Our locked pre-election forecast, scored against the actual count.
Reform UK won 15 council majorities and 1,439 seats.
We called the winner correctly in 55.2%
of the 2,903 wards we have results for.
Headline
What happened on the night
15 Reform UK overall majorities 1,439 Reform UK seats won (across all contesting councils) 10 councils where Reform UK is now largest party without a majority 78 councils with no overall control
All seats won on May 7
Reform UK
1,439
Labour
996
Conservative
769
Lib Dem
747
Green Party
566
Independent / Other
227
Forecast vs reality
How accurate were we?
Two forecasts were live for May 7. The pre-registration (sha256-witnessed 11 days before polling day) and
the as-served forecast, which added one extra step (a Reform realignment uplift) calibrated on the morning
of polling day after we spotted a structural undershoot in northern unitaries.
Forecast version
Winner accuracy
Major-party MAE
Ballots evaluated
Live (as users saw it)
55.2%
7.38pp
2,903
Pre-registration (locked 26 Apr 2026)
49.6%
10.67pp
2,903
Per-party vote-share error (live)
Garforth and Swillington Independents Party
62.45pp MAE -62.45pp bias
Merton Park Ward Independent Residents
59.01pp MAE -59.01pp bias
Ashtead Independents
50.17pp MAE -50.17pp bias
Nork & Tattenhams Residents' Associations
48.97pp MAE -48.97pp bias
Great Yarmouth First
46.04pp MAE -46.04pp bias
Upminster and Cranham Residents Association
45.45pp MAE +45.45pp bias
Labour Co-operative
44.83pp MAE -44.83pp bias
Local Alliance
42.35pp MAE +42.35pp bias
Horwich & Blackrod First Independents
41.77pp MAE -41.77pp bias
Basingstoke & Deane Independents
38.10pp MAE -38.10pp bias
Lydiate and Maghull Community Independents
36.44pp MAE -36.44pp bias
Hornchurch Residents Association
35.23pp MAE +35.23pp bias
Our West Lancashire
32.60pp MAE -32.60pp bias
Camden People's Alliance
32.15pp MAE -32.15pp bias
Workers Party - Oldham Group
32.11pp MAE -32.11pp bias
Hinchley Wood Residents
31.69pp MAE -31.69pp bias
Morley Borough Independents
31.11pp MAE -31.11pp bias
Whitehill & Bordon Community Party
30.37pp MAE -30.37pp bias
The People's Independent Party
30.04pp MAE -30.04pp bias
Residents for Guildford and Villages
30.00pp MAE -30.00pp bias
Farnham Residents
29.35pp MAE -29.35pp bias
Local
27.95pp MAE -27.95pp bias
Your Bradford Independent group
26.41pp MAE -26.41pp bias
Runnymede Independent Residents' Group
25.89pp MAE -25.89pp bias
Rother Association of Independent Councillors
25.21pp MAE -25.21pp bias
Harold Wood Hill Park Residents Association
25.15pp MAE +25.15pp bias
Redbridge and Ilford Independents
25.10pp MAE -25.10pp bias
The Oldham Group Party
23.97pp MAE -23.97pp bias
Chislehurst Matters
23.13pp MAE -23.13pp bias
Residents Associations of Epsom and Ewell
22.78pp MAE -22.78pp bias
The All In Party
22.77pp MAE -22.77pp bias
Walsall Community Independents Party
21.11pp MAE -21.11pp bias
The Conservative and Unionist Party
18.99pp MAE -18.99pp bias
Newcastle Independents
18.45pp MAE -18.45pp bias
Canvey Island Independent Party
18.39pp MAE -18.39pp bias
North Tyneside Community Independents
16.48pp MAE -16.48pp bias
