Are any of you all following the election forecast at FiveThirtyEight ?

[Le Roi de Pot]

Trump's odds have improved to nearly one in three chances of winning electoral college based on simulation

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




I don't know what the underlying distributions (for the variables affecting the result such as voter turnout among different subgroups) are for the simulation - and it would have been nice if they provided the sensitivity analysis of the variables to the output .. but I dont see it in the data they have made avaible for the output .. But lets take it seriously .. most likely the scenarios that help Trump are increased white blue collar turnout in key states. I dont know if we can blunt that with our messaging .. there is no way to reach them .. only hope is to get our folks in force as well.


But the bottom line is We need to get the bloody vote out .. what strategies do we have other than phone banking? .. can we write letters to get people to register and vote early?

On Edit: I did not mean this to be a downer --- We still have a much better chance of winning and even have decent change ( 33% ) of a BIDEN LANDSLIDE ( defined as double digit popular vote margin) -- we can make it happen -- we need to crush those fucking bastards ..

[pwb]

1. 33% for Trump. 66% chance for Joe?

What is the problem?

[bearsfootball516]

3. It's actually 70/30 right now.

Slightly better than Hillary in 2016.

[mr_lebowski]

4. I thought Hillary was around 72% the day of the vote, per 538 ... was it lower? (nt)

[Polybius]

14. That's correct

That one was 72-28 on Election Day.

[mr_lebowski]

19. Exactly. Which means being up 66%-33% is in no way a comfortable lead (nt)

[Le Roi de Pot]

5. That is not the spead of votes -- it is odds of winnng.

imagine having a jar with three thousand balls .. right now roughly 2000 have Biden's name and 1000 has Trump's name .. and a ball will be drawn at random ..

So Trump has a very good shot at winning the electoral college.

[Thekaspervote]

9. Oh pleassse!!

[Le Roi de Pot]

10. and we have the same odds of making it a landslide ( 1 in 3)

based on the data

[Music Man]

30. Well said.

On the whole, I'd rather be Biden right now. But there's some denialism by many on this site about the fact that only a few factors can swing things.

Because of the states in play, one of two things is going to happen on Election Day: Biden will win in a landslide or lose in a squeaker. Democratic turnout in the swing states will determine that.

[obamanut2012]

33. lol that isn't what that means

You have zero idea how to read odds.

Well, actually I think you know what you are doing.

Your concern is noted.

[Le Roi de Pot]

34. If the polls were say 90 - 10 or even 60-40 ...the chance of Trump winning would be ZERO

however - if the polls were like say 52-46 ... the chance that Trump actually wins start to go up

The two are different

[whopis01]

36. Actually, that is what it means exactly.

Fivethirtyeight is saying that Biden has a 70% chance of winning and Trump has a 30% chance of winning. (As of the time when I am posting this).

That’s slightly different than the 67% to 33% quoted by the other poster. But that aside, their analogy is correct.

Right now, according to fivethirtyeight, it is like have 2100 balls with Biden’s name and 900 balls with Trump’s name in a jar and you pick one randomly.

(Not sure why the poster above went wi a total of 3000 - but I used their numbers for consistency)

[whopis01]

35. It was 73% Biden, 27% Trump not very long ago.

The direction is troubling.

[sweetloukillbot]

2. I noticed it this morning too

It had peaked around 74/26, then dropped back down, but the polls don’t seem to support it, unless it’s from the C- weighted polls in Minnesota and Wisconsin

[Chainfire]

6. I gave up on both polling and gambling odds after the last presidential election.

On the day preceding the election, there was no doubt that Clinton would win....

What do you think, in regards to polling, has changed since then?

Remember that Trump is infinitely more powerful than he was four years ago.......

[Thekaspervote]

7. Nate Silver libertarian sometimes dotard defender is not the best pollster to follow....period!!

Last edited Wed Aug 26, 2020, 12:40 PM - Edit history (1)

https://www.commondreams.org/views/2019/10/30/nate-silver-making-he-goes

https://www.electoral-vote.com/

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


Right now, our model thinks Joe Biden is very likely to beat Donald Trump in the electoral college.
Chance of winning
the electoral college Chance of winning
the most votes Predicted range of electoral college votes (270 to win)
Joe Biden
Democrat
around 9 in 10
or 90%
better than 19 in 20
or 98%
220-428
Donald Trump
Republican
around 1 in 10
or 9%
less than 1 in 20
or 2%
110-318

More at the links
https://projects.economist.com/us-2020-forecast/president?utm_campaign

[Fiendish Thingy]

20. NAte Silver is not a pollster, he's an analyst with a good record

He was on the money in predicting the 2016 popular vote, and was within the MOE for the rust belt states.

We ignore 538 at our own peril.

[Awsi Dooger]

27. Nate Silver is light years superior to 270towin.com

Current polling is only partially relevant. Nate understands that while other compilers/analysts fall prey to the stupidity of absorbing current/recent polling as all you need to know

66% is the equivalent of a 4 point favorite in a football game. The mistake that was made around here in recent months was insisting that Trump's support could only collapse, that he would drop to 35% or even 30% approval. That was 100% ignorant of partisan mindset as well as the applicable math. Those stacks of self-identified conservatives and right-leaning moderates do not budge, especially not when an incumbent is running for re-election in the most favorable scenario in American politics...his party in power only one term.

