1 · Data and the 2022 baseline

Primary votes are the VEC’s 2022 Legislative Assembly results for all 88 districts, reconciled against the VEC’s per-district formal-vote totals. 58 districts matched exactly; thirteen were corrected, nine of them out by an exact round thousand (Melton, Broadmeadows, Cranbourne, Lowan, Mildura, Bass, Pakenham, Shepparton, Point Cook), a keying signature rather than rounding. Malvern’s whole row was the pre-recheck count and was replaced with the rechecked figures, and Mulgrave’s independents (Ian Cook and three others) were 2,047 votes short. Every district now matches the VEC formal vote to within 50 votes; the statewide total is 3,654,121.

The model has fourteen parties, each on its own 2022 ballot figures: Labor, the Liberals, the Nationals, One Nation, the Greens, Animal Justice, the Victorian Socialists, Democratic Labour, Legalise Cannabis, Family First, Freedom, the Libertarians, the Shooters, and Independents. Parties that have since been deregistered (Justice, New Democrats, Angry Victorians, Health Australia, Transport Matters, Companions and Pets) are shown as Other, split out of the independents district by district from the VEC’s counts (except Narracan, Pakenham and Sydenham, where they stay with the independents); the Reason Party’s 2022 vote is counted with the Greens. Two parties new in 2026, the Socialist Alliance and the West Party, come in with the projection only (section 6). With nothing applied the page shows the 2022 result: Labor 56, Liberal 19, Nationals 9, Greens 4.

Narracan*’s 2022 result is the January 2023 supplementary election: the November 2022 poll there failed when the Nationals candidate died, and Labor did not stand at the supplementary. On the 2022 view the seat shows that result, Liberal 63.0 v independent Tony Wolfe 37.0, with Labor’s vote passed on by its preferences. Labor is standing in 2026, so the projection starts from a notional 2022 vote that puts Labor on its 2018 primary (35.5%) moved by the 2018–22 statewide Labor swing (−6.2), so 29.3%, with every other party scaled down to make room.

By-elections since 2022 are noted on their seats but do not replace the 2022 general-election baseline: Warrandyte (Aug 2023, Liberal hold), Mulgrave (Nov 2023, Labor hold), Prahran (Feb 2025, Liberal gain from the Greens), Werribee (Feb 2025, Labor hold) and Nepean (May 2026, Liberal hold). Prahran is the one where this matters: the 2022 baseline has the Greens holding it.

2 · District classes

Every district has four classifications. Geography and the Suburban Rail Loop were carried over from the original scenario model and checked against the 2021 Census and the SRL East station areas; diversity and income are set from the 2021 Census (see the Swing Lab below): geography (Inner, Suburban, Outer Suburban, Regional, Rural, Tree change), diversity (European, Diverse, Chinese), economic (higher or lower income) and whether it sits on the Suburban Rail Loop corridor. They are shown in each seat’s hover. The poll average does not use them (it moves every district the same way, as on the Australian model); geography sets which preference flows apply, below.

3 · How the count works

Each district is counted from first preferences by full sequential exclusion: the lowest party is excluded, its votes go to the parties still standing, and this repeats until two remain. The three-candidate count shown in the hover is the state of play when three were left.

Where excluded votes go

A ballot goes to whichever survivor the voter ranks highest, so a party’s share of an excluded pile is the probability of beating every other survivor: the product of its pairwise flows, not their sum. Summing them handed the Coalition a third of Greens ballots while Labor was still in the count; the product gives a metro Greens vote in a Labor/Coalition/One Nation three-way roughly 85 / 13 / 2.

On top of that, 9% of each excluded pile scatters evenly across the survivors whatever the party’s ordering: donkey votes, voters who number every box without caring, ticket confusion. This puts that same Greens three-way at about 80 / 15 / 5, which matches how these counts look in practice.

Final two

The two-candidate result is the direct pairwise transfer of every other party’s vote between the final two, using the preference chart. With nothing applied, each district is corrected by the gap between the chart’s count of the 2022 vote and the real 2022 two-candidate result, so every 2022 result is exact. No projection carries that correction, on either basis: both count on the chart’s flows alone (until 27 October 2026 the 2022 basis carried it). Narracan is pinned to its 2023 supplementary election on the 2022 view only.

