Models starting to converge (update 10/17)
       |           

Welcome, Guest. Please login or register.
Did you miss your activation email?
March 28, 2024, 02:28:31 PM
News: Election Simulator 2.0 Released. Senate/Gubernatorial maps, proportional electoral votes, and more - Read more

  Talk Elections
  Election Archive
  Election Archive
  2020 U.S. Presidential Election (Moderators: Likely Voter, YE)
  Models starting to converge (update 10/17)
« previous next »
Pages: [1] 2
Author Topic: Models starting to converge (update 10/17)  (Read 2539 times)
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« on: September 26, 2020, 08:19:51 AM »
« edited: October 17, 2020, 09:46:49 AM by GeorgiaModerate »

As of this morning (9/26), here are Biden's and Trump's chances of winning in each model:

Economist: 84-15
DDHQ: 83.5-15.9
JHK: 79.4-20.6
538: 77-23
Logged
Non Swing Voter
Junior Chimp
*****
Posts: 8,181


Show only this user's posts in this thread
« Reply #1 on: September 26, 2020, 08:46:15 AM »

is that the highest 538 has ever been for Biden?
Logged
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« Reply #2 on: September 26, 2020, 08:47:52 AM »

is that the highest 538 has ever been for Biden?

It briefly ticked up to 78 for a few hours earlier this week, but otherwise 77 is the highest value since they launched it.
Logged
ProgressiveModerate
Atlas Icon
*****
Posts: 13,569


Show only this user's posts in this thread
« Reply #3 on: September 26, 2020, 09:08:39 AM »

This makes sense. As election day approaches, uncertainty will begin to narrow in, and therefore, models like 538 which once had high uncertainty, and models like the Economists, which were always pretty confident, will begin to converge. My model also falls in this range with Biden at an 83% chance of victory, but if I used the uncertainty from 2 months ago that number is closer to 70%. I'll be curious to see what the uncertainty is for 538's model by election day, because to me, it still seems a bit high. I'm also curious what fundementals in JHK's model make Trump slightly outperform polls; the economy is bad, Biden has raised and spent more money than Trump, polling suggests that Trump has more soft support and that undecides lean slightly towards Biden, ect. Always appreciate the models though, and it'll be interesting to see how they hold up come November.
Logged
TiltsAreUnderrated
Junior Chimp
*****
Posts: 9,771


Show only this user's posts in this thread
« Reply #4 on: September 26, 2020, 09:47:22 AM »

It also makes sense for 538 because their "economic projections" bet-hedging element grows weaker in the model as we draw closer to election day. This was very optimistic for Trump and based off post-lockdown rapid economic growth being treated as high GDP growth in an otherwise normal environment.
Logged
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« Reply #5 on: September 28, 2020, 06:05:20 PM »

Current numbers, change for Biden from initial post:

Economist: 85-14 (+1)
DDHQ: 80.4-19.0 (-3.1)
JHK: 79.7-20.3 (+0.3)
538: 78-22 (+1)

Weird that DDHQ has moved in the opposite direction from the others.
Logged
Donald Trump’s Toupée
GOP_Represent
Jr. Member
***
Posts: 1,536


Show only this user's posts in this thread
« Reply #6 on: September 28, 2020, 06:12:25 PM »

You can’t model this election as no one can possibly predict COVID election turnout. We all know this - stop lying to yourself.
Logged
pbrower2a
Atlas Star
*****
Posts: 26,842
United States


Show only this user's posts in this thread
« Reply #7 on: September 28, 2020, 06:54:59 PM »

People will watch the Tuesday debate, and then the last people to make up their minds will mostly do so then. People will tolerate stuttering more than they will any "Captain Queeg" performance.

I don't ordinarily predict psychological effects... but what happened to Brad Parscale could easily shake President Trump up. We do not know why he did what he did, and anything that he says while under psychiatric treatment is strictly confidential between him and the staff. We won't know whether such is job-related or the consequence of some medical calamity.

