Electoral Calculus
Electoral Calculus is a British political consultancy and polling analysis service best known for its election forecasting website, which attempts to predict the outcome of United Kingdom general elections seat by seat. Based in London and listed in this directory under Oxford, the service was founded and is run by Martin Baxter, a former financial analyst who specialises in mathematical modelling. Its predictions, published freely online and updated as new opinion polls appear, are routinely cited by newspapers, broadcasters and political commentators whenever the electoral map of Britain is being discussed. Electoral Calculus Ltd was incorporated as a private limited company in 2016, although Baxter had been running the forecasting site itself since the mid-1990s, long before election modelling became a staple of British political journalism.
Origins and Martin Baxter
The service grew out of Baxter's background in the City of London, where he worked as a financial analyst building mathematical models. He applied the same quantitative instincts to elections, developing a website that used scientific techniques on data about Britain's electoral geography to estimate how national voting trends would translate into seats in the House of Commons. The site built a following gradually among political enthusiasts, journalists and campaigners. In 2004 The Guardian listed it among its hundred most useful websites, describing it as the best for election predictions, and in 2012 an Oxford doctoral researcher described it as probably the leading vote and seat predictor on the internet. That early recognition established Electoral Calculus as a reference point well before MRP modelling entered the mainstream of British polling.
What the service publishes
The Electoral Calculus website brings together election data, predictions and analysis in one place. Its core product is a seat-by-seat prediction for the next general election, showing estimated vote shares and the likely winner in each constituency, alongside national totals for seats and votes. Around that core sit maps, charts and written analysis pieces explaining movements in the polls and their implications. The site carries separate sections for elections in Scotland and in Northern Ireland, reflecting the distinct party systems in those parts of the United Kingdom. It also maintains historical election data, allowing users to compare current standings with past results, and offers tools that let visitors feed in their own assumed vote shares and see how the Commons arithmetic would change.
Methodology: from swing models to MRP
For most of its history, Electoral Calculus based its forecasts on modified versions of uniform national swing. Its early approach, known as the Transition Model, was derived from the additive uniform swing model but applied national swings proportionally to predict local effects. In October 2007 it introduced the Strong Transition Model, which took account of the behaviour of strong and weak party supporters when estimating how votes move between parties. The models drew on national polls and trends while deliberately excluding purely local factors. Since 2019, however, the service has used MRP — multilevel regression and post-stratification — the technique that has become the industry standard for constituency-level forecasting. Its MRP model combines demographic information, past voting behaviour and geographic data to estimate vote shares in each seat individually, rather than assuming a single national swing falls evenly everywhere. The workings of the models are explained in detail on the website, a transparency that has contributed to the service's credibility with specialist audiences.
Role in election coverage
Electoral Calculus occupies a distinctive place in British political media because it translates polling — which measures national vote share — into the currency that actually decides elections under first-past-the-post: seats. Journalists use its regularly updated predictions to test storylines about possible majorities, hung parliaments and the fate of individual parties, and its figures have been cited by commentators including Andrew Rawnsley and Michael White in The Guardian and John Rentoul in The Independent. Baxter himself appears as an expert guest in broadcast and podcast coverage of elections, explaining, for example, how voters who backed the Conservatives in 2019 dispersed to other parties or to abstention in later contests. Each month during election periods, his updated projection based on polling averages has been a fixture of specialist sites such as PoliticalBetting.com.
Elections and track record
The service has published predictions across successive electoral cycles, including the general elections of 2010, 2015, 2017, 2019 and 2024, as well as devolved and local contests. Its record, like that of every forecaster, is mixed in detail: after the 2005 election, for instance, Baxter acknowledged that while the model had predicted the winner correctly, it had overestimated the governing party's majority, attributing the gap to opinion poll error, regional variation in swing and local factors the model set aside. That pattern — dependence on the quality of the underlying polls — remains the central caveat of all seat forecasting, and Electoral Calculus has generally been candid about it. Its longevity across three decades of elections, through very different political eras, is itself a notable part of its record.
Polls, averages and the wider forecasting field
Electoral Calculus does not mainly conduct its own fieldwork; it sits downstream of the polling industry, taking published voting intention polls, averaging and interpreting them, and converting them into seat estimates. Baxter has at times criticised pollsters for a lack of transparency about methodology, pointing to the British Polling Council's disclosure rules, which require members to publish detailed data tables shortly after a poll's release. The service's monthly projections have historically been built from such polling averages. In comparative assessments of forecasting methods, its Strong Transition approach is typically classed among the more established alternatives to the naive uniform swing baseline, alongside academic and broadcaster models such as the BBC's projected national share and university-based regression forecasts.
Scottish and Northern Irish coverage
Because British elections are really several different contests running in parallel, Electoral Calculus maintains dedicated coverage of Scotland, where the SNP, Labour, the Conservatives and the Liberal Democrats compete on a different axis from England and Wales, and of Northern Ireland, where the party system is entirely distinct. This matters for the overall Commons calculation: a forecasting model that treated Great Britain as a single uniform market would misread both nations badly. The service's seat-level approach allows Scottish and Northern Irish dynamics — such as the collapse or recovery of a party in one nation — to feed into the UK-wide prediction without distorting it.
How readers and campaigns use it
Beyond journalists, the site's audience includes political activists, candidates, academics, students of psephology and election bettors, all of whom use its constituency-level detail to identify marginal seats and plausible outcomes. The ability to model scenarios — what happens if a party's vote share rises or falls by a few points, or if tactical voting patterns shift — makes it a planning tool as well as a news source. Teachers and lecturers have long used it to demonstrate how first-past-the-post can deliver large majorities on modest vote leads, or deny a party seats despite a substantial national vote, because the model's outputs make the system's disproportionality visible seat by seat.
Limitations and transparency
Electoral Calculus is explicit that its predictions are only as good as their inputs. Polling error, differential turnout, boundary changes, by-election shocks and genuinely local campaigns can all move individual seats away from a model's expectation, and a model built on national data will always struggle with seats decided by local personalities or controversies. Its published methodology pages — setting out the transition models and, later, the MRP approach in full — are part of its offer: readers are invited to understand the machinery rather than treat the headline seat totals as certainties. That openness distinguishes it from commentary-led forecasting and aligns it with the academic psephology tradition.
Place in the UK media landscape
Within this directory's agencies category, Electoral Calculus represents the analytical end of political information services: not a newswire and not a conventional pollster, but an interpreter whose output shapes how news organisations report the state of the parties. Its combination of a free public website, a long publishing history stretching back to the 1990s, a named and accountable founder-analyst, and a willingness to document its methods has made it one of the most durable independent voices in British election forecasting. As MRP modelling has spread across larger rivals with bigger budgets, Electoral Calculus has adopted the same techniques while retaining the lean, specialist character of its origins.
At a glance
Electoral Calculus is a London-based political consultancy and forecasting service founded by former financial analyst Martin Baxter, with the website dating to the mid-1990s and the limited company incorporated in 2016. It publishes seat-by-seat general election predictions, polling analysis, maps and historical data for the whole United Kingdom, with dedicated Scottish and Northern Irish sections. Its methods evolved from proportional swing models — the Transition and Strong Transition models — to MRP modelling from 2019. Its official website is electoralcalculus.co.uk.