With nothing changed on the 2022 view, the hover’s four- and three-candidate counts are the VEC’s actual 2022 counts. Once anything is moved (the poll average, a party figure, the Swing Lab, a preference flow or another basis) they are the model’s own count.

4 · Preference flows

Flows vary by location only, never by diversity, income or the rail corridor, and in three tiers: Inner Melbourne; suburban and outer suburban; regional, rural and tree-change. A district with two geographies (Preston is Inner and Suburban, Bass Regional and Outer Suburban) takes the average of its two tiers, as it takes half of each in the Swing Lab. Labor v Coalition, Labor v Greens, Labor v One Nation, Coalition v One Nation and Coalition v independent have location variants; every other contest is the same statewide. All defaults are multiples of five, and the spread between tiers for the same flowing party is at most 15 points, except independents in a Labor v One Nation count (70 / 60 / 50).

Coalition preferences in a Labor v One Nation count started from the DemosAU MRP’s respondent rankings (60 / 69 / 79 to One Nation from the inner city outward), compressed to the 15-point spread. On 23 October 2026 they moved 5 points towards Labor (55 / 65 / 70 to One Nation; on 24 Oct 2026 the inner city went to 60, so now 60 / 65 / 70). In the same contest the Greens now send 90% to Labor everywhere, and independents 70% in the inner city, 60% in the suburbs and 50% in the regions (65% everywhere until 26 October 2026). RedBridge and Roy Morgan publish two-party figures from their respondents’ own preferences; run on each poll’s own primaries, the chart as it stands on 24 October 2026 comes within about a point of them on average (Labor v Coalition 1.0 point kinder to Labor than the polls, Labor v One Nation 1.1 points kinder to One Nation), across RedBridge’s February, June and October polls and Roy Morgan’s August poll. One Nation’s preferences go 80% to the Coalition over Labor everywhere: a little above its actual 2025 federal flow in Victoria (75%) and RedBridge’s 77%, since today’s much larger One Nation vote draws heavily on former Coalition voters (RedBridge’s October primaries and two-party figure imply about 87%). On 24 October 2026 the chart was checked against the AEC’s actual 2025 Victorian flows: the Greens now send 90% to Labor over the Coalition (89.7% in 2025), independents and others 55% (67–70% in 2025, but that came mostly from outer-suburban independents; the teal voters of Kew and Hawthorn are largely former Liberal moderates, and country independents draw on Coalition voters too; since 26 October 2026 55% in the inner city, 50% in the suburbs and 45% in the regions, against the VEC’s 2022 67 / 45 / 39), One Nation 70% to Labor over the Greens (71%), and in a Coalition v independent count the Greens 85%, Labor 80% and One Nation 30% to the independent (88 / 79 / 31). Between Labor and the Coalition the right-wing minor parties now split as they did in 2025: Family First 65%, Freedom 70%, the Libertarians 75% and the Shooters 70% to the Coalition (the Shooters 75% until 26 October 2026; 2022 about 60%), the Shooters 70% to One Nation over Labor too. On 26 October 2026 seven flows were moved part of the way towards what the VEC’s 2022 counts measured: the Socialists 85% to Labor over the Coalition (was 90%; 2022 about 66%, 61–74% in the western suburbs); Family First, Freedom, the Libertarians and the Coalition 65% to Labor over the Greens (was 75%, and 70% for the Coalition; 2022: 39%, 39%, 45–61% and 35%, when the right put Labor last); Family First and Freedom 80% to the Coalition over the Greens (was 85%; 2022: 63–72%). Most of these flows stay kinder to Labor than 2022 measured, on purpose: voters of parties that do not back the Coalition or One Nation are taken to be less angry with Labor than at the last election, and right-wing voters who like the Coalition or One Nation are more likely to vote for them directly this time, which leaves fewer hard anti-Labor preferences in the minor parties’ piles. RedBridge’s June three-way test (which of Labor, the Coalition and One Nation Greens and minor-party voters rank highest) comes out within a point of the chart. The right-wing minor parties send 75% of their preferences to One Nation in every contest One Nation is in, everywhere (Democratic Labour is not counted with them and keeps its own flows). That is a little above the 2025 federal seats where One Nation made the final two (Hunter and Maranoa: 59–76%, about 68% overall), allowing for One Nation’s much larger vote now. They were set at 95% and then 90% earlier on 24 October 2026; with five parties to choose between, those put about 85% of a non-standing party’s vote into One Nation, against about 62% now. The independent pairings are sourced under Known limitations: Labor v independent from the AEC’s 2025 counts, the others derived through a third party and checked against real counts. Every value can be changed in the Preference flows tab.

