Jump to content
270soft Forum

Nate Silver guarantees there is a 10% chance Trump wins!!!


PoliticalPundit
 Share

Recommended Posts

2 minutes ago, ThePotatoWalrus said:

A lot of us entertained the realm of possibility concerning a Trump presidency. I forget if I did but in particular @vcczar gave him a good chance and @TheMiddlePolitical I think actually called Trump's victory in the rust belt super accurately.

However no one who predicted him winning in 2016 says that today. There's no chance.

Hell, I actually supported him in 2016. Lot has changed in 4 years.

  • Upvote 1
Link to comment
Share on other sites

Just now, Reagan04 said:

Hell, I actually supported him in 2016. Lot has changed in 4 years.

Yeah watching your change during his presidency was interesting. You tried to support him in 17, became undecided in 18, and supported Biden during the primaries I think. I would have voted for Gabbard, and did in the Ohio Primaries.

Link to comment
Share on other sites

Just now, ThePotatoWalrus said:

Yeah watching your change during his presidency was interesting. You tried to support him in 17, became undecided in 18, and supported Biden during the primaries I think. I would have voted for Gabbard, and did in the Ohio Primaries.

Yeah I tried in 17, but a number of things shook me by 18. I supported Delaney, Klobuchar, or Biden in the primaries but really I would have loved a serious GOP challenge. That or Justin Amash seeking the Libertarian nomination.

Link to comment
Share on other sites

Just now, Reagan04 said:

Yeah I tried in 17, but a number of things shook me by 18. I supported Delaney, Klobuchar, or Biden in the primaries but really I would have loved a serious GOP challenge. That or Justin Amash seeking the Libertarian nomination.

What's wrong with mama Jo? No real government experience?

Link to comment
Share on other sites

1 minute ago, PoliticalPundit said:

Where did I say it was?? 🧐

 

Trump's policies and being a Libertarian were inherently going to end w conflict and division over 4 years. 

 

Frankly I'm surprised as a Libertarian he didn't vote 3rd party in 2016 

He was more conservative back then. Ted Cruz, Fiorina and Ben Carson in that order I think right? @Reagan04

Link to comment
Share on other sites

17 minutes ago, PoliticalPundit said:

Let's be honest

 

You're a Libertarian 

 

there was no way you'd end up supporting him long term once he won 

I’m a conservative Libertarian. Supported Cruz in 2016. If Trump were an actual conservative, a Reaganite, etc, I’d support him. But he’s not, he’s an alt-right authoritarian. And he is the death of the conservative movement. That’s why I oppose him so viscerally. If he is re-elected, Reagan’s legacy and the conservative movement is destroyed. Liberty isn’t passed down through our bloodstream, we’ve got to fight for it every generation. I’ve got that quote hanging on my wall, beating Trump is how we save conservatism this generation.

  • Like 1
Link to comment
Share on other sites

14 minutes ago, ThePotatoWalrus said:

He was more conservative back then. Ted Cruz, Fiorina and Ben Carson in that order I think right? @Reagan04

Walker, Fiorina, Carson, Cruz over the course of the primaries.

I remain a Fiorina Stan.

  • Upvote 1
Link to comment
Share on other sites

22 minutes ago, PoliticalPundit said:

Damn he really did not like Trump 😂

It’s not that I didn’t like him, it’s that he was the worst, least conservative candidate running.

Link to comment
Share on other sites

On 11/1/2020 at 11:35 AM, Defiant said:

More like people don't understand basic probability.

Something with 70% odds doesn't always happen all the time.  In fact, almost a third of the time it *doesn't* happen.

If he says that Biden has a 90% chance of winning, he is predicting a Biden victory.

On 11/1/2020 at 12:04 PM, Actinguy said:

He wasn’t wrong.  People think anything less than a 50% chance means a 0% chance.  He’s not responsible for people not knowing what a percentage chance means.

Yes he is because the number means nothing.  If he thinks Trump only has a small percentage chance to win, then he is predicting a Biden victory.  The percentage is just a copout to be able to cover up being way off with his prediction.  That's what happened in 2016, and that is what'll happen this year if Trump wins.

  • Upvote 1
Link to comment
Share on other sites

17 minutes ago, jvikings1 said:

The percentage is just a copout to be able to cover up being way off with his prediction.

On the one hand, with a computer sim, the forecaster can be just stating the % of runs where a candidate wins. So, Candidate A wins 56% of the time, say. I think it's completely legitimate for someone basing a prediction on a computer model to say "It had this result x% of the time." If that's all they say, fine.

The problem is going from that, to saying in reality there therefore was a 56% chance of Candidate A winning. What does that even mean? We're no longer talking about a computer sim being run a certain number of times - presumably reality is only being run once.

