Perfect-Pass Rate and the Data Gap in Vietnamese Volleyball

At 10:40 p.m. I opened the scouting file for a group-stage match in Vietnam's national volleyball championship. The file had every heading: spike success rate, blocks per set, ace-to-error ratio, perfect-pass rate, dig rate. Every cell was empty. Not empty because I had not filled it in, but empty because the collection pipeline upstream had returned a hollow shell: correct template, correct format, missing every number.

The empty file is not the interesting part. The interesting part is that a file like that can pass through three layers of processing and come out as a report that reads beautifully. The end reader sees a document with headings, sections, conclusions and recommendations. Nobody sees that underneath every section lies nothing.

In volleyball analysis, an empty cell is data. An empty cell filled with guesswork is error.

I learned this long ago. 2026 taught me to listen to what the model cannot measure. That April I sat in front of three screens in Saigon rewatching Leicester City against Everton. The press praised a goalscorer, while the expected-goals table showed the home side created 1.2 units against 3.8. I stopped trusting live commentary and hand-counted all 380 matches of that season. Vietnamese volleyball sits exactly where English football sat then: plenty of feeling, very few definitions.

Context: many matches, thin data

The national championship runs in two phases. The Hung Vuong Cup opens the year, the VTV Cup is an invitational, the SEA V.League has two legs for men and women, and national teams also play the AVC Challenge Cup, the Asian championship and, for the women, the road to world qualification. Add youth events, regional cups and overseas training camps.

The match volume is not small. Public data, however, is thin to the point of disbelief. Football has international event-data providers and opponent-adjusted metrics. Vietnamese volleyball lives almost entirely on the scorer's sheet and the electronic board hanging in the arena.

Three problems stack on top of each other. There is no shared definition: a first pass logged as "good" in one arena can be logged as "average" in another. There is no positional data: nobody records how far from the net the ball was delivered, which is precisely what determines how many options the setter still has. And the sample is tiny: a match runs three to five sets, 25 points each, roughly 150 to 200 rallies, which split across six rotations leaves 25 to 30 rallies per rotation per match. That is a mathematical limit, not a failure of effort by the person recording.

When the pandemic shattered the 2026 calendar, teams had to cram fixtures into brutal conditions. Mid-crisis, I counted history again and saw that every cycle wears a familiar face: after each compressed schedule, injuries rise, ball quality improves in patchwork fashion, and statistical tables stop meaning anything because the sample no longer compares to itself.

Before discussing any team, I have to discuss the measuring frame. Get the frame wrong and every conclusion is wrong, even when the conclusion is elegantly written.

The root metric: perfect-pass rate

Everything in volleyball starts with the first contact. A perfect pass puts the ball exactly where the setter can run the full tactical menu: middle attacking behind, middle attacking in front, outside hitter, opposite, and a back-row attack.

The perfect-pass rate is the only metric that forecasts attacking quality before the ball is even hit.

In international women's volleyball I use these reference bands: above 45 percent is the foundation of a team that plays a system; 38 to 45 percent is decent, enough to survive but not to impose; below 32 percent and the team is pushed out of its system entirely, at which point any debate about the outside hitter is beside the point.

This is where Vietnamese volleyball misreads the game most. Post-match coverage usually publishes one fact: who scored how many points. An outside hitter with 22 points is always called brilliant. But if that took 58 attempts, it was not brilliance, it was a system funnelling balls to one person because no other option existed.

Based on my experience tracking matches live and on video, I always add a column the official sheet does not have: the number of first passes pushed outside the net zone, meaning the ball landed more than two metres from the net. That column says more than spike success rate, because it measures the real pressure the serving team created.

A citable example for comparison: in the SEA V.League, Thailand and Indonesia routinely post higher perfect-pass rates than Vietnam despite no height advantage. The reason is a trade-off they accept: less aggressive serving to preserve structure, and a reception system kept intact with the libero and two covering players. I do not use that as proof of individual quality; I use it as proof of serving philosophy.

Serving: the mislabelled metric

The common way to measure serving in Vietnam is counting aces. That number is close to useless.

A good serve does not have to win the point. It forces a bad first pass, and the consequence arrives two or three rallies later. The right metric is not ace count but the opponent's perfect-pass rate after that specific serve, measured across a player's entire service turn.

I use a ratio I call the serve trade-off: direct points plus rallies where the opponent was forced out of system, divided by service errors. Below 1 means the player is risking more than he returns. Many Vietnamese players serve hard and beautifully and still sit below 1 against opponents who can pass.

Six rotations: where small samples fool the eye

Volleyball has six rotations in service order. In three of them the setter is in the front row. In those three, the front row holds only two attackers and the team must compensate with back-row attacks. In the other three, with the setter in the back row, three front-row attackers are available and the team plays at its easiest.