Also, as I have long emphasized, Hispanics are loyal to the presidential incumbent. Issues mean nothing. Trump is going to gain among Hispanics even after treating them even worse than anticipated.

We need the right track/wrong track number to stay low. We need independents to remain on Biden's side. We need effective messaging down the stretch. In particular we need to understand that Texas is not going to happen and those funds should be redirected to Florida.

If those variables fall in line Biden should hold on for a narrow victory

[marble falls]

8. Not me.

[radius777]

11. He's an incumbent president with high support amongst his own (racist) party.

And he will do anything to win - legal and illegal.

I read somewhere that the Rasmussen polls and the state polls in the Midwest are affecting things.

[Thekaspervote]

12. Links please

[radius777]

22. Check out this Morning Joe segment.

There's a new CNBC poll showing tightening in the swing states.

[Klaralven]

13. They're Monte Carlo simulation electoral college results based on the error distributions of polls

So if a poll for a given state polled 1000 voters and 520 chose Biden and 480 chose Trump, the poll would be reported as 52% for Biden and 48% for Trump.

But the sampling error on 520 samples is something like +-23. So there is about a 68% chance that the true situation is that Biden is between 497 and 543. In fact there is about a 16% chance that Biden's support is actually less than Trumps.

So in a Monte Carlo simulation, in 16% of the runs, that state would fall into the Trump column instead of the Biden column.

I'm sure 538s analysis is more completed and accounts for historical errors of each pollster, as well as correlation of errors between states, but that is the gist of it.

[Le Roi de Pot]

15. I thought it included turnout models

That way I can use a poll of registered voters and simulate election resutls..

But you could be right.. I ve done bunch of discrete event Monte Carlo simulations in grad school for a Port Authority NY/NJ study .. We had many sub models ... Ship size / draft , Tides , port capacity etc.. We did a multi year simulation to determine capacity expansion stratgies ... One of the constraints was that we couldnt dredge a lot of Newark Bay due to PCB pollution ( Thanks Jack Welch)

[KS Toronado]

16. Is 538 looking at the post office's efforts in blue states/areas to not deliver mail-in ballots,

until after Nov 4th? Oh it's coming alright, this election will have the largest number of
undelivered mail-in ballots we have ever seen. Mailing your ballot this year is like voting
for trumpig.

[LisaL]

18. Oh give it a rest.

I still get mail.
People should make sure they ask for ballots early, fill them out quickly, and drop them into the drop boxes if possible (some states don't allow that).

[KS Toronado]

24. NO! trumpie put dejoy in charge of the P.O. to help him win the election, & he's going to.

[pwb]

29. Relax our Postal workers are professionals

What is at the top matters not. Every postal worker takes an oath and they are a force on their own.

[texasfiddler]

21. Statistics is like a bikini

What it shows is interesting. What it hides is vital.

[Fiendish Thingy]

23. Currently 69/30 odds in favor of Biden, with 1/100 chance of a tie

Forecast has been changed from Biden being favoured to win, to being slightly favoured to win.

Interestingly, the national poll aggregate hasn’t changed- it’s still 52/46, although Biden’s projected EV’s have dropped from 324 to 317.

Remember that 538’s forecast model is only about 50% based on polls, with the other 50% based on other factors.

[Awsi Dooger]

31. "538's forecast model is only about 50% based on polls, with the other 50% based on other factors"

I wish I had seen your sentence before posting earlier in this thread. Nate's approach is markedly superior to other models because he understands that aspect, that polling is only part of the story, and you cannot take polling at face value.

As a political bettor since 1992 I cannot imagine wagering based on the polls alone. That would be incomparably stupid. If all you know is the polling then you don't know much of anything, because all you know is exactly what everyone else knows. How could there possibly be an edge? There are situational variables all over the place that detour the polling in one direction or another.

In my early years on this site I used to post numerical threads included a PAN or Partisan Adjustment Number for each state. It was an analysis of how the polling in each state tended to err toward one side or another, and normally consistently in the same direction. I quickly gave up on posting it because it was a complete waste of time. Nobody cared. Nobody posted in the threads. Or the posters who did take interest and post in the threads, like Jiacinto and Dolstein, were no longer posting here.

Regardless, I still have those numbers and the states Biden is relying on have red PAN, meaning the polling in that state is overly favorable to blue so the PAN adjustment is in the red direction. That type of thing makes Biden's edge more tenuous than it appears at first glance.

[Javaman]

25. We are still two months out

Plenty of time for the orange asshole to continue to fuck up

[Fiendish Thingy]

26. 3 new national polls out today, including Rasmussen, which has Biden 46/Trump 45 nt

[Le Roi de Pot]

28. Rasmussen has all the credibility of Donald Trump

[LincolnRossiter]

32. What about this is so difficult to grasp?

Right now Biden is favored (by about 2:1) to win the election. That approximately 1/3 chance that Trump has is not insubstantial. If I told you there was a 1/3 chance that you’d die in a car accident on your way to work tomorrow, you’d probably have the good sense to call in sick. Silver gave Trump about a 1/4 chance in 2016 and when Trump won that somehow made the forecast “wrong.” It’s like people don’t know one damned thing about odds or statistical probability. The odds right now are saying that not only may Trump win in 3 of every 10 elections run under the current circumstances. He actually should win in 3 of every 10.

Biden has the upper hand in this race, but this is by no means a done deal. Trump very well could get re-elected. Donate and volunteer as though you understand that.