Trump knows more than we do. Parscale will not be talking for a while.     
Logged
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« Reply #8 on: September 30, 2020, 05:39:53 PM »

Current numbers, change for Biden from two days ago:

Economist: 87-13 (+2)
DDHQ: 82.2-17.3 (+1.8
JHK: 81.7-18.4 (+2.0)
538: 78-21 (no change for Biden, but Trump lost 1 point)

I plan to update this every couple of days until the election.  If anyone has other models to add, feel free to suggest them.
Logged
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« Reply #9 on: October 04, 2020, 06:27:45 PM »

Current models, changes from last update (Sep. 30):

Economist: 89-11 (+2/-2)
DDHQ: 82.2-17.2 (nc/-0.1)
JHK: 81.4-18.6 (-0.3/+0.2)
538: 81-19 (+3/-2)
Logged
ProgressiveModerate
Atlas Icon
*****
Posts: 13,569


Show only this user's posts in this thread
« Reply #10 on: October 04, 2020, 07:47:03 PM »

Current models, changes from last update (Sep. 30):

Economist: 89-11 (+2/-2)
DDHQ: 82.2-17.2 (nc/-0.1)
JHK: 81.4-18.6 (-0.3/+0.2)
538: 81-19 (+3/-2)

Funny how on a state by state level, 538 went from being the most bullish on Trump to now the most bullish on Biden. Obviously, as we well know their model has higher uncertainty, so it makes sense they still give Biden the lowest chance, but it's just interesting
Logged
Real Texan Politics
EEllis02
Sr. Member
****
Posts: 2,610
United States


Political Matrix
E: 0.39, S: -1.57

P P P
WWW Show only this user's posts in this thread
« Reply #11 on: October 04, 2020, 10:36:41 PM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?
Logged
Buzz
Sr. Member
****
Posts: 2,186


Show only this user's posts in this thread
« Reply #12 on: October 04, 2020, 10:37:45 PM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?
almost exactly the same lol
Logged
Horus
Sheliak5
Atlas Icon
*****
Posts: 11,556
United States


Show only this user's posts in this thread
« Reply #13 on: October 04, 2020, 10:38:21 PM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?

I think 538 had her lower than where Biden is (though not by much) but at least one forecaster had her at 99% lol
Logged
Real Texan Politics
EEllis02
Sr. Member
****
Posts: 2,610
United States


Political Matrix
E: 0.39, S: -1.57

P P P
WWW Show only this user's posts in this thread
« Reply #14 on: October 04, 2020, 10:57:22 PM »

Well then I guess we'll see where it is by halloween lol
Logged
Pericles
Atlas Icon
*****
Posts: 17,066


Show only this user's posts in this thread
« Reply #15 on: October 04, 2020, 10:58:33 PM »

You can’t model this election as no one can possibly predict COVID election turnout. We all know this - stop lying to yourself.

A pretty insane turnout differential would be required to overturn Biden's 8-point lead.
Logged
politics_king
Jr. Member
***
Posts: 1,591
United States


Show only this user's posts in this thread
« Reply #16 on: October 04, 2020, 11:00:37 PM »

You can’t model this election as no one can possibly predict COVID election turnout. We all know this - stop lying to yourself.

Like Trump and his comb over?
Logged
ProgressiveModerate
Atlas Icon
*****
Posts: 13,569


Show only this user's posts in this thread
« Reply #17 on: October 04, 2020, 11:12:18 PM »

You can’t model this election as no one can possibly predict COVID election turnout. We all know this - stop lying to yourself.

That's why I decided to use a t-distribution for my model; it makes the tails on the range of the outcome a bit fatter, hence why Biden still has a 0.02% chance in WY and Trump has a 0.05% chance in NY, unlike other models.

