Election models megathread

[Figueira]

From Fivethirtyeight: We Fixed An Error In Our Deluxe Midterm Forecast

Basically they deleted a bunch of Sabato ratings accidentally in June, so a bunch of races (mostly House) showed incorrect odds until they were corrected. Overall the correction increases Democratic odds in the House slightly, but notably causes some swing district odds to swing massively (a full list exists in the article). Meanwhile in the Senate Democratic odds have gone down slightly due to a slight change in Wisconsin.

[Figueira]

From Fivethirtyeight: We Fixed An Error In Our Deluxe Midterm Forecast

Basically they deleted a bunch of Sabato ratings accidentally in June, so a bunch of races (mostly House) showed incorrect odds until they were corrected. Overall the correction increases Democratic odds in the House slightly, but notably causes some swing district odds to swing massively (a full list exists in the article). Meanwhile in the Senate Democratic odds have gone down slightly due to a slight change in Wisconsin.

These changes appear to have been reversed, and now a completely different group of races have shifted.

[Figueira]

Not that it means much, but if you take the 538 Deluxe model and assign each of the tossups to the party who has the slight lead, Republicans win the House by one seat. Of course, their model has some weirdness (Tilt D TX-15 and AZ-02, Likely R OH-01 and OH-13) but this is just something to note as of the moment.

I noticed this as well. If you do this and give 50/50s to Democrats, then Democrats actually win. Of course, I think they have Kiggans slightly favored in VA-2, and it just rounds to 50, but that wasn't true a few minutes ago.

[Figueira]

I just noticed that while Fivethirtyeight has Republicans winning an average of 230 seats, if you go by the individual seat count, it actually has Republicans winning 223 seats, which is not that far off from what the actual result will be.