538 model & poll tracker thread

[ProgressiveModerate]

While this model is better than the crappy RCP model, I'm still underwhelmed. Release the F***ING FORECAST!

[ProgressiveModerate]

When your best number is being behind by 9 points nationally, that's not a good sign

[ProgressiveModerate]

538 is being super lazy this year. There's no excuse for the model not being out yet when the primaries have been done for months and even the 2016 version came out sooner.

I'm guessing they're either waiting for the conventions of 100 days before the election. Making the model itself really isnt that hard. I'm in 9th grade, and by myself I was able to create a full presidential and senate forecasts model in under a week that weights different factors like polling, incumbency, fundraising, and so much more. Having a bunch people who are actually getting paid working to create a model shouldn't take very long at face value. I think the COVID stuff they were working on caused them to start the model late, and now they just want to find a good date to release that isn't completely random. The model is probably already done if I'm being honest. Also, as someone else mentioned, I think that Nate thinks that Trump's chances will be perceived as too low by the general public. There are other little kinks that can be annoying to deal with; getting Doug Jones below a 10% chance using only statistics, and getting Bullock above a 35% chance of winning in MT at the same time is quite difficult. He's probably also being picky about those problems, and trying to adjust the model in a fair way to fix those sorts of issues.

[ProgressiveModerate]

https://projects.fivethirtyeight.com/2020-election-forecast/

This link sends you to a blank page that is not an error page. Literally any other simillar link send you to an error page. It's the same address they used for the 2016 forecast, just swap 2016 with 2020. Keep an eye on the page, because we might be able to get an early glimpse of the forecasts in the works before it will be released on the 27th of July, but just a wild guess. Also, why would they make this page if they weren't going to release a forecast? This page didn't work on June 27th when I last tried it so clearly something is up. Also, the page has a reserved server too, which might indicate they'll be some simulator, or some function where a lot of information will need to be sent back and forth. Also, through inspecting the webpage I found:



Not sure what all the numbers mean; maybe those long numbers that start with 15 and 16 indicate Trump's chance are around 16%?

[ProgressiveModerate]

Did they say it would be released on the 27th?

Nope, just my guess

[ProgressiveModerate]

https://projects.fivethirtyeight.com/2020-election-forecast/

This link sends you to a blank page that is not an error page. Literally any other simillar link send you to an error page. It's the same address they used for the 2016 forecast, just swap 2016 with 2020. Keep an eye on the page, because we might be able to get an early glimpse of the forecasts in the works before it will be released on the 27th of July, but just a wild guess. Also, why would they make this page if they weren't going to release a forecast? This page didn't work on June 27th when I last tried it so clearly something is up. Also, the page has a reserved server too, which might indicate they'll be some simulator, or some function where a lot of information will need to be sent back and forth. Also, through inspecting the webpage I found:



Not sure what all the numbers mean; maybe those long numbers that start with 15 and 16 indicate Trump's chance are around 16%?

Today, rows 3 and 18 went away. Looking at the states listed, it seemed to align with the class 1 senate map from 2016. My guess is they will replace it with the states from the class 3 senate for 2020. Some of the other random numbers went away as well. Now it looks like:

[ProgressiveModerate]

Does anyone else feel like their national tracker is a bit jumpy? New Biden +7 and +8 polls this morning made their average drop nearly an entire point, yet when Biden got the +15 WaPo poll, it barely budged the average.  It feels like the good polls don't push it up nearly as much as the "lower" polls drop it.

Not to mention, yesterday, Trump was at 41.6%, then he got a Trump 38 and Trump 40 polls, but somehow his average got pushed *up* to 41.9%.... am I missing something?

Some of that is because last time those polls had him lower than 38-40, and since they trash the old version of a poll after the new version come out, it's possible that a number below the average could push the average up because it replaces an even lower number

[ProgressiveModerate]

1% is a pretty significant drop in 1 day. I wonder if this holds or if later this week there's polls that bring the average back up

[ProgressiveModerate]

When Biden gains in the polls, the polls must be fake and wrong. When Trump gains in polls Biden is doomed to fail. The electoral map is still quite rosy to Biden:

[ProgressiveModerate]

When Biden gains in the polls, the polls must be fake and wrong. When Trump gains in polls Biden is doomed to fail. The electoral map is still quite rosy to Biden:

I don't see why GA should be so R for the Senate. Let's be generous to Republicans and say both are Tilt R, that still means there's a very real possibility at least one race flips. Texas seems underrated too as a competitive race, based on current margins it would probably be a low single digit win for Cornyn.

