Why no website can compute your exact banding: standardisation and school adjustment explained

What this page answers

Students in Primary 6 in September 2026 are in the 2025/2027 Secondary School Places Allocation cycle. They enter Secondary 1 in September 2027, and their internal assessment is still scored 999632. This page answers why no website, tutor or school can tell a family the child's exact band.

What the Bureau explains to families is that mechanism. The briefing says the three terms of internal results are submitted, standardised and adjusted. It does not state how the three submissions weigh against each other, or how a school converts its own exam marks into the marks it submits. The individual band itself is never released.

Four steps from internal marks to a band

Step 1: the weighted composite

There is no territory-wide paper. Schools submit internal-assessment marks. For the 2025/2027 cycle the weights are 999632, on a base of 38: Chinese 9, English 9, Mathematics 9, General Studies 6, Visual Arts 3, Music 2. Religious Education (2) and Putonghua (1) are optional and are the school's decision. A subject's weight divided by 38 is its share of the composite, so Chinese, English and Mathematics are each 9/38. The cohort entering Primary 5 in September 2026 is the first under 88844222 (cycle 2026/2028, Secondary 1 in September 2028). The subjects differ, but they still form a composite before the standardisation below. The composite is only the starting point. What the computer later ranks is not the raw school average.

Step 2: standardisation within the school

The computer first standardises each student's Primary 5 second-term and Primary 6 first- and second-term internal marks, so that schools which mark generously and schools which mark strictly can be compared. The Bureau describes this step as standardisation. The briefing does not set out a formula a family can substitute marks into. The same mark on the page need not stay the same after standardisation at two schools. All three submissions enter this step. The Bureau does not state that any one of them counts for more.

Step 3: adjustment by the school's sample average

After standardisation, marks are adjusted by the average of the primary school's two most recent Pre-S1 HKAT samples. Since 2006 the Bureau has sampled answer scripts in even years only, so 2027 is not a sampling year. The 2025/2027 cycle uses that school's 2024 and 2026 samples. The previous cycle (2024/2026) used the 2018 and 2024 samples. The sample average adjusts later cohorts at that primary. It does not score the sitting student for allocation, and the Bureau says schools and parents need not drill for the test. For 2025/2027 the test is on 13 July 2027, after results on 6 July. The test is confidential: parents cannot see the scripts.

Step 4: ranking and cutting into thirds

Adjusted scores are ranked and split into three groups of equal size. The territory band ranks all students together and is used for Central Allocation Part A. The net band ranks students inside each of the 18 secondary school nets and is used for Part B. Both bands are calculated only when the allocation programme runs, after the final participant count is fixed. Individual bands are never released to parents. Until that count is fixed, there is no personal band already cut and available to look up.

Why a school cannot calculate it either

The 2025/2027 briefing states this directly: even with past cohort profiles plus a student's internal marks or rank, a school cannot precisely predict an individual student's band. The actual band may differ from what the school or the parents predict.

The band is calculated only when the allocation programme runs, and the final number of participants is fixed only then. Part A uses the territory band and Part B uses the net band. The rank is against the relevant students of that year, not against an earlier cohort at the same primary. The adjustment uses the average of that school's two samples. Internal rank on its own does not include that adjustment, and it does not include the territory-wide or net-wide ranking.

What a family can know is the internal marks and rank the school holds. If a teacher or the school offers a general indication from past profiles, that is a prediction. The Bureau has already said the actual band may differ from it.

Territory band and net band

Each student has two bands. The territory band splits every student's adjusted score into three equal thirds and is used for Central Allocation Part A (unrestricted by school net). Of the places left after repeaters and Discretionary Places, Part A is 10%. Parents may list up to 3 schools in any net, including their own. The programme processes all territory Band 1 choices in order, then Band 2, then Band 3. A student placed in Part A is not processed in Part B.

The net band splits adjusted scores into three equal thirds inside each of the 18 secondary school nets and is used for Part B (by school net). The 18 nets are the 18 administrative districts. In principle a student's net is the district where the primary school is located. Part B is 90% of the remaining places, with up to 30 schools in the student's own net. The programme processes net Band 1, then Band 2, then Band 3. Within the same band, if a school is oversubscribed, a random number decides the order, smaller number first. The number is generated before the run and is unrelated to the student number or to personal data. Allocation uses the band, the order of parents' choices, and this number.

A cross-net student keeps the same territory band. The band in the new net is decided by comparing that student's adjusted marks with students in the new net. Once the move is approved, Part B runs in the new net, so the two bands for one student can differ.

What an estimate can and cannot do

General guidance, not an Education Bureau rule. At best, an estimate places a composite beside one primary school's past intake pattern. It cannot know that school's sample adjustment, and it cannot see other students' adjusted scores or the final headcount.The banding estimatorturns a within-school position the family enters into a reference. It is not the Education Bureau band.

An example

Take two students at different primary schools, with identical marks on the page for all three submissions. This is a scenario only.

Each composite is standardised inside its own school. If the two schools mark on different scales, the standardised results can differ. Each is then adjusted by that primary's own sample average: for 2025/2027, the average of the school's 2024 and 2026 samples. Those two averages need not match. The next step ranks the adjusted scores with every student in the territory, or in that school net, and cuts them into three equal thirds. Identical marks on the page can therefore sit in different places after adjustment, and the bands can differ. Until the programme runs and the headcount is fixed, that comparison has not happened, so nobody can say in advance which third each student is in.

Common misunderstandings

• A school's "Band 1 rate" is my child's band. The Bureau states that past cohort profiles plus internal marks or rank still cannot precisely predict an individual band. How many students in an earlier year were in the first third is not this student's band. The band is cut from adjusted scores into three equal thirds only after that year's headcount is fixed.
• The HKAT the child sits decides the child's band. The test does not directly affect the sitting student's own allocation. Sampling happens in even years only and adjusts later cohorts at that primary. The 2025/2027 cycle uses the 2024 and 2026 samples. The test on 13 July 2027 is after results on 6 July, and 2027 is not a sampling year.
• The band is already fixed before the Central Allocation form. Bands are calculated only when the allocation programme runs, after the final participant count is fixed. For 2025/2027 the e-platform window is 8 April to 5 May 2027. Submitting the form does not display a band to parents. The three submissions are inputs to standardisation, not a personal band already published.
• The net band equals the territory band. They are two separate cuts into thirds. The territory band is used for Part A and the net band for Part B. After a cross-net approval the territory band stays the same, while the new net band is compared afresh with students in the new net, so the two can differ.
• A website with more data could calculate the exact band. Individual bands are never released to parents. The cut needs every relevant student's adjusted score for that year, and a participant count fixed only when the programme runs. A school that already has past profiles and internal rank still cannot predict precisely. Extra public data is still not the input to that run.

Next steps

To see which figures the estimator reads from the family, use thebanding estimator. For how 999632 and 88844222 build the composite, seeSSPA subject weights. For how the band is used when listing Part A and Part B choices, seeCA choice-form strategy.

FAQ