Hi everyone, and welcome to my newsletter! To everyone who’s subscribed, thank you so much, and sorry it’s taken me so long to sit down and write something for you.
We’re six weeks out from election day on 28 November, and things have been quiet. No party has released a how-to-vote card yet, so one of the biggest unknowns is still open: how every party will order its preferences. Where the Coalition puts One Nation gets the headlines, but the Labor, Greens and One Nation cards all matter too.
The polls have stayed steady too. Every party has held firm, and my baseline has barely moved in a month. So this is a good moment to explain what that baseline actually is, and how I build it.
Where the numbers sit
My current baseline is Coalition 29.0, Labor 24.4, One Nation 23.1 and Greens 14.6.
The Greens are the only party showing much movement in the baseline, up 0.6 points to 14.6 since late September. The two newest polls, RedBridge (2–8 Oct) and Resolve (6–10 Oct), put them at 16 and 15 respectively.
Here are the last five polls:
| Pollster | Fieldwork | Sample | Coalition | Labor | One Nation | Greens |
|---|---|---|---|---|---|---|
| Resolve | 6–10 Oct | 1,000 | 28 | 25 | 23 | 15 |
| RedBridge | 2–8 Oct | 1,008 | 30 | 24 | 24 | 16 |
| Freshwater | 24–28 Sep | 1,030 | 29 | 24 | 22 | 14 |
| DemosAU | 10–20 Sep | 1,020 | 30 | 25 | 20 | 14 |
| RedBridge | 1–14 Sep | 2,371 | 29 | 24 | 25 | 15 |
Put them side by side and every poll tells roughly the same story. Five polls from four pollsters all have the Coalition between 28 and 30 and Labor between 24 and 25. One Nation ranges from 20 to 25, and the Greens from 14 to 16.
Most of these polls sampled around 1,000 people, which gives a margin of error of about 3 points either way — small moves in a single poll are mostly just noise, not a real shift. That’s why I don’t trust any single poll alone; I look at all of them together.
How I build the baseline
The baseline is my poll aggregate: one statewide number per party, built from every public Victorian voting-intention poll. It starts from the raw poll table: fieldwork dates, pollster, sample size, and first preferences for Labor, the Coalition, the Greens and One Nation.
From there, every poll gets a weight. Recent polls count for more: only the last 60 days are used, and older polls fade out smoothly rather than dropping off a cliff. Bigger samples count more too, so a poll of 2,000 people outweighs one of 500.
Two more checks stop any one source taking over. If a pollster releases far more often than the rest, each of its polls is scaled down, so the average reflects the spread of pollsters rather than whoever polls most. And a poll that sits well away from all the others gets less say, so one odd result can’t drag the average around.
I don’t currently adjust for house effects, the tendency for some pollsters to lean one way or the other. Ideally I would. But there aren’t enough Victorian polls yet to measure those leans with any confidence, and I’d rather treat every pollster equally than build in guesses that could make the average less accurate.
Those statewide numbers don’t tell you who wins each seat on their own. That’s where the seat model comes in: it simulates 10,000 elections across all 88 seats, and that’s where the numbers below come from.
What it means for seats
Run those numbers through the seat model and nobody gets to the 45 seats needed for a majority. The Coalition is the clear frontrunner for most seats, but it’s well short of governing alone.
| Party | Seats (middle result) | Likely range |
|---|---|---|
| Coalition | 38 | 27–50 |
| Labor | 26 | 14–40 |
| One Nation | 14 | 3–32 |
| Greens | 5 | 4–7 |
| Independents | 2 | 0–4 |
The likely range covers the middle 80% of my simulations, so one time in ten a party does worse and one time in ten it does better. One Nation’s range is huge because so many seats hinge on how big its vote ends up being. The Greens’ range is tight because their seats are concentrated in a handful of inner-Melbourne contests.
Across the simulations, nobody wins a majority about three times in four. The Coalition wins one outright about 23% of the time, and Labor only about 3%.
Three seats have swapped favourites since the last update, all of them close. Bellarine now leans Labor, while Pakenham and Mildura now lean One Nation. Mildura is a genuine three-way race, with One Nation ahead on just a 35% chance.
What could still move it
The statewide numbers haven’t changed much, but there are still a few things that could make a real difference before election day.
The big one is the how-to-vote cards. Most of One Nation’s likely wins are seats where the final count is One Nation against Labor, so where the Coalition puts One Nation could swing a handful of seats either way. The other cards count too: whether the Greens and Labor swap preferences, how the Liberals and Nationals preference each other in three-cornered contests, and where One Nation sends its own preferences when it’s knocked out.
I’m also watching One Nation’s best seats. In South Australia, the model was a few points too low in One Nation’s strongest areas. If that happens again, One Nation could win more seats than I show, and the Nationals heartland is where I’d look first: Gippsland East, Murray Plains and Shepparton all have One Nation above 36%, but the Nationals are still favoured.
If the Greens keep polling 15 or more, a few of their inner-Melbourne seats firm up.
And then there are the independents, always the hardest part to call.
I’ll update the baseline as new polls land, and rerun the seat model when the how-to-vote cards are released. Those cards could matter more than the next small movement in the polls. You can see every seat on the interactive map.
Thanks for reading, Caleb