When the VBA Box Score Goes Quiet: Analytics as a Trade and the Price of Saying 'I Don't Know

Hai Phong, an evening in August. After the final buzzer, the A4 sheet came down the scorer's table into my hands: 94–88, four quarters, every name present. But the minutes-played column for two visiting players was blank. The scorer explained in half a sentence: “The tablet died midway through the third.” Twelve minutes of basketball had vanished from the record.

The next morning, three articles appeared. All three explained why the visitors lost: tired legs on defense, slow rotations, no second option when the ball handler was pressed. None mentioned the twelve missing minutes, because none had a reason to. I read three explanations for a silence, and realised my trade runs on an assumption nobody has tested: that wherever data is missing, a story will fill the gap, and the story that fills it will be correct.

It is not correct. It is merely fluent.

A league with enough numbers to play, not enough to conclude

The VBA tipped off in 2026 with five teams and now sits around seven or eight: Saigon Heat, Hanoi Buffaloes, Danang Dragons, Cantho Catfish, Hochiminh City Wings, Thang Long Warriors, Nha Trang Dolphins. A double round-robin plus playoffs, fewer than twenty regular-season games per team before the knockout rounds. Every game is played indoors under air conditioning, with spectators close to the floor, and every game is streamed on the league's own digital channels.

That is enough to explain why the VBA's public data stops at the lowest tier of the analytics ladder. The official box score gives points, rebounds, assists, steals, blocks, turnovers and shooting percentages. That is all. No play-by-play, no on-court lineups by minute, no tracking data. Offensive rating, defensive rating, all-in-one composites such as EPM or LEBRON exist only in the internal files of a handful of clubs that employ a video coordinator.

The gap between what the league records and what people want to conclude is far wider than the equivalent gap in the NBA. That is not inherently bad. A small league recording tier one honestly is more useful than a big league recording tier three impressively but with holes everywhere. The problem lies elsewhere: people use tier one to answer questions only tier three can answer, and nobody footnotes the borrowing.

Across eighteen years watching basketball, eight of them tied to Southeast Asian competitions and the VBA, I have learned something rarely written down: in small leagues, scarce data has never been the biggest problem. The biggest problem is that the scarcity gets covered over by a confident voice.

Who chooses the number

Every figure in a box score has an author. In the VBA the author is usually one or two people at the scorer's table by the sideline, tapping software, eyes on the ball. That person decides whether a pass counts as an assist. Whether a hand on the ball counts as a block or merely a change of trajectory. Whether a rebound was contested.

Those decisions involve no dishonesty. They simply have an owner. And owners always have incentives, even unconscious ones. The phenomenon of home scorers being more generous to the home team has been documented across professional leagues, most visibly in judgement-dependent categories like assists and blocks. In a league with few camera angles and few people checking the work, that bias tends to live longer.

Numbers do not lie, but the people who choose them do.

I am not saying the VBA box score is wrong. I am saying it is a document with a signature on it, and a professional reader has to read it the way one reads any signed document: ask who drafted it, under which conventions, and which way those conventions lean.

A small example I have met many times. A long outlet pass in transition, a teammate collects it, takes one dribble, finishes. At many scorer's tables that still gets logged as an assist, because it is “almost” an assist. Over twenty games, a primary ball handler can pick up two or three extra assists per night and nobody calls it an error. But when you use assist counts to compare two players on two teams with two different scorers, you are comparing two conventions, not two people.

Three tiers of data

Analysts split basketball data into three tiers. Tier one is the final box score: points, rebounds, assists, shooting. Tier two is play-by-play and on-court lineups by minute, enough to compute the plus-minus of a five-man unit. Tier three is tracking data: positions, speeds, distances, shot angles.

The VBA lives in tier one, occasionally touching tier one-and-a-half at a few strong clubs with someone logging video and re-charting possessions. Almost every public analysis of Vietnamese basketball is written with tier-one material and concluded at tier three. Writers talk about “spacing”, about “ball rotation”, about “shot quality” — concepts that can only be measured with positional data — while holding nothing but points and percentages.

A sentence like “this team creates gaps on the weak side” requires data on where the opposing guards stood on each possession. The VBA does not have that data. The only observable substitute is weak-side three-point percentage — a composite that merges shot quality, pass quality and luck. The box score cannot separate those three.

That borrowing is not always harmful. It is harmless when the writer knows they are borrowing. It becomes dangerous when the writer forgets, and starts believing they have measured what they only described.

The Four Factors and the limits of applying them

In 2026 Dean Oliver published “Basketball on Paper”, systematising the four factors that decide games: effective field