Method
The list that came back, and left a bloc
Three polls arrived in two days, and one of them seated the Reservists — a list this page reported on Sunday as having left the ballot. It is back with a new partner, and both sources now count it outside every bloc. The printed table is level at 54 and 54, with 12 for the Arab parties.
The front of this page reads Opposition bloc and Netanyahu bloc are close, at about 54 and 54 seats, and Arab parties with 12, from 21 polls. Two days ago it read 55 and 53. Three polls arrived in between — and one decision that is not a poll at all.
Three polls
Kantar for Kan 11, fielded 6 September, sample 551: Yashar 23, Likud 20, Together 13, the Democrats 8, United Torah Judaism 8, the Joint List 7, Otzma Yehudit 7, Shas 7, Yisrael Beiteinu 7, Ra’am 6, Religious Zionism 6, Amcha Israel 4, the Reservists-New Economic Party 4. Blocs: 52 Netanyahu, 51 opposition, 13 Arab parties — and 4 in no bloc, which is the rest of this piece.
Filber for Channel 14, fielded 7 September, sample 1,057: Likud 30, Yashar 22, Shas 10, the Democrats 10, the Joint List 8, Otzma Yehudit 8, Religious Zionism 8, United Torah Judaism 7, Together 6, Yisrael Beiteinu 6, Ra’am 5. Blocs 63, 44 and 13.
Midgam R&C for Channel 12, fielded 7 September, sample 503: Yashar 25, Likud 21, Together 13, the Democrats 11, United Torah Judaism 8, Yisrael Beiteinu 8, the Joint List 7, Shas 7, Otzma Yehudit 6, Religious Zionism 6, Amcha Israel 4, Ra’am 4. Blocs 52, 57 and 11.
Filber’s 63 is the highest Netanyahu figure of the twenty-one polls the model reads, and the only three readings that high in the window are all Filber’s. Kantar’s 52 and Midgam’s 52 are the median of the twenty-one. The model reads all three the same way and adjusts none of them.
The list that came back
Two days ago this page reported that Hili Tropper had left Zionist Home for Yashar, that the half he left behind — Yoaz Hendel’s Reservists — had merged with Yaron Zelekha’s New Economic Party, and that no poll had been fielded since either change. It named two lists the forecast was still reading as though they were running: Unity, which withdrew on 4 September, and Zionist Home.
One of the two has resolved, and not the way it read. The Reservists did not leave the ballot. They changed a partner and came back, and Kantar gave the merged list four seats on 6 September, the first figure it has been given under its new name. Filber and Midgam then put it at 2.1% and 2.7%: below the threshold, but asked about.
Its chance of four seats or more went from 3.7% to 6.5%, and its chance of any seats at all from 8.3% to 14.9%. The printed table still gives it nothing, because a median below 4 is below what the threshold allows.
Unity is still a phantom. Nothing has asked about it since it withdrew, all 21 polls the model reads predate that, and it takes four seats or more in 0.5% of simulations. Candidate lists were filed with the Central Elections Committee today, between 4 PM and 10 PM — after this run was made. That is where the last of it goes.
Which bloc it sits in
Until Sunday the column was Tropper and Hendel, and the poll compilation coded every row of it as opposing Netanyahu. This forecast put it in the opposition bloc on that reading, and has printed it there all campaign.
The merged list is coded differently, by both sources that code blocs at all. The compilation files all three of its readings as unaligned while leaving the older rows where they were, which is a distinction drawn rather than a coding dropped. And Haaretz reported the Kan 11 poll as the Netanyahu bloc with 52 without Hendel and Zelekha and Gadi Eisenkot’s bloc with 51 — 52, 51, 4 and 13, a full chamber with this list outside every bloc.
So it is outside every bloc here too, from this run onward. The mapping file holds one membership per election and the election is October, so the ballot decides it.
What that costs, in both directions, because one entry cannot be right about both ends of a campaign. Ahead: the list’s median is nothing, so the printed table does not move, but P(Opposition ≥ 61) falls from 7.7% to 6.4% and P(Opposition + Arab parties ≥ 61) from 92.4% to 90.1%. Behind: a 2025 reading of Tropper-and-Hendel really was an opposition reading, and the monthly averages restate — by a tenth of a seat or less in most months, by 1.7 seats in September 2025 and 1.2 in July 2026. Inside the window the model reads, exactly two of the twenty-one polls give this list a seat: Yossi Tatika on 3 September, when it was still Tropper’s, and Kantar on 6 September, when it was not.
The change also reached something with nothing to do with the ballot. Polling firms are compared with one another here on their bloc profiles, so redefining a bloc redefines every profile — and Yossi Tatika’s nearest firm moved from Panel Project HaMidgam to Kantar, the two now 0.02 seats apart with 34% of resamples agreeing. The record of which firm Tatika resembles has been removed for failing its own support test: the second such removal, and the first not caused by a poll. Nothing in the forecast reads that file, which is switched off. It is written down because a mapping edit reaching that far is worth seeing.
None of this is a claim about who Hendel or Zelekha will recommend for prime minister. Neither has said, and when either does, this is the first entry to revisit.
Where that leaves the order
The printed table is 54 for the Netanyahu bloc, 54 for the opposition and 12 for the Arab parties, level where it read 53 and 55 two days ago. The headline says nobody leads, and the simulations agree: the opposition is the larger bloc in 53.7% of them, from 59.2%.
Neither is near 61. The Netanyahu bloc reaches it in 4.3% of simulations and the opposition in 6.4%. The opposition and the Arab parties reach 61 between them in 90.1%, which is arithmetic about seats and nothing more: the 61 would come from six lists that do not act as one, and no coalition is modelled here.
Above the blocs, almost nothing moved. Yashar is the largest list in 51.8% of simulations, Likud in 40.1%, and they are level in 7.7% — 50.7%, 40.9% and 7.9% two days ago.
The threshold is where the round did its work. Religious Zionism, running with Zehut, is printed at 5 again, which it last did on 27 August: the three new polls seat it at 6, 8 and 6, and its chance of four seats or more went from 59.8% to 67.9%, its chance of nothing from 24.3% to 19.4%. Amcha Israel stays at 4, with 52.3% from 55.0%. Those two are still the lists whose fall would take most off the Netanyahu bloc, and they have moved apart: one is now clear of the coin toss and the other is still on it.