B/C of Georgia going to runoff, and Biden wins most of the time, so the environment is less good for Ds. Osoff can't win outright unless he gets above 50% which is a pretty demanding task in a state as polarized as GA

[ProgressiveModerate]

Why is everyone so upset about the model not being released. 538 is under no obligation to create a model; it just goes against tradition. We have other models like the Economists, JHK. If people are so upset, just go ahead and make your own model and see how easy/difficult it is to make.

[ProgressiveModerate]

https://twitter.com/wiederkehra/status/1287872080474603521

https://twitter.com/wiederkehra/status/1287872082991169538

https://twitter.com/wiederkehra/status/1287872085675536385

https://twitter.com/wiederkehra/status/1287872115585081344

https://twitter.com/wiederkehra/status/1287872089051955201

https://twitter.com/wiederkehra/status/1287872092990365697

Why are MN and NM deep red on all those maps!!??

[ProgressiveModerate]

Also, if this is the actual forecast I'd puke. IL, WA both more competative than AZ!? Give me a break.

[ProgressiveModerate]

They are just discussing their graphics and how they are going to present their model. I imagine the data is fake. Chill.

The tweet thread even says it has fake data.

Oh ok good. I saw the image and then I started freaking out for a bit

[ProgressiveModerate]

Also, if this is the actual forecast I'd puke. IL, WA both more competative than AZ!? Give me a break.

Vermont and New York are taking over.

Maybe the left really does take over the US, and there is a backlash against them

[ProgressiveModerate]

Maybe it'll be released tomorrow; the first of August

[ProgressiveModerate]

Nate Silver didn't post much on twitter today; maybe he's busy finalize it

[ProgressiveModerate]

Maybe it'll be released tomorrow; the first of August

If it gets released over the weekend it would be on Sunday during This Week With George Stephanopoulos on ABC.

Welp that didn't happen. However, they're clearly very close to releasing it, they added some more stuff in "Local Storage" upon inspecting the web page, however don't know what to make sense of what was added; it's mostly random numbers and lists of states. Could be "'fake data" for test, because I wasn't able too correlate the numbers with anything

[ProgressiveModerate]

At this point I just want it b/c I want a better aggregate for senate polls too, since RCP doesn't include even half the ones 538 does.

I made an aggregate of polls for the senate using 538 polling page to collect polls for my model:

https://docs.google.com/spreadsheets/d/1odwJDvqAY3UXreN8XY4Hmql9cJG5r7MCiPCTwzrAVA0/edit?usp=sharing

Not sure if these are what you're looking for, but feel free to use them

[ProgressiveModerate]

https://twitter.com/NateSilver538/status/1291462510332256262?s=20

Seems like they've completed the model, and based on the tweets from last week the graphics are at the very least almost done. Can't see why it's taking them so long

[ProgressiveModerate]

This is starting to become a bit of a joke, really. If Silver doesn't think his model is important enough to release a few months before the election, then why even release it at all.

Honestly, this should have been out for at least a month already, we have been getting plenty of polls for a long time now.

yeah, the "it's finished but i'm in no rush" thing truly has me gobsmacked. like what?? lmao.

Maybe he's worried it's too bullish on Biden? I just can't see how a fair statistical model that accounts for many factors can give Trump over a 1/4 chance of winning, although the general public seems to think the election is more of a tossup, so he's worried for being hammered. It could also be him trying to adjust the senate forecast; senate forecasts tend to be harder to code because there are unique circumstances in every race that are hard to account for in a purely statistical manner.

[ProgressiveModerate]

Honestly, the delayed release of the model is a blessing in disguise in some ways. It forces people to have more independent takes instead of always resorting to the model. It also encouraged me to make a model of my own which gave me valuable skills in Excel and taught me a lot about how the math I learned in school is actually applicable.

[ProgressiveModerate]

Release date: Wednesday morning

Ah thank you! JHK, the Economists, and my model will have to do until then

[ProgressiveModerate]

https://projects.fivethirtyeight.com/2020-election-forecast/

Here's the likely link to the forecast

[ProgressiveModerate]

https://twitter.com/NateSilver538/status/1292947361950564353?s=20

Wow... sad that they had to add that