🎓 JUPAS Simulator

JUPAS Simulator Insights · Data Analysis

2026 JUPAS Cut-off Forecast: the Birth Wave, Surging DSE Candidates & What the Data Says

JUPAS received 45,273 applications for 2026 entry — up 5.5% and rising for a second straight year — while 58,500+ candidates sat the DSE. Will admission scores climb again? We walk through six years of official data, from birth statistics to cut-offs.

Three demand lines: born, sitting the DSE, applying via JUPAS

A popular rule of thumb predicts DSE competition from "births 17–18 years ago". Put the three lines side by side and the story turns out to be far more interesting:

80k 70k 60k 50k 40k 202120222023 202420252026 Births (−18y) 78,822 DSE ~58,500 JUPAS 45,273 — — Births 18 years prior— DSE candidates— JUPAS applicants
Birth line = exam year minus 18 (most 2026 candidates were born around 2008). DSE 2026 is HKEAA's announced "over 58,500" registration; JUPAS figures are official applicant counts per cycle.
JUPAS applicants 202645,273▲ +5.5% (2nd year rising)
DSE candidates 202658,500+▲ about +4.9%
Births in 200878,822the 2026 cohort

Why the birth wave never became a candidate wave

Illustration: within the 2003-2011 birth wave, most babies born to non-resident parents returned to the mainland; only some stayed for school in Hong Kong
Hong Kong births climbed from 46,965 in 2003 to 95,451 in 2011, a large share born to parents who were not Hong Kong residents. Many of those children never entered Hong Kong schools — so the birth wave did not translate one-for-one into DSE candidates.

The actual numbers: births rose 44% from 2005 to 2009, yet the corresponding DSE cohorts (2023–2027) have grown only about 12% so far. A forecast that keys directly on birth statistics would badly overestimate the squeeze — which is why our model uses actual registration numbers, not births, as its demand signal.

The real engine of the rebound: mainland DSE cram schools

Illustration: a full-day DSE cram-school classroom in Shenzhen, an orange arrow pointing to a Hong Kong university exam hall — the mainland tutoring industry feeding students into JUPAS
Private DSE candidates surged 37% year-on-year to over 10,000 in 2025 — students who are not in Hong Kong schools at all. The mainland DSE cram-school industry is the real engine of the rebound.

A popular explanation credits children of Top Talent Pass families entering local schools — but the data points elsewhere. The fastest-growing group is private candidates (+37% to over 10,000 in 2025), who are not in Hong Kong schools at all. The real engine is a booming mainland DSE cram-school industry: a May 2026 investigation by HKBU's San Po Yan found 50+ institutions offering DSE courses on a single mainland tutoring platform (28 of them in Guangdong), with major chains clustered in Shenzhen's Futian district compressing the three-year DSE syllabus into 8–18-month full-time programmes at fees of RMB 200,000–300,000 a year.

The incentive is plain: the gaokao is a single-file bridge crossed by over ten million candidates, while the DSE has 58,000 worldwide and Hong Kong universities admit at far higher rates than elite mainland ones. Some operators also partner with local private schools in shell-school arrangements (借殼辦學) so students can sit as school candidates — or even claim local-student status in JUPAS — and the Education Bureau has announced a package of measures to crack down.

The effect is unmistakable in the data. After hovering around 50,000 through 2022–2024, DSE candidature jumped to 55,781 in 2025 (+9.8%) and to 58,500+ in 2026; JUPAS applications climbed from 39,298 in 2024 to 45,273 in 2026 — +15% in two years. Frontline teachers report that these candidates are exceptionally strong in mathematics and Chinese, lifting the bar for Levels 4 and 5 (the HKEAA responds that DSE grading is criterion-referenced and unaffected by private-candidate results). Funded first-year places have not grown to match. More students, same seats: competition tightens.

Does more demand actually raise cut-offs? 2025 was the experiment

We compared the official admission medians of the 262 programmes with consistent scoring scales across 2024 and 2025, programme by programme:

61% rose 20% flat 19% fell Change in admission median, 2024 → 2025 (262 programmes) Typical rise ≈ +3.5%, while JUPAS applications grew +9.2% and DSE candidature +9.8% Control: in 2023→2024, when demand dipped, medians broadly fell by 1–2 points
Sample: programmes with officially published medians in both years on a consistent scale. PolyU switched to publishing per-subject grades in 2025; those programmes are excluded rather than misread as "collapsing" scores.

Two findings:

1. At the macro level, yes. In the demand-surge year (2025), six in ten programmes raised the bar; in the demand-dip year (2024), medians broadly slipped. The direction is consistent.

2. At the micro level, no. Programme by programme, the correlation between "how much applications grew" and "how much the median moved" is just −0.09 — essentially zero. A programme's cut-off is set by the quality of its marginal applicant, not by headcount. Any claim of precisely forecasting each programme's rise does not survive contact with the data.

Our approach

We anchor on the actual 2025 median as the baseline, then apply a demand adjustment of about 2% for 2026 (demand growth is running at ~60% of last year's) — roughly +0.5 to +1 point on Best-5-style scales — presented as a range, never as false precision. Each programme also gets its own application-trend indicator as competition context, but we deliberately do not scale individual programmes by their own application growth.

The next few years: the birth data says this isn't over

Births reached 82,095 in 2009, 88,584 in 2010 and 95,451 in 2011 — so the college-age population keeps growing through the 2027–2029 cohorts. Even with limited retention of cross-border-born children, as long as the DSE stays an easier route to university than the gaokao and the mainland cram-school pipeline keeps producing candidates, further growth in DSE and JUPAS numbers is the base case — until the post-2013 "zero quota" birth cliff reaches university age around 2031. Regulation is the wild card: if the Education Bureau's crackdown on shell schools tightens candidate eligibility, the climb could slow.

In one sentence: today's Form 4 and Form 5 students will face stiffer competition than this year's cohort. Knowing each programme's realistic entry bar early is the most practical preparation.

FAQ

Will JUPAS admission scores rise or fall in 2026?
The data points to a moderate rise: about +2% overall (+0.5 to +1 point on Best-5-style scales), smaller than 2025's jump, because demand growth has slowed from +9% to +5.5%.
Will every programme go up?
No. Even in 2025, 19% of programmes fell. Programme-level moves are not precisely predictable (the empirical correlation is near zero), so think "baseline + buffer" rather than trusting any single number.
Why do some PolyU programmes look like they "crashed"?
PolyU changed its disclosure basis in 2025 to per-subject grades — a different scale from the previous weighted totals. Comparing across the break produces a phantom collapse. Our data handles the break explicitly.
How do I actually use these numbers for my choices?
Take the 2025 median as your baseline and add a half-point-to-one-point buffer for Band A stretch picks; put at-the-line programmes in Band B; keep one or two clearly safe choices. And score yourself with each programme's official weighting formula rather than guessing.
Check your own score with JUPAS Simulator422 programmes · official weighting formulas · free Browse all programmes' scores & requirements →