Today we are launching our Westminster projection. As with our other systems this is customisable and covers all 650 Westminster constituencies. It sits alongside our existing projections for Holyrood, the Senedd, Stormont and the London Assembly making this (I believe) the only site with a complete editable projection for every major chamber in the UK and the only Westminster projection to fully include Northern Ireland.
Our hexmap based on today’s polling average (9/10/26)
How it works
We take each constituency’s actual 2024 result and apply a swing based on the difference between that and your input. If you double a party’s vote share from last time our system will roughly double their support everywhere, but with some diminishing returns and a series of controls to keep things in check. We have put in some constituency specific factors to account for popular incumbents and other local issues but this is tricky.
You can either type in your own numbers or click one of our presets, which include the 2024 results, a live polling average (Nowcast) and a Margin of Error button which randomises results around the polling average.
Monte Carlo Simulation
We also have a Monte Carlo simulator which runs one thousand random simulations around the polling average to show the level of uncertainty both overall and for every seat. We randomise national vote shares, tactical voting and local swings to see what could happen. The result is a probability score for every seat which you can view per party and by constituency. Every seat gets sorted into a category based on that probability so you can see if your seat is a Safely Conservative, a Labour/Reform coinflip or a Long Shot for the Lib Dems.
The results of 1000 simulations around today’s Nowcast (9/10/26)
We have the same set up for the devolved chambers in Scotland, Wales, NI and London.
Tactical voting
In a close seat some people switch their vote from their favourite party to whichever of the front-runners they see as being better (or less bad) rather than “wasting” their vote.Our model does this computationally by squeezing some of the support from smaller parties towards the most competitive parties. We account for the ideology of the parties, for example: more Green supporters will vote for Labour against Reform than for the Conservatives against Reform.
The effect scales with how close the race actually is; a safe seat isn’t squeezed at all but one which is quite close will have a lot of tactical shifts. A big part of my PhD involved understanding how and why people vote tactically so I’ve poured this experience into it but it will not be perfect.
Local variation
A single national swing figure also can’t capture the fact that some places are just more receptive to a given party’s message than others, independent of how they voted last time. We apply regional swings across England and seprate swings in Scotland, Wales and Northern Ireland. We also vary swings by urban/rural character and by clusters of nearby seats so that nearby seats or similar seats tend to move in similar directions.Missing candidates and local pacts
Not every party stood in every seat in 2024 which makes factoring them in hard. For example, there were several seats that Reform didn’t contest last time but where they would have won a solid number of votes. We have created baselines for them by comparing those seats to other similar seats, both nearby and nationally.Northern Ireland is a bit different. We do not expect the DUP to stand in Fermanagh & South Tyrone because unionist parties tend to unite behind the UUP candidate so we keep them on 0% there. We keep the competing candidates as they were in 2024, but in the Margin of Error and Monte Carlo system we randomly add-in some more unionist and nationalist pacts in places like Lagan Valley. Once we know who is standing we will turn this off.
Coalition builder
Once you’ve got a seat projection, our Coalition Builder lets you drag parties together to see whether they’d cross the line for a majority. We have an automated guide which says if this particular coalition is workable in terms of both ideology and size, this is our opinion so feel free to disagree. Sinn Féin’s MPs don’t take their seats at Westminster by convention, so they, along with the Speaker who doesn’t vote, are greyed out.
The easiest way to reach the majority line from today’s Nowcast (9/10/2026)
The usual caveat
This is a projection, not a prediction. We’ve tried to account some real-world, but correctly estimating tactical behaviour and local factors across 650 seats lots of competing parties is impossible to do perfectly. We will keep amending the model to improve the methodology as more data becomes available and as events take place which could add even more crucual local factors. The model is based on historic trends and a series of mathematical formulas, trends may not continue or they may strengthen.As with all our projections. Please don’t bet on any of this.