In 2016 and 2020, the news organization FiveThirtyEight put out results from a statistical model predicting the results of the elections in those years. The 2020 model https://projects.fivethirtyeight.com/2020-election-forecast/ came out in Early June showing Biden with a 70% chance of winning, and was frozen the day of the election, showing Biden with a 89% chance of winning - a difference of 19 percentage points. This question asks if a similar model put out in 2024 will show a difference of more than 10 points between the earliest time listed on the homepage for the model, and the model result on the day of the election (or otherwise, the last day the model was active before being frozen). If FiveThirtyEight doesn't put out a model, this resolves N/A.
As I read the criteria, this would resolve on the Harris vs Trump probabilities yes, or any future model if yet another candidate leaves the race. I think it's fair to call this current model separate from the one before, since they show different start dates. So the Trump win probability of 53% currently shown for July 26 is what the final Trump number will be compared to.
I am also noticing the title says "at least 10" and the criteria say "more than 10". This is ambiguous, so I am going to change the title to match the criteria. Admittedly it seems Trump changed from 41 to 51 over the course of the model as it was frozen when Biden dropped out. But I think that this shouldn't resolve Yes yet because the criteria dominate.
Silver has left 538 leaving me in a little bit of doubt. I'm not certain, but I assume that one way or another, 538 will get some kind of model up on their site. If so that's what I'll use.
If not, there's a chance that Silver's model could still appear somewhere. Would traders on this market like me to use that model if 538 doesn't put one out?