched 15 matches again before believing it.

The layers covering competition system, schedule and rules fail in the same way. To judge schedule pressure I need fixture density, the rest window between rounds and travel hours. To judge the impact of rules I need to know which code table applied to which match. Video review duration is a variable too: two minutes of waiting is enough to cool a rally down, and if I do not log the length of every review, I will blame a scoring drought on form when the real culprit is a cut in match rhythm.

Team building and personnel require player files: age, sets played, club and national-team load, injury history. Without match names, I cannot link that load to any change in form. Risk and public narrative depend on media data: number of articles, engagement level, the moment a controversy breaks. The industry transmission layer, the chain from youth development to the professional league and on to the broadcasting market, needs a far longer time frame, and it cannot run on an empty dataset.

Looking at all nine categories together, I see a rule simpler than any model I have built. The value of an analysis lies not in the complexity of the model, but in the honesty of the source data layer. Before publishing any model, I try to break it first.

My way of breaking it rests on three charting habits. I record three situations that contradict my own hypothesis; if I cannot find three, the hypothesis drops to the level of an observation. I keep the dead-ball notebook separate from the open-play notebook, because the habit of counting dead balls makes me overlook long attacking sequences. And I log the duration of every video review, so that I do not blame form when match rhythm was cut by two minutes of waiting.

Some will say silence is waste. I understand that pressure, because the incentive structure of the industry runs backwards: writers are rewarded for decisive predictions and punished for saying there is not enough data. The result is that every gap gets filled with three cheap things: a small sample, belief and confident language. The romantic story of a small club beating the giants runs on exactly that mechanism; it drapes an emotional layer over financial gaps and operational reality that cannot be checked. An honest blank return is always cheaper than a wrong one, because the wrong one keeps getting cited.

At the individual level the pressure is quieter. Representation contracts mean many athletes speak only from approved scripts, so most interview data I collect has already been filtered. I learned not to use it as evidence, only as the ground for the next question. A hitter such as Tran Thi Thanh Thuy or Nguyen Thi Bich Tuyen can be quoted for attack efficiency while nobody mentions first-pass quality and the number of opponent blocks; the quotation is still literally true, and useless for reading a match.

Next week, if my dataset still lacks match names, dates and at least three data points, I will publish a blank return again. What I want to know is whether the people working in Vietnamese volleyball will agree on a shared code table for first pass, digs and block touches before we talk about predictive models. If the answer is yes, the first task is not buying software, but redefining what counts as a perfect pass.