5 · Poll average

The Apply poll average button moves every district’s vote to a weighted average of the latest Victorian state polls before counting. Only the most recent poll from each company is used, as long as it ended within 90 days of the newest poll. Each poll is rescaled to 100% and weighted by pollster tier (Newspoll and YouGov 1.0; RedBridge and Resolve 0.8; DemosAU and Freshwater 0.6; Roy Morgan and other firms 0.4) times recency, which halves every 14 days (). The exact rule is under “Polls used and weights”.

A server job reads Wikipedia’s poll table every hour. A new poll is used once its cited article confirms its figures, or once it has stood unchanged on Wikipedia for 3 hours. The page falls back to the polls built into it if that file can’t be read.

Every party is fitted on its own. The polls’ L/NP is shared between the Liberals and Nationals by their 2022 statewide votes. Not standing. A party or independent that stood in a district in 2022 and is not standing there in 2026 (or has not announced yet) keeps its 2022 vote there, swings with everyone else, and then goes first in the count, before anyone else, passed on by its own preferences: a party by its own column of the chart, an independent by the independents’ column. Nobody passes votes to them. A new candidate’s starting vote comes out of theirs first, so they shrink as candidates are announced. Other (the white card) is the parties since deregistered (Justice, New Democrats, Angry Victorians, Health Australia, Transport Matters, Companions and Pets: 0.5% of the state in 2022). They show on the 2022 view and count there as the independents did; with the swing applied they are deleted, everyone else filling the room. A figure typed into the card puts a vote back, spread evenly, which exhausts in the count. The polls’ Others is shared over the independents and every minor party, standing or not, in proportion to their starting votes, and the swing moves them all alike, returning independents included: named independents standing again from their 2022 vote, new independents from 1.5%, new parties from 2.5%. Every party’s statewide primary matches the poll average exactly. Labor, the Greens, One Nation and one Coalition party are assumed to stand in every district; one that would otherwise start from exactly nothing gets a 0.5% notional base, except One Nation on the 2022 basis, which starts from its 2022 median vote, 4.6%, outside the six districts it contested in 2022 (see 2022 state basis below). The swing method sets how each party’s change is spread:

• Uniform: each group gains or loses the same number of points in every district.
• Logit (default): the same shift is added to each group’s log-odds in every district, so moves are largest where a party sits mid-range and taper towards 0% and 100%. It lands between uniform and proportional and can never push a vote below zero.
• Proportional: each group’s vote is multiplied by the same factor everywhere, so parties grow most where they were already strong.

Each district is rescaled to 100%, and the swing is solved so the statewide primaries match the average exactly. The ▲/▼ arrows and the boxes on the party cards (and on each Liberal, Nationals and minor-party part) move a statewide primary from there; the change comes out of, or goes into, Independents and the minor parties.

Two starting points

The swing to the poll average can start from either of two maps of where each party is strong, chosen with the buttons beside the swing method:

