538 model & poll tracker thread

[The Mikado]

Monday will be the 4th anniversary of the launch of the 2016 model.

I do think that it'll take longer this time. If I had to take a guess: they do like their numerically significant stunts, so maybe July 23rd, AKA the "100 days until the election" point.

[The Mikado]

A week from tomorrow makes sense as a launch date. A week from today is the 100 days until the election milestone, so the following Monday is a good start date for the model.

[The Mikado]

I do think that uncertainty is especially high at the moment because we're in a polling drought. I'd expect uncertainty to shrink quite a bit if there's a big wave of polls this weekend (which I think is likely as people want to get their pre-convention snapshots to compare with post-convention).

I do really miss the density of state polls we had in the 00s. I remember in the late 2000s, SUSA literally did a "Bush approval in all 50 states" monthly tracker for a while. No pollster would dare touch something like that now. If we had more actual polls of, say, Mississippi, it'd help 538's model get to 99% for Trump and eliminate some of the uncertainty.

[The Mikado]

I'm not going to complain about any of the given results.

I just want to:



Make 538 Aesthetics Great Again.

[The Mikado]

Could we please just have a f**king map on the front page like we did in 2008, 2012, and 2016? And much, much, much less bulls**t?

Jack Kersting's model is way prettier than this and he's just one teenager designing a website from scratch.

https://projects.jhkforecasts.com/presidential-forecast/

Less is more. Minimalism is always the way to go in website design, IMO.

I keep coming back to the 538 2008 model:



Why can't we just do this? That's all you need. Why is there all this extraneous stuff?

[The Mikado]

Biden's chances have been inching up...he's gone from 71 to 72 to 73 since the model launched.

At least this model seems more stable than the 2016 one where the chances of winning changed dramatically week by week

That race changed dramatically week to week, though, especially towards the end.

2020 is really steady.

The model is backdated every date to June 1st. Lowest it's had Biden's chances was 69% on June 8th, and highest it's had him was 79% for several days around June 27th. It's been static in that 10% range since June 1st.

[The Mikado]

In the vein of ridiculous 538 maps I'm seeing this one:

This is where Brad Pascale uses a monkeys paw to have Trump win Oregon

I'd be perfect if they also carried NM, Parscale's other great white whale.

[The Mikado]

My new favorite. We can all blame HockeyDude for Trump for four years.

[The Mikado]

It’s really bothersome how little information the design of this model makes easily available to viewers.  It seems like the only message they want to communicate is “I guess anything could happen?!?”.

Where’s the “polls only” forecast?
Where’s the “now cast”?
Where’s the easy link to the database of polls used to generate the projections?
Where’s the accounting for how individual “fundamentals” factors influence the projections?
Where’s the map??

All of this was available in the 2016 model, which actually performed quite well.  This is such a huge step backward.

He has said since 2017 that he's never doing another Nowcast and it was a mistake to make it, so I'm not surprised it's gone.

The rest, though, should be there. I really want that list of probabilities of various interesting things (Electoral tie, landslide, EV/PV split, etc) that they had in 2016.

[The Mikado]

It’s really bothersome how little information the design of this model makes easily available to viewers.  It seems like the only message they want to communicate is “I guess anything could happen?!?”.

Where’s the “polls only” forecast?
Where’s the “now cast”?
Where’s the easy link to the database of polls used to generate the projections?
Where’s the accounting for how individual “fundamentals” factors influence the projections?
Where’s the map??

All of this was available in the 2016 model, which actually performed quite well.  This is such a huge step backward.

He has said since 2017 that he's never doing another Nowcast and it was a mistake to make it, so I'm not surprised it's gone.

The rest, though, should be there. I really want that list of probabilities of various interesting things (Electoral tie, landslide, EV/PV split, etc) that they had in 2016.

