41 of 158 official cut-offs matched our October prediction
- 41Matched
- 45Up to 5 points off
- 336 to 10 points off
- 39More than 10 points off
- We assumed
- 8,400–9,400
- Home Affairs invited
- 20,000
The short answer
The 15 October 2026 round invited 20,000 people across 159 occupations. Our October prediction, frozen on 9 October, assumed a round of 8,400–9,400. Because the real round was more than twice that size, cut-offs fell further than we predicted in most occupations.
- October edition: matched the official minimum for 41 of 158 occupations (26%), within 5 points for 54%, average error 8.1 points.
- Direction: where we were wrong in October, we were usually too high. If we told you your occupation needed 85, there was a good chance it went lower.
- Earlier editions: August (36%) and September (32%) matched slightly more often than October.
Three editions, one round
Each edition was frozen before the round and kept unchanged. August and September published a range for some occupations; October published a single cut-off, so for October “inside our range” means exact.
| Edition | In range | Within 5 | Avg error |
|---|---|---|---|
| Augustfrozen 7 Aug · Previous method | 36% | 72% | 6.0 |
| Septemberfrozen 4 Sep · Previous method | 32% | 69% | 6.5 |
| Octoberfrozen 9 Oct · New methodlive on round day | 26% | 54% | 8.1 |
| Edition | Method | Official inside our range | Within 5 points | Average error |
|---|---|---|---|---|
| August frozen 7 Aug | Previous method | 36% (57 / 158) | 72% | 6.0 pts |
| September frozen 4 Sep | Previous method | 32% (50 / 158) | 69% | 6.5 pts |
| October frozen 9 Octlive on round day | New method | 26% (41 / 158) | 54% | 8.1 pts |
- we said 10+ pts lower
- 5 lower
- exact
- 5 higher
- we said 10+ pts higher
The August and September editions leaned low: they more often said a cut-off would be easier than it was. The October edition leaned the other way.
Why we missed: the round was twice the size
Every edition started from the same assumption: a first round of about 9,000 invitations, close to the 10,000 invited in each of the two rounds before it. Home Affairs invited 20,000.
Round size is the first thing a cut-off depends on. A larger round reaches further down each occupation’s queue, so minimum points fall. With 20,000 invitations, 80 occupations were invited at 65 or 70 points. We had predicted 80 or 85 for 64 of them.
Home Affairs does not announce the size of a round in advance. We chose one size and published one answer. That was the mistake.
Where the October lines landed
The October edition put occupations on three lines.
| Our October line | Official minimum, 15 October | ||||||
|---|---|---|---|---|---|---|---|
| 65 | 70 | 75 | 80 | 85 | 90+ | ||
| 85 · Most occupations · 95 | 95 | 28 | 8 | 3 | 28 | 17 | 11 |
| 80 · Health and education · 44 | 44 | 1 | 27 | 6 | 7 | 2 | 1 |
| 65 · Construction trades · 17 | 17 | 15 | 1 | · | 1 | · | · |
- Construction trades held. 15 of 17 were invited at 65, as predicted.
- Health and education went lower. We said 80; 28 of 44 were invited at 65 or 70.
- Most other occupations split in two. We said 85. 36 were invited at 65 or 70, 56 at 80 or above, and only 3 at 75. In past rounds these occupations moved together. This time one part cleared to the bottom of its queue and the other stayed high. Nothing in the earlier rounds showed this split, and we do not know the reason for it.
The new method did worse here than the old one
On 9 October we changed how the prediction is made. In tests on five past rounds the new method was exact for 63% of occupations. In its first live round it was exact for 26% (41 of 158).
The previous method, last frozen on 2 October, had the official minimum inside its range for 32% (51 of 157), with an average error of 6.3 points against 8.1 for the new one. On this round the old method was closer.
The reason is the round size again. The new method depends more directly on the number of invitations, so a wrong size hurts it more. When we re-ran it after the round with 20,000 invitations, it was exact for 34% of occupations with an average error of 5.3 points. That re-run uses a number nobody had before the round, so it does not count as a prediction. It tells us where the error came from.
One live round is not enough to choose between the two methods. We are keeping the new one, and we will keep publishing both results.
Every occupation
- 85official inside our range
- 855 points out
- 8510 or more out
Official minimums as published by Home Affairs on 2 November 2026. Sorted by the invitations we expected.
| Occupation | Aug | Sep | Oct | Official | |
|---|---|---|---|---|---|
| Registered Nurses (nec) | 254499 | 75 | 75 | 80 | 70 |
| Carpenter | 331212 | 65 | 70 | 65 | 65 |
| Early Childhood (Pre-primary School) Teacher | 241111 | 90 | 90 | 85 | 90 |
| Secondary School Teacher | 241411 | 75 | 75 | 80 | 80 |
| Statistician | 224113 | 85–90 | 90 | 85 | 85 |
| Multimedia Specialist | 261211 | 90 | 90 | 85 | 90 |
| Chemical Engineer | 233111 | 90 | 90 | 85 | 85 |
| General Practitioner | 253111 | 65 | 65–70 | 80 | 70 |
| Social Worker | 272511 | 75 | 80 | 80 | 80 |
| Registered Nurse (Aged Care) | 254412 | 75 | 75 | 80 | 70 |
| Medical Practitioners nec | 253999 | 75 | 80 | 80 | 70 |
| Electrician (General) | 341111 | 65 | 65 | 65 | 65 |
| Electronics Engineer | 233411 | 95 | 100 | 95 | 95 |
| Telecommunications Engineer | 263311 | 90 | 90 | 85 | 85 |
| Bricklayer | 331111 | 65 | 65 | 65 | 65 |
| Wall and Floor Tiler | 333411 | 65 | 65 | 65 | 65 |
| Architect | 232111 | 85 | 85 | 85 | 85 |
| Construction Project Manager | 133111 | 75–80 | 80 | 85 | 80 |
| Management Consultant | 224711 | 75 | 80 | 85 | 80 |
| Registered Nurse (Medical) | 254418 | 75 | 75 | 80 | 70 |
›Show the other 139 occupations
University Lecturer was invited at 90. None of our editions published a cut-off for it, so 158 of the 159 invited occupations are graded.
What changes for the next prediction
- More than one round size. We are working on showing cut-offs for a small, a medium and a large round, so you can see how much your occupation depends on it.
- The 15 October results go in. The next prediction starts from the pool after this round and counts the 20,000 invitations against each occupation’s places for the year.
- The same rule as before. We freeze the prediction before the round, keep it unchanged, and publish the grade afterwards.
FAQ
›How accurate was ImmiTrend's prediction for the 15 October 2026 189 round?
The prediction frozen on 9 October matched the official minimum for 41 of 158 occupations (26%) and was within 5 points for 54%. The average error was 8.1 points.
›Why was the prediction off?
We assumed a round of 8,400–9,400 invitations. Home Affairs invited 20,000, so cut-offs fell further than predicted in most occupations.
›Were the predictions changed after the round?
No. Each edition was frozen before the round with its date, and is shown here as published.
›Can I still rely on the 189 prediction?
Treat it as a reference. It shows where a round of a given size would reach in the current queue. The size of the next round is not known in advance, which is why we are working on showing more than one size.
General information only, not migration advice. Home Affairs decides invitations; for your own case, speak to a MARA-registered migration agent.