Here is the link to the current probabilities: https://docs.google.com/spreadsheets/d/1HLtdIjLW05pzDwXz0aneLRsg8Jdg2x8JaBtkodOCf_8/edit?usp=sharing

Remember; a model is just a good estimate based on the available data, and while it isn't perfect, does offer a purely statistical analysis on the picture of the race
Logged
The Mikado
Moderators
Atlas Star
*****
Posts: 21,677


Show only this user's posts in this thread
« Reply #18 on: October 04, 2020, 11:18:28 PM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?

At this far from the election? 538 had Clinton at 77% chance to win. She'd peak at 85% chance to win about a week later before falling to ~65% and then rebounding to the 71% she ended up at on election day.
Logged
OBD
Junior Chimp
*****
Posts: 5,575
Ukraine


Political Matrix
E: -5.16, S: -6.26

Show only this user's posts in this thread
« Reply #19 on: October 04, 2020, 11:20:38 PM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?

At this far from the election? 538 had Clinton at 77% chance to win. She'd peak at 85% chance to win about a week later before falling to ~65% and then rebounding to the 71% she ended up at on election day.
I mean, after 2016 it's likely all these models are far more cautious (thanks to the shock Trump win), so it's really comparing apples to oranges. Hence why there aren't any 99% Biden forecasts despite conditions being objectively better for him than Hillary.
Logged
Non Swing Voter
Junior Chimp
*****
Posts: 8,181


Show only this user's posts in this thread
« Reply #20 on: October 05, 2020, 03:23:36 AM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?

At this far from the election? 538 had Clinton at 77% chance to win. She'd peak at 85% chance to win about a week later before falling to ~65% and then rebounding to the 71% she ended up at on election day.

Hmm, the 538 model seems pretty reasonable.  This makes me want to follow them more this year.
Logged
Tintrlvr
Junior Chimp
*****
Posts: 5,284


Show only this user's posts in this thread
« Reply #21 on: October 05, 2020, 06:42:18 AM »

I hate to be "that guy", but for the record what did the models look like for Hillary at this time in 2016?

At this far from the election? 538 had Clinton at 77% chance to win. She'd peak at 85% chance to win about a week later before falling to ~65% and then rebounding to the 71% she ended up at on election day.

As others noted, though, it’s not an apples-to-apples comparison, as the 2016 538 model was much more confident generally. Plug its polls into the 2020 model and you’d probably get around 65% odds right now (and if you put today’s polls into the 2016 model, you’d get close to 90% odds).
Logged
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« Reply #22 on: October 06, 2020, 05:52:34 PM »

Today's probabilities, with change since last update (Oct. 4):

Economist: 90-10 (+1/-1)
JHK: 86.6-13.3 (+5.2/-5.3)
538: 83-17 (+2/-2)
DDHQ: 79.9-19.5 (-2.3/+2.3)

I don't know what's going on with DDHQ.
Logged
ProgressiveModerate
Atlas Icon
*****
Posts: 13,569


Show only this user's posts in this thread
« Reply #23 on: October 06, 2020, 05:53:15 PM »

Today's probabilities, with change since last update (Oct. 4):

Economist: 90-10 (+1/-1)
JHK: 86.8-13.3 (+5.2/-5.3)
538: 83-17 (+2/-2)
DDHQ: 79.9-19.5 (-2.3/+2.3)

I don't know what's going on with DDHQ.


It seems they're model doesn't have it so taht the uncertainty decreases as election day approaches
Logged
GeorgiaModerate
Moderators
Atlas Superstar
*****
Posts: 32,350


Show only this user's posts in this thread
« Reply #24 on: October 08, 2020, 12:24:05 PM »

Current percentages, with changes from last update (10/6):

Economist: 91-9 (+1/-1)
JHK: 86.4-13.6 (-0.2/+0.3)
538: 85-15 (+2/-2)
DDHQ: 81.3-18.1 (+1.4/-1.4)
Logged
Pages: [1] 2  
« previous next »
Jump to:  


Login with username, password and session length

Terms of Service - DMCA Agent and Policy - Privacy Policy and Cookies

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Page created in 0.051 seconds with 13 queries.