• 2022 state basis: the 2022 vote as cast, and nothing from 2025. One Nation stood in only a handful of districts in 2022 and polled 0.3% statewide. It starts from its 2022 vote in the six districts where it stood (Morwell, Bendigo East, Monbulk, Narracan, Macedon and Pakenham) and, everywhere else, from its median vote across those six, 4.6% (a typical district it contested), taken out of Labor, the Coalition and the Greens in proportion, as the Senate basis gives it its Senate share of the four; minor parties and independents keep their 2022 votes on both bases. The swing then takes it to its poll figure like every other party. The 2025 Senate basis is the one with One Nation’s statewide map. Like the Senate basis, a projection on this basis counts on the chart’s flows alone, with no 2022 calibration.
• 2022 Council basis: the 2022 Legislative Council first preferences in each district (the VEC’s district totals) say where Labor, the Coalition, the Greens and One Nation are each strong, used exactly as the Senate basis uses the 2025 Senate: the four’s Council shares of each other share out each district’s 2022 four-party Assembly vote, the Liberals and Nationals divide the Coalition’s share as they did there in the Assembly, and minor parties and independents keep their 2022 Assembly votes. One Nation ran a Council ticket in every region (2.0% statewide), so it starts from its own Council vote in every district rather than from a median. Nothing from 2025, and no 2022 calibration.
• 2025 Senate basis (what Apply poll average uses unless you pick another). The AEC’s 2025 Senate first preferences, booth by booth, are put onto the 2022 state districts (each booth’s votes split across census SA1s the way its House voters were, then SA1s into districts by population). From those we take the first preferences of the Greens, Labor, the Coalition and One Nation as shares of the four, and those shares split each district’s 2022 vote for the four. Nothing else in the Senate count goes in: other Senate parties’ votes are left out, whether or not a state counterpart is standing (the minor parties’ and independents’ own 2022 state votes stand for those voters, so counting their Senate votes too would count them twice). Postal and other declaration votes can’t be placed by booth and are left out. Until 26 October 2026 Labor, the Coalition and the Greens were first moved onto a curve to their 2022 statewide state levels (as in Alex Jago’s calculator), and Senate parties with a state counterpart not standing were passed to the four by their preferences; the curve was a second reshaping on top of the Senate’s own map, and the poll swing sets every party’s level afterwards anyway. One Nation was always on its own Senate share: moved to its 2022 state level (0.3%, when it barely stood) its Senate geography had disappeared. The Liberals and Nationals divide the Coalition’s share as they did there in 2022; minor parties and independents keep their 2022 votes. The Senate has no local candidates, so this basis carries no 2022 calibration and no incumbents’ personal votes. With uniform swing it makes little difference, since a uniform swing adds the same points everywhere; logit and proportional carry its geography through. Picking this basis with the poll average off shows its starting map rather than the 2022 result: the four on their 2025 Senate shares of each other within each district’s 2022 four-party vote (One Nation about 4.2% statewide), and minor parties and independents on their 2022 votes. Press Reset to 2022 result to see the 2022 result itself.

6 · Candidate assumptions

Retiring members

This applies only with the poll average on, so the 2022 baseline stays the 2022 result. (The scenario model also cut the independent vote by hand in nine seats; those cuts were taken out on 23 October 2026 in favour of the button below.)

• Retiring members: fifteen sitting members are not recontesting. Labor: Bass, Bayswater, Essendon, Macedon, Melton, Pakenham, Sydenham. Coalition: Benambra, Croydon, Gippsland East, Malvern, Murray Plains, Rowville, Shepparton. Greens: Brunswick. Nothing is taken off their party’s vote by default; the Swing Lab’s Retiring member slider moves it by points (until 24 October 2026 a fixed ×0.925 was applied).
• Sophomores: 28 members face their first re-election, elected new in 2022 or at a by-election since (Warrandyte, Mulgrave, Narracan, Prahran, Werribee, Nepean). The Swing Lab’s Sophomore slider moves their party’s vote; 0 by default.

Parties not standing again

Assume all minor parties run (off by default) counts every minor party that stood in a district in 2022 as standing there again until told otherwise. Most hadn’t announced candidates when this was written, so it is the other extreme from reading the list as final: the minor parties’ part of the polls’ Others is then shared over all of them, and One Nation starts from its Senate vote alone, without the right-wing minor parties’ vote passed to it. With it off, any minor party, left or right, that stood in a district in 2022 and isn’t on the 2026 candidate list there (yet) keeps its 2022 vote there, swings with everyone else and is then passed on first in the count by its own preferences (section 5). This only sets the starting votes the swing works from; the poll average is still the target for every party.

