From today to polling day in the Netherlands
Abstract. We test the British pull-back towards the previous result on Dutch polls at ten distances up to three years before the elections of 2012 to 2025. Unlike in Italy it helps: the best curve keeps a little over half of each party's change since the last election, and it was closer than the polls at nine of ten distances. The polling-day forecast moves today's corrected shares part of the way back to 2025 and widens the ranges for the time left.
Summary
Everything else in the Netherlands section answers the question "what if the election were held today?". The next regular Tweede Kamer election is on 15 May 2030, 1271 days after the most recent poll, though three of the last four elections came early. This paper describes how we get from today to polling day.
In Britain the older parties finish much closer to their previous result than their mid-term polls suggest, and we pull them part of the way back (From today to polling day: how far British polls move before a general election). We ran the same test on Dutch polls, at ten distances from one month to three years before the elections of 2012 to 2025. Unlike in Italy, the pull-back helps. The best-fitting curve keeps a little over half of each party's change since the last election and pulls the rest back, and with each election predicted from the others it was closer to the result than the polls as they stood at most distances. So the polling-day forecast moves today's corrected shares part of the way back towards 2025, and widens the ranges for the time left: polls this far out have missed each party by about 4.2 points by the time votes were cast.
What has changed since this paper was written
Updated 9 November 2026. The rest of the paper is as first published, except where a sentence below says it was corrected.
- Only elections from 2000 shape the forecast (28 September). Earlier elections stay in the long view and on the history pages but no longer feed any test, correction or error size. In the Netherlands the gemeente swing is now tested on the eight pairs of elections from 2002; proportional swing was closer at all eight, so it stays.
The evidence: polls at fixed distances before five elections
The evidence is every national poll in our series for the three years before each of the Tweede Kamer elections of 2012, 2017, 2021, 2023 and 2025, from Wikipedia's poll tables. Coverage before the tables start is missing: for 2012 the series begins in January 2012, for 2017 in March 2016. Where a poll gives only seats, its share is seats ÷ 150 (Polls that publish only seats: deriving shares from seat headlines).
For each election we average the polls at 30, 61, 91, 182, 273, 365, 548, 730, 913 and 1,095 days before polling day, over a 60-day window centred on each point, averaging within each pollster first and then across pollsters with equal weight. Each record holds the party, the poll average, the party's result at the previous election and its result at this one, with parties measured by their line to today (GroenLinks and the PvdA together as PRO). A party with no previous result is left out. That gives 546 horizon points. Because 2021, 2023 and 2025 were less than three years apart, some polls appear in the windows of two elections.
What the data show
| Days before polling day | Elections | Polls' miss | Previous result's miss | Pull-back's miss | RMSE |
|---|---|---|---|---|---|
| 30 | 5 | 1.85 | 2.80 | 1.88 | 2.85 |
| 61 | 5 | 2.39 | 2.80 | 2.35 | 3.62 |
| 91 | 5 | 2.44 | 2.80 | 2.39 | 3.67 |
| 182 | 5 | 2.46 | 2.82 | 2.42 | 4.05 |
| 273 | 5 | 2.48 | 2.82 | 2.38 | 3.66 |
| 365 | 4 | 2.43 | 2.97 | 2.40 | 3.70 |
| 548 | 3 | 2.85 | 2.90 | 2.72 | 4.40 |
| 730 | 3 | 3.16 | 2.83 | 2.88 | 4.74 |
| 913 | 3 | 3.30 | 2.64 | 2.78 | 5.01 |
| 1095 | 3 | 2.73 | 2.78 | 2.50 | 3.81 |
The previous result is a moderate guide. It missed each party by about 2.8 points, at every distance, since it does not change with the distance.
The polls are better close in and not much better far out. A month out they missed by 1.85 points per party, from three months to a year out by about 2.4 to 2.5, and two to three years out by about 3. At the longest distances the polls are no better than the previous result.
The pull-back beats both at most distances. A weighted average of the two was closer than the polls at nine of the ten distances, all but the closest, and the gain is largest far out.
Election by election:
| Days out | 2012 | 2017 | 2021 | 2023 | 2025 |
|---|---|---|---|---|---|
| 30 | 3.00 | 1.64 | 1.72 | 1.26 | 1.62 |
| 61 | 3.87 | 2.22 | 2.09 | 1.61 | 2.16 |
| 91 | 3.89 | 2.38 | 2.08 | 1.83 | 2.00 |
| 182 | 3.31 | 1.91 | 2.00 | 2.90 | 2.18 |
| 273 | 3.39 | 2.05 | 2.21 | 2.41 | 2.33 |
| 365 | – | 1.86 | 2.69 | 2.35 | 2.79 |
| 548 | – | – | 2.48 | 2.48 | 3.59 |
| 730 | – | – | 3.07 | 2.81 | 3.59 |
| 913 | – | – | 2.50 | 3.42 | 3.97 |
| 1095 | – | – | 2.40 | 3.03 | 2.76 |
Table: Polls' average miss per party at each distance and election, points.
2012 was the hardest: the polls a month out still missed by three points per party, after the SP's summer surge collapsed in the last weeks of the campaign. 2023 was the easiest a month out and among the hardest six months out.
The pull-back curve and its test
The polling-day share is a weighted average of today's share and the previous result,
with the floor \(f\) and time constant \(\tau\) chosen by least squares over all horizon points. The best fit is
With \(\tau\) this short, the weight falls to its floor within a few months: from about half a year out, a little over half of each party's change since 2025 is kept and the rest pulled back. The test, each election predicted from a curve fitted on the other four, beat the polls as they stood on average, so the curve is used.