None of this is new. What is new is how fast people conclude from it. If a team concedes three straight points in a two-attacker rotation, commentary declares a systemic problem. But a set is only 25 points and each rotation comes round about four times. Three points in four rallies is statistical noise, not pathology.

The correct way to separate signal is to pool matches and compute point differential by rotation. I have done this by hand across several domestic seasons. The results are usually uncomfortable: most rotations fans call "death rotations" differ by minus 0.8 to 1.4 points per turn, which sits inside normal variance when the opponent happens to serve better.

A genuinely sick rotation needs two conditions: a sustained negative differential across many matches, and a visible cause on video, for instance a middle who cannot run a quick ball, letting the opposing block collapse onto the wings.

Spike points and the trap of the pretty number

The most common evaluation error in Vietnamese volleyball is reading raw spike points.

Twenty points from 60 attempts is a completely different player from 20 points from 35 attempts. The first runs around 33 percent efficiency before errors and blocks are subtracted; the second runs around 57 percent. Both get called "scored 20".

True spiking efficiency is points minus attack errors minus times blocked, divided by total attempts. In international women's volleyball, above 40 percent is elite, 30 to 35 percent is good, and below 20 percent signals a war of attrition.

I always add one more: the share of team attacks a player receives. If an outside hitter takes 45 percent of the team's swings, the opponent barely needs to read tactics, only to send a two-person block that way. The opponent's block success rises not because their middles improved, but because they knew where the ball was going.

Middles and blocks: the biggest blind spot

Blocks per set is the most misleading number in the whole statistical sheet. It depends directly on who the opponent attacks. A middle standing next to a weak opposing hitter will post far more blocks than one standing next to a strong opposite. Without opponent adjustment, the figure is not comparable between two players.

A better substitute is block touches, including deflections that stay in play. A touch that earns no point can still decide a rally by turning a losing block into a counterattack.

The same logic applies to a libero's dig rate. A high dig rate is usually a sign of a bad block, not a sign of a good libero. The libero flies because the ball already passed overhead. It is a lagging indicator, and I never use it to rank individuals.

The real job of a modern middle is not points. It is stretching the block and creating one-on-one situations on the wing. No column in a Vietnamese scoresheet measures that. That is the gap a good middle lives in.

Correlation is not causation

Now the uncomfortable part. Win rates and perfect-pass rates correlate strongly. Many people read that and conclude good passing causes winning. That is half right.

The other half is that leading teams force opponents to serve more aggressively, which produces more errors and less pressure, which inflates the leader's perfect-pass rate. To separate the directions you must split the data by score, for example counting only rallies within three points and only rallies beyond five points.

I have run that split repeatedly. The result is stable enough to publish: most of the passing advantage of winning teams appears after they are already ahead. Only about a third exists beforehand.

Croatia was not a miracle story, they were a problem that needed solving from scratch. Solving it taught me that a good model is one that says "I don't know" in the right place.

And then there is the silence of the model. Video cannot measure how a team tenses up after losing a set 25-14, when every metric has already given up before the humans have. It cannot measure what is said in a timeout at 18-20. It cannot measure the accumulated fatigue of a key player who finishes the domestic season, boards a bus to national camp, then a plane to an Asian tournament.

With Vietnamese volleyball, the model knows less than it does not know. No positional data, no shared definition of a first pass, no opponent adjustment, no individual workload tracking. Anyone offering a firm verdict on a team's "identity" from five recent matches is describing something else.

Empty pipeline, full conclusions

The danger of a process returning an empty shell is not the missing data. It is that the empty shell still has the right shape to be read as finished work. Data engineers call it consuming an empty input as a valid one. In volleyball it has another name: a statistical table with every heading and no definitions.

When that happens in a league, three steps follow. The sheet is published with professional-looking columns. Fans read, compare and argue. Coaches, under result pressure, use those same numbers to justify decisions. By season's end a review is written on three stacked layers of error: inconsistent definitions, tiny samples, zero opponent adjustment.

The fix is not to abandon statistics. It is to state assumptions up front and accept that some conclusions must be postponed. An honest table with three empty cells beats a complete table with three inferred cells. Readers deserve to know what was measured and what was guessed.

Signals for the next round

Perfect-pass rate split by rotation, pooled over at least eight matches. Not to hunt a death rotation, but to see which reception structures survive when serves are aimed at the libero. Back-row attack volume, which reveals whether a team truly plays a system or is just covering a hole. The share of attacks funnelled into one individual: above 40 percent for three straight matches and risk is accumulating. Individual workload counted in sets, not matches. And the number of empty cells in the tables I collect, the only metric that measures the quality of the analysis itself.

Volleyball does not lack stories. It lacks people willing to record how many empty cells those stories were built on.