Mixed districts

Each district has a geography (Inner, Suburban, Outer Suburban, Tree change, Regional, Rural), a background (European, Diverse, Chinese) and an income band. Districts that straddle two take both, as in the federal model, and take 50% of each class’s slider in the Swing Lab (the button beside Preference flows). The lab moves One Nation, the Coalition (together or each party alone), the Greens, the Socialists and independents by geography, background, income, social profile, housing tenure, workforce and the Suburban Rail Loop, in points of the district vote, on top of whatever the page is showing; Labor balances what actually moves in each district. It also holds the retiring-member penalty. The seat hover lists both. Geography: the coastal and bushland fringes of Geelong, the Peninsula and the Yarra Ranges, growth corridors that run out into country towns, inner seats that run into the middle ring, and regional seats with a rural hinterland (26 districts). Background follows one rule on 2021 Census ancestry: where 44% or more of ancestry responses are neither European nor English-speaking (Chinese, Indian, Vietnamese and the rest; Greek, Italian and other European ancestries count as European) a district is Diverse; 30–44% is half Diverse (or Chinese) and half European; under 30% is European; and it takes a Chinese half wherever Chinese ancestry is 12% or more (18 districts are mixed). Income comes straight from 2021 Census median household income: higher at 110% of the state median or more (30 districts), lower under 90% (30), middle in between (28), which take no income swing, as in the federal model. A district with 4% or more public housing is lower income only under 85% of the median, since the public-housing slider already carries it. Housing tenure uses the 2021 Census as the federal model does: each district takes its largest of owning outright, mortgage and renting, the next if within 5 points of it, and so on, plus public housing where that is 4% or more of households; its tags share the slider equally. Workforce is not exhaustive: a district is tagged only where it stands out from the state (white collar at 1.2 times the state share of employed people, 22 districts; blue collar at 1.3 times, 20; pink collar at 1.08 times, 17; unemployment at 1.3 times the state rate, 14). The federal model’s small-business-owner tag has no census measure here. The social profile uses each district’s 2023 Voice referendum No vote, as the federal model does: the AEC’s booth results placed on the 2022 districts through census areas (statewide this comes to 54.3% No, against the actual 54.15%). Progressive under 50% No (27 districts), moderate 50–60% (33), conservative over 60% (28); the federal model’s four tiers left most Victorian districts in the middle two.

Which Coalition party is standing

With the poll average on, the projection follows the Coalition candidates on Wikipedia’s 2026 list. Where a Coalition party stood in 2022 but isn’t standing now, its 2022 vote is shared out among the parties standing by its preferences before any swing, by the Coalition partner column: about two-thirds to its partner, the rest leaking as it would in a real three-cornered contest. That is the Liberals in Bendigo East, Bendigo West, Euroa, Mildura, Morwell and Shepparton, where only the Nationals are standing, and the Nationals in Bass, where only the Liberals are. A Coalition party standing where it had no 2022 vote (the Nationals in Bendigo East and West) starts from its partner's shared-out vote. Ripon is a new three-cornered contest: the Liberals, who stood there in 2022, keep the Coalition vote and the Nationals newcomer starts from none, with nothing invented for them. Districts with no Coalition candidate listed yet (Brunswick, Lara and Pascoe Vale when checked) keep 2022’s arrangement.

Independents standing again, or not

Independents not standing again. By default (with the poll average on; the 2022 view is always the 2022 result) the model takes out every independent who polled 2% or more in 2022 and is not on Wikipedia’s list of 2026 candidates (checked ; nominations had not closed, so this means “not announced”), and their 2022 vote swings with everyone else and is then passed on first in the count by the independents’ column of the chart (section 5). The independents standing again don’t take over the departing ones’ share. Assume all independents run (off by default) keeps them all in, as the counterpart of Assume all minor parties run.

New independents. Every independent on the 2026 candidate list who wasn’t a 2022 independent of 2% or more starts from a notional 1.5% of their district’s vote, taken first from the vote of those not standing there, before the swing, and the swing moves them with the other independents and minor parties; the returning independents start from their own 2022 vote and swing the same way. Where a seat has more than one independent, the one with the larger vote leads the count and the others’ votes go to them first by the independent-to-independent flows on the preference board (the Independents column of each “v Independents” row); until 26 October 2026 a returning independent always led, however small, and those flows were fixed in the code. The 1.5% is the median 2022 vote of every independent Assembly candidate who wasn’t a sitting MP (96 candidates in 59 districts; Narracan’s count was the supplementary election and is left out): the typical new independent. Their mean was 3.9% (used until 26 October 2026), carried by eight organised community and teal campaigns that polled between 12% and 32%; a well-organised campaign will do far better than the median. Bernard Quince (Ripon) polled 0.7% in 2022, under the 2% cut, and is standing again; he starts from his own vote.

