Thursday, October 15, 2026
B.C. votes in 9 days

U.K. by-election race desk · Holborn and St Pancras

Labour has to defend Starmer’s seat without Starmer.

Keir Starmer resigned from the Commons on 1 September 2026, ten weeks after leaving Downing Street. The writ has not been moved, no date is set, and nobody has declared. This page is what a forecast honestly looks like at that stage.

Add Vote-Scope to your preferred sources on Google

Where the race stands

WritNot moved
Polling dayNot set
Constituency pollsNone
Vacant since1 September 2026

Why there is no single number here

Vote-Scope does not publish a point estimate for a by-election with no constituency poll — the rule was set at Clacton and it holds here. What the model can do without one is bound the plausible range from the 2024 result and the live national posterior, and then say which unknown actually moves the outcome. In this race that unknown is not the swing. It is who stands.

Projected shares

74.4% Labour holds the seat unconditional, across 50,000 simulations

10th–90th percentile, median marked

Labour
36.6% [29–44]
Green
29.6% [23–37]
Reform UK
11.5% [9–15]
Conservative
6.4% [5–8]
Workers Party (George Galloway)
6.2% [3–11]
Liberal Democrats
4.6% [4–6]
Other candidates
4.1% [3–6]

Path to victory

Each row conditions on one field event, using the same draws as the headline number — so the probabilities are directly comparable to it and to each other.

If…Share of drawsp(Labour hold)p(Green win)
Top decile of the left-revolt factor: simultaneous Labour demobilisation and Green mobilisation. 10% 41.9% 58.1%

The national anchor

The seat baseline is swung by the live U.K. posterior, weighted toward proportional rather than uniform swing: a bastion at 48.9% does not shed the national points the way an average seat does.

Party2024 nationalModel nowSwing
Labour 33.7% 26.5% −7.2
Conservative 23.7% 22.1% −1.6
Liberal Democrats 12.2% 10.3% −1.9
Reform UK 14.3% 22.0% +7.7
Green 6.8% 9.9% +3.1
Other candidates 3.8% 5.3% +1.5

The field

Nobody has formally declared. The names below are reported intentions, tracked so that the model follows the field rather than a guess about its size.

  • Sagal Abdi-Wali Labour reported intention
  • Zack Polanski Green Party reported intention
  • George Galloway Workers Party reported intention
  • Peter Newman Reform UK reported intention
  • Ewan Cameron Conservative reported intention
  • Patrick Stillman Liberal Democrats reported intention
  • Barmy Brunch Monster Raving Loony reported intention
  • Tom Darwood No description reported intention
  • Clare Fischer Climate reported intention
  • Tyler Johnson Restore reported intention
  • Briony Kapoor Rejoin EU reported intention
  • Ketankumar Pipaliya UK Voice reported intention
  • Jonathan Silberman Communist League reported intention
  • Helen Spiby-Vann CPA reported intention
  • Ben Walker UKIP reported intention

What happened here in 2024

  1. Keir Starmer Labour 18,884 48.9%
  2. Andrew Feinstein Independent 7,312 18.9%
  3. David Stansell Green 4,030 10.4%
  4. Mehreen Malik Conservative 2,776 7.2%
  5. David Roberts Reform UK 2,371 6.1%
  6. Charlie Clinton Liberal Democrat 2,236 5.8%
  7. Six other candidates 993 2.6%

July 4, 2024 general election result — 38,602 votes cast, 54.1% turnout.

Read this before quoting a number

  • No constituency poll, no polling date and no candidate list. Three first-order unknowns; the intervals shown are the model’s, not the world’s.
  • The probability that Feinstein or Polanski stands is a documented judgement, not a measurement. Those two numbers drive most of the spread, and any news about the field should update them before anything else in the model is touched.
  • Starmer’s personal vote is stripped with a prior centred on zero and an undetermined sign. The 2024 boundaries are new, so the repo has no earlier line for this seat and no London comparator — the level is not identified, and the model refuses to invent one.
  • There is no new-prime-minister honeymoon term. Burnham has been prime minister since July and the national posterior already measures him; adding a bounce would count it twice.

Methodology after the signal

A by-election model that says what it does not know.

The engine simulates flows from the 2024 voter blocs, swings them by the live national posterior, and treats the two candidacy questions as explicit events rather than as variance folded into a wider interval. That is what lets the page answer “what would change this race?” instead of only “what is the number?”

  • Projected shares — updated October 15, 2026 · 50,000 simulations
  • The national anchor — web_data/uk/latest.json