What do People Think Will Happen in 2026?
The combined wisdom of a 1108-strong crowd
A massive thank you to everyone who entered, shared or otherwise supported the 2026 Forecasting Contest.
This year it received a massive 1108 entries1 - almost three times as many as last year and significantly above my best hopes. To give context, it’s more than a third as many people as entered Scott Alexander's ACX/Metaculus forecasting contest, which given Scott is far more famous than I am, with about a million times as many subscribers, makes me very pleased.2
I’m also grateful to the i Paper for writing it up again this year.
In this post I’ll be presenting the full Wisdom of Crowds, my own forecasts and a couple of other things of interest.
The Wisdom of Crowds
The 1108 entries included 378 people who work in journalism, politics, public policy or similar - but the strength of the wisdom of crowds is that it doesn’t rely on the expertise of any single individual, but on aggregating the answers from all.
Last year, the wisdom of crowds would have come 28th, beating 93% of contestants, including the vast majority of those who work in related fields. This year, with three times as many people, it should be even more accurate!
So what did it say?
And here are my own forecasts (presented alongside the wisdom of crowds for comparison).
Regarding the leadership of Labour and the Conservatives, at the beginning of the year, I thought that one would last out the year but wasn’t sure which - hence putting 60 for each. Six weeks on, the forecast of last year’s winner, at 35% for Starmer and 65% for Kemi, is looking a lot more prescient!
In terms of other significant divergences from the wisdom of crowds, I am more confident Reform will get the most seats in the locals (70%), more confident inflation will fall (35% it will be over 2.5%), more confident that the cordon sanitaire will hold in Germany (25% the AfD enters a state government) and more optimistic about a cease-fire in Ukraine (60% - would be nice if that one came true). Time will tell!
To look at two other forecasts of possible interest, reader Chris again used an AI - this time Opus 4.5 - to generate predictions.3 Last year the AI got a Brier score of 0.222 which would have put it 136th (35th percentile) - so better than the median human but well below the top performers. I must admit I think its forecasts this year look highly dubious, but its reasoning can be found here.
In addition, I’m interested to see what the best forecasters think will happen this year. Of the 27 people who beat the wisdom of crowds last year, I was able to identify ten who entered this year’s contest and had used the same name/pseudonym. It is a pretty eclectic mix,4 including, in no particular order:
Last year’s winner, a TfL station area manager
Me, a think-tanker and former civil servant and special adviser.
An economics professor who was formerly a Chief Economist in the civil service.
A public health statistician.
A KC who specialises in shipping law.
A supply chain modeller at a major supermarket.
Four other people who I don’t know.5
The average of these ‘experts’ is not that far off the wisdom of crowds in most areas - but it is different, and it will be very interesting to see whether taking only a small number of those who did well at forecasting in a previous year can outweigh the benefits of the much greater numbers in the full wisdom of crowds.
Thank you to everyone who took part - and remember to check back in next January to see the results! (This is the bit where I remind you to subscribe, if you’ve not already).
And as I figured out the answer validation settings on Google Forms this year, they’re all valid.
Last year when I said it was the largest UK-focused forecasting contest, I was secretly hoping no-one would contradict me (no-one did!); this year I’m confident of it.
Potentially other entrants used AI to assist them but this can be thought of as a pure AI guess, rather than an AI-assisted human.
Though with some strong commonalities.
These being Principal P, JoeS, Mike Smith and Nick O’Connor - if any of you are reading this, do say ‘hi’ in the comments!
Nick O'Connor here - social worker. I'd love to believe that I'm great at forecasting, but I suspect I did well because a) I got lucky, and b) I took the Brier scoring seriously - where I didn't know I put 50%, where I had a view but not much evidence to back it up I bumped that up to 60%ish.
Ironically, I may have done better as a result of following the news much less closely than I used to - I knew how ignorant I was compared to my past self, which might have given me a more accurate estimate of how ignorant I was in general.
I wonder what effect Restore Britain will have on questions 6-15: after about a week they're already bigger than the Lib Dems in terms of both membership and bookies' odds of winning the next GE (aggregated on oddschecker). I don't know whether they'll plateau here or continue this trajectory; but either way it's another surprise that probably nobody's predictions accounted for, in an already unusual political landscape.