Today, with more than three years to go, the pull-back moves each party close to half-way back to its 2025 share:
- PRO, D66 and the VVD, the CDA and the PVV move towards their 2025 shares of 12.8%, 16.9%, 14.2%, 11.8% and 16.7%;
- JA21 and FvD, which have risen since 2025, are pulled down towards 5.9% and 4.5%;
- the DNA, which did not stand in 2025, stays as polled, and the PVV is pulled towards the PVV's 2025 share even though seven of its MPs have since left.
The last point is a weakness: the PVV's 2025 vote included voters who may now back the DNA, so pulling the PVV back towards 16.7% probably overstates it.
Drift
The polls' miss grows with the time left, and the polling-day forecast keeps that uncertainty. The root-mean-square miss of the pulled-back shares at each distance is fitted by
over the ten distances. At 1271 days it is 4.2 points. Under 30 days it is scaled down in proportion to the days left, reaching zero on polling day, and it is never less than half a point. Beyond three years, the longest distance tested, the curve is an extrapolation.
In the simulation the drift is scaled by the square root of each party's share relative to 15%, so that large parties get more and small ones less, and combined with the party's final-poll spread:
Polling-day ranges
The polling-day forecast runs the simulation (Simulating Dutch polling error: where the chances come from) with 1,000 runs and equation (4), centred on the pulled-back shares:
| Party | Share today | Share, polling day | Seats today | 80% range today | 80% range, polling day | Largest, polling day |
|---|---|---|---|---|---|---|
| PRO | 19.0 | 16.3 | 30 | 25–34 | 16–34 | 43% |
| PVV | 13.9 | 15.2 | 21 | 16–27 | 13–33 | 33% |
| JA21 | 11.8 | 9.2 | 18 | 16–20 | 8–20 | 1% |
| D66 | 10.4 | 13.4 | 16 | 13–18 | 12–28 | 14% |
| VVD | 9.2 | 11.5 | 14 | 10–18 | 9–25 | 6% |
| CDA | 8.7 | 10.1 | 13 | 9–18 | 8–23 | 4% |
| FvD | 7.6 | 6.2 | 11 | 10–13 | 3–15 | under 1% |
| 50PLUS | 3.3 | 2.5 | 5 | 4–6 | 0–7 | 0% |
| PvdD | 3.0 | 2.6 | 4 | 3–5 | 0–7 | 0% |
| SP | 2.8 | 2.4 | 4 | 1–7 | 0–8 | 0% |
| SGP | 2.8 | 2.6 | 4 | 3–5 | 0–7 | 0% |
| Volt | 2.3 | 1.8 | 3 | 2–4 | 0–5 | 0% |
| CU | 2.1 | 2.0 | 3 | 1–4 | 0–6 | 0% |
| DENK | 2.0 | 2.2 | 3 | 2–3 | 0–6 | 0% |
| BBB | 1.2 | 1.9 | 1 | 0–2 | 0–6 | 0% |
At more than three years out, the drift is larger than any party's final-poll spread. The ranges of the large parties become two or three times wider, the small parties' chances of a seat fall, and the chance of being largest spreads among four or five parties. The chances of 76:
| Combination | Today | Polling day |
|---|---|---|
| The Jetten cabinet (D66, VVD, CDA) | 0% | 1% |
| D66, VVD, CDA and JA21 | under 1% | 17% |
| D66, VVD, CDA and PRO | 25% | 61% |
| PRO, D66 and CDA | 0% | 5% |
| The right: VVD, PVV, JA21, FvD, DNA, BBB, SGP | 11% | 31% |
| The left: PRO, SP, PvdD, Volt, DENK | 0% | 0% |
The date
The Kiesraad gives 15 May 2030 as the next regular election; the site uses it. That is the cautious choice, giving the widest ranges. The Kamer can be dissolved earlier, and three of the last four elections were early: 2012, 2023 and 2025. The Jetten cabinet is a minority cabinet with 66 seats, and a motion of no confidence was tabled against it in September 2026. An earlier election narrows the drift: at a year out it would be about 3.8 points, at six months about 3.6, at three months about 3.4. A Dutch snap campaign has taken about four to five months from the fall of a cabinet to polling day.
Limitations
Five elections. The curve and the test rest on five elections, and on three at the longest distances.
Wikipedia's tables. The series is taken from Wikipedia's poll tables, with seats-derived shares for most polls, and is not checked against releases.
The drift includes the final-poll error, which the simulation also adds, so the ranges err on the wide side, as elsewhere.
Split and merged parties. The pull-back moves PRO towards GroenLinks-PvdA's 2025 share and the PVV towards the undivided PVV's.
New parties. Dutch politics has produced a new national force at most recent elections, from the PVV in 2006 to the BBB and NSC in 2023. A party that forms after today's polls cannot be anticipated.
References
Erikson, R. S. and Wlezien, C. (2012). The Timeline of Presidential Elections: How Campaigns Do (and Do Not) Matter. Chicago: University of Chicago Press.
Stone, M. (1974). Cross-validatory choice and assessment of statistical predictions. Journal of the Royal Statistical Society: Series B, 36(2), 111–147.
How to cite
Polling Forecast (2026). From today to polling day in the Netherlands. Retrieved 21 November 2026 from https://www.pollingforecast.com/research/netherlands-election-day. Live figures change as new polls arrive; the date above is when the text was last revised.