Haringey Socialist Alliance
16.14pp MAE -16.14pp bias
Thames Ditton / Weston Green Residents' Association
The Peace Party - Non-violence, Justice, Environment
1.89pp MAE +1.02pp bias
Confelicity
1.83pp MAE -0.61pp bias
Communist Party of Britain
1.73pp MAE +1.21pp bias
Social Democratic Party
1.71pp MAE -1.71pp bias
The Official Monster Raving Loony Party
1.68pp MAE +0.62pp bias
Wandle Independent Group
1.64pp MAE -1.64pp bias
People's Alliance for Change and Equality
1.61pp MAE -1.61pp bias
SDP
1.58pp MAE +1.58pp bias
Heritage Party
1.56pp MAE +1.13pp bias
Vectis Party
1.41pp MAE +1.41pp bias
Taking The Initiative Party
1.37pp MAE -0.62pp bias
Animal Welfare Party
1.37pp MAE -1.37pp bias
Advance UK
1.31pp MAE -1.31pp bias
Christian Peoples Alliance
1.28pp MAE +0.28pp bias
Rejoin EU
1.23pp MAE +0.03pp bias
Build
0.89pp MAE -0.89pp bias
Farnworth and Kearsley First Party
0.81pp MAE +0.78pp bias
Pirate Party UK
0.81pp MAE -0.81pp bias
Working for Sidcup
0.78pp MAE -0.78pp bias
UK Independence Party (UKIP)
0.78pp MAE -0.33pp bias
Chesterfield And North Derbyshire Independents (CANDI)
0.77pp MAE -0.77pp bias
UK Voice
0.75pp MAE -0.75pp bias
UKIP
0.72pp MAE -0.72pp bias
TUSC
0.70pp MAE -0.70pp bias
Libertarian Party
0.60pp MAE -0.56pp bias
The Moon And Serpent Party
0.47pp MAE -0.47pp bias
The Revolting Party
0.47pp MAE +0.47pp bias
Wessex Regionalists
0.43pp MAE -0.43pp bias
Forward Party
0.42pp MAE +0.42pp bias
National Rebirth Party
0.41pp MAE -0.41pp bias
The Revolting Party. Join The Revolt
0.35pp MAE -0.35pp bias
Equality Party
0.31pp MAE -0.31pp bias
Imagine
0.29pp MAE -0.29pp bias
The Motoring Party
0.27pp MAE -0.27pp bias
The Christian Party Christian Peoples Alliance
0.23pp MAE -0.23pp bias
The Socialist Party of Great Britain
0.22pp MAE -0.07pp bias
Save Us Now
0.22pp MAE -0.22pp bias
Yeshua
0.20pp MAE +0.20pp bias
Climate Party
0.18pp MAE +0.18pp bias
Communist League Election Campaign
0.17pp MAE -0.02pp bias
Communities United Party
0.16pp MAE +0.16pp bias
The Forward Party
0.06pp MAE -0.06pp bias
Patria
0.06pp MAE +0.06pp bias
Bar width = MAE relative to the worst per-party error. A positive signed bias (red) means we
predicted that party would win more share than they did; negative (blue) means we
ran them cold. Reform was +3.35pp (slightly hot); Greens were −4.04pp (cold).
Methodology audit
Step 9b: the Reform realignment uplift
On polling day, after spotting Blackburn-with-Darwen at 4.8% Reform across 17 wards while Burnley/Hyndburn/Pendle
next door were at 30-40%, we added a per-ward "realignment uplift" calibrated on Asian-population share against
observed Reform share in Lancashire. It was applied as an upward floor only, so it never reduced a predicted Reform
share. The shadow column shows what would have happened without it.
Metric
Live (with 9b)
Shadow (no 9b)
Δ
Reform UK MAE (uplift cohort)
9.00pp
16.73pp
7.73pp
Wards helped / hurt / neutral
1061 helped · 467 hurt · 45 neutral
Verdict: net positive nationally. The London (0.50 multiplier) tier hurt Reform MAE by 3.43pp and was retuned
to 0.00 on 20 May 2026 (see methodology Step 9b). Full audit JSON: postaudit-2026-05-10.json.