But another big problem is then going on to say "Whatever happens, as long as I have a chance that isn't 100% or 0%, the result corroborates my computer model." In some weak sense this might be true, but it sounds more like charlatanry to me. Silver tries to make this into a basic point about probabilities, but what it actually does is make his model unfalsifiable.

Anyone can play that game, as long as they say a particular result doesn't have a 100% or 0% likelihood.

You can see the problem with this because multiple forecasters will typically have different %s, and they can all say they were correct in that sense. But they all can't be correct in their % assignment if they have different %s. So either it's not a very useful claim, or it's incorrect.

 

  • Upvote 1
Link to comment
Share on other sites

8 minutes ago, admin_270 said:

 

You can see the problem with this because multiple forecasters will typically have different %s, and they can all say they were correct in that sense. But they all can't be correct in their % assignment if they have different %s. So either it's not a very useful claim, or it's incorrect.

 

Indeed. That's why I tend to give ranges for outcomes.  If I see one candidate have a 70% or higher chance of victory, that tells me they think he will win.  If it's something like 60%, then its a lean but not definite prediction.

Take this election for instance.  If Trump wins despite having a 10% chance of victory, I am gonna call BS on the model (even though it technically predicted a win in 10% of the scenarios).

  • Thanks 1
Link to comment
Share on other sites

1 minute ago, jvikings1 said:

If he says that Biden has a 90% chance of winning, he is predicting a Biden victory.

 

If I say that I'm going to roll a die, and there's 66% chance that the number rolled will be a 3 or higher, is that a prediction that a 3, 4, 5 or 6 will be the result of the die rolled?

Link to comment
Share on other sites

38 minutes ago, admin_270 said:

How do you verify your % chance is correct?

Either by looking at the physics involved or, more simply, run a very large number of trials of dice rolling (preferably while playing a game), and seeing that the odds approximate the estimate that we know it is.  (If it's dramatically different, than either the die is a loaded die, or the laws of the physics have changed).

 

And you can examine the estimates by election models the same way.  Run it on a very large number of trials, and, for example, if you run it on 100 different elections where the estimates are that 60% of the time, one of the candidates win and 40% the other candidate, then check to see if, indeed, about 60% of the time the person ahead wins, and the other 40% of the time, the underdog ends up winning.

But my point, going back, is that I could just as easily have states that there was a 66% chance that the result would be 4 or less.  Which would be a very different "prediction" even though they're both correct.

Similarly, I could say that there was ~16% chance for each side.  And in that case, with no one side getting more than 50%, would that mean I was predicting that the die would never land?

Link to comment
Share on other sites

32 minutes ago, Defiant said:

Either by looking at the physics involved or, more simply, run a very large number of trials of dice rolling (preferably while playing a game), and seeing that the odds approximate the estimate that we know it is.  (If it's dramatically different, than either the die is a loaded die, or the laws of the physics have changed).

Correct.

32 minutes ago, Defiant said:

And you can examine the estimates by election models the same way.  Run it on a very large number of trials, and, for example, if you run it on 100 different elections where the estimates are that 60% of the time, one of the candidates win and 40% the other candidate, then check to see if, indeed, about 60% of the time the person ahead wins, and the other 40% of the time, the underdog ends up winning.

But how can you test Presidential election prediction %s this way? They only happen once every 4 years. How do you get a 'very large number of trials'?

To make matters worse, Silver's model changes from cycle to cycle. So his 2016 model can't be verified in 2020, because it's a different model!

So either 1. Silver is making a claim about who will win, or 2. He's making a claim about the % chance someone will win.

If it's 1., then we know he was wrong in 2016, or if it's 2., then it's practically unfalsifiable - either result is compatible with his model, and it will never be tested again.

  • Upvote 1
Link to comment
Share on other sites

Actually, no, 51 elections happen every four years, and another 468 or so elections happen every two years.  And then there are primaries and other potential races, etc.   Well, maybe not that many, since I imagine it's mostly the competitive races that get polled significantly.  But that's still got to be, what,  maybe 75 races every couple of years?

(You can probably also take a look at the margins they estimate and compare them to the results.)

Link to comment
Share on other sites

8 minutes ago, Defiant said:

Actually, no, 51 elections happen every four years, and another 468 or so elections happen every two years.  And then there are primaries and other potential races, etc.   Well, maybe not that many, since I imagine it's mostly the competitive races that get polled significantly.  But that's still got to be, what,  maybe 75 races every couple of years?

(You can probably also take a look at the margins they estimate and compare them to the results.)

No, wrong. Primaries, House races, Senate races, Gubernatorial races - they are all different kinds.

What are these 51 elections you are referring to that you think are relevant? Do you mean state-by-state results for the Presidential election?

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...