First off, thanks for reading and subscribing at the link below!
https://lnkd.in/gfY7cUsT
I've taken a stab at an election forecasting model, similar to what you'd find over at 538 Politics or at the Nate Silver Bulletin. Forecasting elections relies on a lot of assumptions and careful weighing of data. This is also not something that I'm paid to do and rather something I did as a side project for fun. That's a major disclaimer - this can and will change as more data comes in and filters out!
First, this is a tight election. The model (after 10k simulations) gives Harris approximately a 75% chance of winning if the election were held today.
But, she's really just one Pennsylvania away from losing, with her mean & median EC votes hanging at 289. That & Pennsylvania is close. Very close.
The model includes probabilities for each and every state.
I walk through the methodology in more detail in the article. But, I make various weighting decisions like:
LV > RV
More Sample Size -> More Weight
Pollster Rating & Quality -> Higher 538 rating, greater weight
I also apply a small convention bounce adjustment for polls within 3 weeks of their national conventions. We're not seeing huge bounces (so it would seem). This is equivalent to around 0.5% right now. But, that can make a difference, like in the very close state of North Carolina. You can see how tight that election could be in the piece.
There's some other weighting discussion in the post, please read there!
Regardless, know this will be a tight election. You can see all of the state and swing-state results in the piece - but they also may change substantially as new polling comes in.
This model does not include some other assumptions or economic indicators in the way the professional folks do, but I think that it can get us pretty close and again demonstrate how tight this election will be.
In the meantime, I think that the map below seems pretty likely to be close.
How close? Time will tell, and this model will be updated everyday!