New parties. Two parties standing in 2026 did not stand in 2022: the Socialist Alliance (six districts) and the West Party (eight, in the western suburbs). Each starts from a notional 2.5% where it stands, the median 2022 vote of all 336 minor-party Assembly candidates (mean 2.9%), and the swing moves it with every other independent and minor party. In Werribee the West Party’s candidate is Paul Hopper, an independent with 2,278 votes in 2022 who is not new, so there the party starts from his 2022 vote instead. Their preferences follow another party’s column on the preference board, marked in its heading: the Socialist Alliance the Victorian Socialists’, the West Party the independents’.

Candidate lists. Candidates come from Wikipedia’s 2026 candidates page, plus any the ABC’s election guide lists that Wikipedia does not (checked 26 October 2026: five independents, seven minor-party candidates and the two new parties). Where the two disagree the page keeps Mathew Clarke in Melbourne and Family First in Lowan.

7 · Legislative Council

The Council has eight regions of five members, each made up of eleven Assembly districts. Victoria abolished group voting tickets for 2026, and the Council is counted here as plain quotas: a Droop quota of 100/6 = 16.67%, full quotas filled first, then the largest remainders. Independents and ungrouped candidates are left out of contention. A full sequential STV count was built and deliberately not used, because without the parties’ own how-to-vote flows the extra machinery adds precision the inputs don’t have.

Each region’s vote starts from the Assembly projection for its eleven districts. It is then moved by how the region voted differently in its 2022 Council ballot from its 2022 Assembly ballot, so a party that does better in the upper house (Legalise Cannabis, Animal Justice, the Libertarians) keeps that edge. The Liberals and Nationals are pooled as one Coalition, as they run joint Council tickets. Independents’ surplus above each region’s 2022 level is spread across the parties. Two exceptions. A minor party that isn’t standing in every one of a region’s Assembly districts still runs a Council ticket there, so it takes its 2022 Council vote, moved as the polls’ Others has moved since the 2022 Assembly, rather than an Assembly vote that has been passed on to other parties where it isn’t standing. And One Nation doesn’t get the 2022 Council-over-Assembly edge: that came from its having almost no Assembly candidates in 2022, and it stands everywhere now.

With nothing applied on the 2022 view the Council is pinned to the seats actually won in 2022, under the old group voting tickets that let several minor parties win from tiny primaries; counted as plain quotas the same votes would give a different split. Once anything is applied, the quota count takes over. The seat changes on the Council cards are against the 2022 result itself (Labor 15, Coalition 14, Greens 4, Legalise Cannabis 2, Animal Justice 1, Democratic Labour 1, Libertarian 1, One Nation 1, Shooters 1).

8 · Display conventions

• The Liberals and Nationals are separate parties in the Assembly count; the Coalition card shows both. The preference chart has one Coalition, whose rows both use. How a Liberal or Nationals vote treats its partner is the Coalition partner column of the preference chart (since 26 October 2026; a fixed 80%, 65% over One Nation, before): 80% to the partner over Labor, 90% over the Greens, 75% over an independent and 70% over One Nation, from the VEC’s 2022 counts where one was excluded with the other still in (87–89% against Labor in Euroa and Morwell, 69% in Bass, 77% against an independent in Mildura and Shepparton; nothing measured against the Greens or One Nation). In a Liberal v Nationals count every other party splits evenly, except One Nation, Democratic Labour and the minor-right tickets, which lean 55% to the Nationals.
• Hexes deepen from pale (a line-ball result) to the party’s full colour at a 16-point margin. A gold ring marks a seat changing hands since 2022.
• Parties run left to right by the site-wide rule: a 45° line through Victoria’s vote-weighted average party on the compass.
• Thick black lines on the Council map are the boundaries of its eight regions.
• The district panel (hover or tap a hex). With the poll average on, its primaries are the votes after the parties and independents not standing in that district have been passed on (section 6), so none of them has a row; they are listed in one line underneath. Every independent standing has a row under their own name: the one with the larger vote also carries what the independents’ column picks up from those passed on, and a new independent’s change reads “new”. On the 2022 view the rows are the 2022 result as cast, and a party or independent not standing in 2026 is marked “not standing”.

9 · Known limitations

Things the model does not do well, stated plainly.