Another prime minister in 1% of simulations, most often Naftali Bennett, who leads B'Yachad. B'Yachad finishes first in 6% of simulations. Finishing first confers no right: after consulting the elected groups, the President asks the member most likely to succeed to form the government.
| List | enters the Knesset | votes | seats | range |
|---|---|---|---|---|
| > 99% | 18.7% | 23 | 14-38 | |
| > 99% | 16.9% | 21 | 11-37 | |
| 98% | 9.7% | 11 | 5-22 | |
| 99% | 6.9% | 8 | 5-14 | |
| 98% | 6.8% | 8 | 5-14 | |
| 98% | 6.7% | 8 | 5-13 | |
| 97% | 6.0% | 7 | 4-12 | |
| 97% | 6.1% | 7 | 4-12 | |
| 91% | 5.5% | 6 | 0-11 | |
| 90% | 5.1% | 6 | 0-10 | |
| 80% | 4.5% | 5 | 0-9 | |
| AIAmcha Yisrael | 34% | 3.0% | 0 | 0-7 |
| HHendel-Zelekha | 28% | 2.8% | 0 | 0-6 |
A list below 3.25% gets no seats: its votes are lost. Five lists stand between 3 and 6%: much of the election is decided there.
Lists whose entry into the Knesset is not assured. A list that stays out loses all its seats, which are shared among the others.
Each simulation replays, in equal shares, the polling error of one of the five elections from 2019 to 2022. Choosing an election keeps only the simulations that replay its error. Seat differences are measured against the forecast.
-
Tick lists: the page counts the share of the selected simulations in which their combined seats reach 61. This is arithmetic: it does not say whether these lists would agree to govern together.
The other lists stayed above 97% at every date: Yashar, Likud, The Democrats, Shas.
| List | votes | seats | in 2026 |
|---|---|---|---|
| Likud | 23.41% | 32 | Likud |
| Yesh Atid | 17.79% | 24 | B'Yachad |
| Religious Zionism | 10.84% | 14 | Otzma Yehudit, Religious Zionism-Zehut |
| National Unity | 9.08% | 12 | Blue and White, not modelled |
| Shas | 8.25% | 11 | Shas |
| United Torah Judaism | 5.88% | 7 | United Torah Judaism |
| Yisrael Beytenu | 4.48% | 6 | Yisrael Beytenu |
| Ra'am | 4.07% | 5 | Ra'am |
| Hadash-Ta'al | 3.75% | 5 | Hadash-Ta'al-Balad |
| Labor | 3.69% | 4 | The Democrats |
| Meretz | 3.16% | 0 | The Democrats |
| Balad | 2.91% | 0 | Hadash-Ta'al-Balad |
| Jewish Home | 1.19% | 0 | none |
| New Economic Party | 0.29% | 0 | Hendel-Zelekha |
Benjamin Netanyahu's camp won 64 seats. Yashar and Amcha Yisrael did not exist. Benny Gantz's National Unity has broken up: Blue and White runs alone, New Hope joined Likud and Gadi Eisenkot founded Yashar.
| Election | Netanyahu camp at 61 | seats forecast | seats won | majority |
|---|---|---|---|---|
| April 2019 | 84% | 64 | 65 | yes |
| September 2019 | 12% | 59 | 55 | no |
| March 2020 | < 1% | 57 | 58 | no |
| March 2021 | < 1% | 50 | 52 | no |
| November 2022 | 35% | 61 | 64 | yes |
The model rated the actual outcome as the most likely in 4 elections out of 5. Its Brier score, which measures the gap between stated probabilities and what happened, is 0.092; it would be 0.240 by always stating the average frequency of majorities. Lower is better.
A forecasting model is not one more poll. A poll surveys a sample and describes opinion at a given moment. A model surveys no one. It gathers what can be measured, published polls, past election results and the campaign context, and derives a distribution of possible outcomes with their probability. It does not say that a list will win 22 seats. It says within which range it will finish, and how likely each outcome is.
Two pollsters published on the same day can differ by several seats. The model aggregates every poll published since the lists were filed, giving more weight to the most recent, the largest and the pollsters that erred least in the past.
The model does not rest on polls alone. It also starts from the November 2022 election result, list by list, following the mergers and splits since then. New lists have no such anchor. Finally, it can include Benjamin Netanyahu's lead on the question of the best prime minister. The elections from 2019 to 2021 show how that lead translated at the ballot box, beyond what the polls predicted. This measure only counts if a poll has published it in the last seven days. No poll has done so over that period, so it does not count in today's forecast. The closer the vote, the more the polls weigh. When a source provides no reliable measurement, it does not count, and its share goes to the others.
Israeli polls erred at each of the five elections from 2019 to 2022. Four times out of five, they underestimated Benjamin Netanyahu's camp. The model replays these real errors, in their actual direction, on today's polls. This is what gives each list its range and each outcome its probability.
A single forecast would give a false sense of certainty. The model therefore replays the election 20,000 times, varying what is genuinely uncertain: polling error, the gap between pollsters, each list's own score. The probability shown is the share of simulations in which the event occurs.
Each simulation allocates the 120 seats under Israeli electoral law: 3.25% threshold, Bader-Ofer method and surplus agreements. Who governs is then read from the refusals to govern together that list leaders have stated publicly during the campaign.
Uncertainty is this model's product, never its flaw. The computation is exact and reproducible, and the model was replayed on the five elections from 2019 to 2022 without any setting being changed afterwards.
Results: Central Elections Committee. Polls: pollsters and broadcasters, reports filed with the Committee's register. Past elections: public compilations of 2019 to 2022 polls. Statements by list leaders: Jerusalem Post, Times of Israel, Ynet, Israel National News.
All rights belong to Le Millénaire SAS, 47 rue de Vivienne, 75002 Paris.
Model designed by
Project director: William Thay
Deputy project director: Matthieu Hocque
Political and data analysis: Antranig Kevorkian and Pierre Clairé