There is a link at the very bottom of the "National overview" page that will download model outputs. Besides more precise numbers, there is a file called "Scenario Analysis" that includes the chances of:

1: An Electoral College tie. (0.3%)
2: A presidential recount. (0.5%)
3: Trump winning the popular vote. (16.6%)
4: Biden winning the popular vote. (83.4%)
5: Trump winning the popular vote but losing the Electoral College. (0.1%)
6: Biden winning the popular vote but losing the Electoral College. (10.2%)
7: Trump winning a majority. (12.3%)
8: Biden winning a majority. (78.5%)
9: Trump winning by over ten points. (1.2%)
10: Biden winning by over ten points. (31.9%)
11: The same map as 2016. (0.0%)
12: Trump winning a Clinton state. (41.1%)
13: Biden winning a Trump state. (88.0%)

All of this is interesting, and it's valuable to have, but it is definitely odd that it requires a separate file download to find.

This is very cool. Also lol re: 0.0% same map as 2016.

Trump by over 10 points at even 1.2% is stunning to me. If he's saying Trump has a 12.3% chance of even winning the popular vote at all, that "Trump wins by 10" makes up a tenth of those scenarios is shocking.

[The Mikado]

Posting this here rather than the weird maps thread because I think it makes an interesting point: if Trump loses FL he's basically totally screwed even in relatively favorable situations elsewhere.

He wins MN and NH and holds onto PA and WI and still loses the election.

[The Mikado]

I noticed in the current "cloud" figure of 100 sample elections that Biden wins >350 electoral votes about 37% of the time.  Trumps wins >350 electoral votes 1% of the time.  So many the uncertainty finally is contracting somewhat.

If you create a map where Biden wins everything in their "Very likely" category (95%+ chances) and Trump wins everything else, you end up with this:

362-176 Trump. It's on the verge of plummeting to 349-189 because Virginia is RIGHT on the cutoff line of that Very Likely category.

[The Mikado]

Model is now showing Biden at an 87% chance of winning Michigan while he still shows an 86% change of winning CO.

This model is weird.

I mean, it makes sense in a way, that states with a lot of polling the model is a lot more confident in than states with just a little polling. If CO had the polling density of MI Biden's chances there would probably be 95%+.

[The Mikado]

Why do they have an option for a polls-only model for the Senate but not the presidency?

Because it would show Biden with higher chance of winning.

Nate is deathly afraid of making it look like Trump has no chance, so he's cooked the model this year to ensure Trump gets all of the favors... including gutting the polls-only model

I'm wondering what happens when we get to just one or two days before the election, assuming state polls stay steady. Silver said that Biden would have more than a 90-something % chance of winning if the election were held tomorrow, but I doubt he's comfortable with that being the last prediction on his model the day of the election.

I think he'd let it through if it was the final model output. 2012 ended up at Obama with a 91% win chance on 538 and that race looked close on paper because there was a pretty big polling error in Romney's favor. (Final polling averages had Obama ~+1 nationally rather than the Obama ~+4 result)

[The Mikado]

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

Biden, apparently, has a 14% chance of winning Mississippi.

1 in 7 chance. Of winning Mississippi.

1 in 7.

EDIT:

For comparison: Trump has a 12% chance of winning Michigan. So 538 thinks its more likely that Biden wins Mississippi than that Trump wins Michigan.

[The Mikado]

This is just embarrassing.

40,000 simulations or not, this shouldn't show up if you do 40,000,000 simulations.

[The Mikado]

He said in the aritcle that went up last night that Biden's odds would be 91% on Election Day if things stay consistent.

Given that 91% was exactly the odds he gave Obama 2012 on Election Day when Obama only had a 1.5% national polling average lead over Romney, it says a lot about how much more cautious this model is that it'd give Biden's much larger lead the same chance.

[The Mikado]

There's discussion on the Congressional boards about the new House model, but I wanted to take a moment to look at this, which Nathaniel Rakich highlighted:



63% Chance Dem Trifecta
32% Chance Divided Government
5% Chance GOP Trifecta

Does make some intuitive sense to me, though I'd put GOP Trifecta even lower. I also think that, while it's certainly not LIKELY, Trump winning but D House and Senate seems more likely to me than a 2% chance.

[The Mikado]

Today or tomorrow, Biden will probably cross from "Favored" to "Clearly Favored"

[The Mikado]

IBD's at least a real poll.

The Spry poll is really, really weird and I'd love to see a deep dive into this "Women's Liberation Front."