Do not pay attention to 538 Liveblog %

[Johnny2X2X]

The inputs are swinging wildly, this model will drive you crazy tonight.

Just key on the handful of key House races we've known about for weeks.

[dhill926]

1. yep....

this place is an insane asylum tonight....

[JonLP24]

3. Im going crazy

I though Gillum was going to win but Desantis up with 91% in

[llmart]

2. I agree!

I won't click on any threads that have 538 in the subject line.

[Kaleva]

4. Yes.

[still_one]

5. The polls haven't even closed on the West Coast, and us in California have a lot of districts

[AdamGG]

6. That switched from a 39 seat gain to a 20 seat gain awfully fast

They have it back at 24 now, but that corresponds to what I just heard James Carville saying on MSNBC. It's apparently going to be close if we're going to take the house. Though, I wonder if the Democrats do well out west, if the numbers can pick up again. 538 projected 6 Democratic pick ups in Cali.

[Hoyt]

7. This is what an apparent 538 staffer is saying

You may have noticed that our real-time forecast has moved toward Republicans in the House. It’s being too aggressive, in my opinion. The model sees that a bunch of “likely Republican” districts (particularly in Florida) are now 100 percent likely to go red. But there hasn’t been the chance for Democrats to clinch many equivalent likely Democratic districts.

[treestar]

8. I agree some posters seem to

Think it decides and not the real polls which are actually being counted. It’s insane.

[Takket]

9. yes, it is bad data. they are using early returns, 1% vote counts.

surprised they would present such poor data.

[cally]

10. Important explanation

Link to tweet

?s=20

Sorry I don’t know how to copy tweet but it basically says it is weighted wrong. FiveThirtyEight is trying to correct