Trump could win presidency: Yes or No?

[PartSkeptic]

(snip)can suddenly become a responsible politician.

Name one!

 

[The Great Zaganza]

Bu

Name one!

Hillary Clinton: she is being held responsible for everything she's ever been involved with, no matter how loosely.

 

[The Don]

On the polls only forecast Trump is up to 48.5% at fivethirtyeight.com

http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

IMO Trump as President is almost inevitable now

IMO nothing Hillary or the Democratic Party can do will change a majority of the US electorate's mind that Hillary is untrustworthy and crooked and that Trump is a first class businessman with a plan for the US economy - despite all the evidence to the contrary on both counts.

I am as disappointed about the US electorate as I was about the UK electorate about Brexit but it's the same thing....

- Wanting anything different
- Believing the lies
- Not believing experts

Trump will also have the house and senate so a clear mandate to install a far-right SCOTUS

 

[aleCcowaN]

On the polls only forecast Trump is up to 48.5% at fivethirtyeight.com

http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

IMO Trump as President is almost inevitable now

Hold your horses!

It's still Clinton + 1.5 to 2 and Trump is two battleground states short.

I admit that Trump might become the first president of a democracy selected by ISIS and/or by Anthropogenic Global Warming (and/or bacteriological warfare ), but unlikely yet.

 

[PartSkeptic]

(snip)
IMO Trump as President is almost inevitable now

(snip)

Ah. A realist.

Today CNN had a political commentator say that Trump would push the "perceived but incorrect view" that the US has economic trouble. CNN then played a clip of one of the poor West Virginia towns in serious trouble.

The commentator looked stressed.

Which is the reality? That interest rates at zero, incomes stagnant, and growing income disparity is just a Trump fantasy?

I will be setting my alarm clock for 2 am to watch the debate live on CNN.

 

[aleCcowaN]

Today's list of battleground states at RealClearPolitics matches my list (they added Maine to theirs, I had dropped Texas from mine)

Recent polls in battleground states (don't expect new polls for a couple of days until the debate causes a statistical effect)

Pennsylvania: Clinton 50%, Trump 47% / Clinton 44%, Trump 41%
Colorado: Clinton 49%, Trump 47% / Trump 41%, Clinton 37% / Clinton 40%, Trump 39%
Ohio: Trump 43%, Clinton 42%
Virginia: Clinton 45%, Trump 37% / Clinton 48%, Trump 38%
Maine (whole state): Clinton 40%, Trump 37% (CD1 sure for Clinton, CD2 sure for Trump)

 

[aleCcowaN]

(snip)
IMO Trump as President is almost inevitable now

(snip)

Ah. A realist.

Neither of you in that particular matter. One for reading that threefourteenwhatever site and panic, and the other one for calling "realism" any confirmation of his wishful thinking

 

[Ladewig]

On the polls only forecast Trump is up to 48.5% at fivethirtyeight.com

http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

IMO Trump as President is almost inevitable now (

You're not even going to wait 24 hours to see the results of the first debate before using the words almost inevitable?

The UK betting markets have Cinton around 1:2 and Trump around 3:2.

If you bet $20 on each your winnings from Clinton would be $10 and your winnings from Trump would be $30. That is a very big difference.


ETA

Plus, your link now shows 54/46. Don't put too much faith in numbers that change every few hours.

 

[Dr. Keith]

You're not even going to wait 24 hours to see the results of the first debate before using the words almost inevitable?

The UK betting markets have Cinton around 1:2 and Trump around 3:2.

If you bet $20 on each your winnings from Clinton would be $10 and your winnings from Trump would be $30. That is a very big difference.


ETA

Plus, your link now shows 54/46. Don't put too much faith in numbers that change every few hours.

Now may be the best time to bet on Hillary if you have some spare coin.

 

[aleCcowaN]

If I were to bet 10 quids on the U.S. having a bad president for the next four years, how much would I win?

 

[Childlike Empress]

IMO nothing Hillary or the Democratic Party can do will change a majority of the US electorate's mind that Hillary is untrustworthy and crooked and that Trump is a first class businessman with a plan for the US economy - despite all the evidence to the contrary on both counts.

That's not what the polls show. Killary and Teh Donald are both widely despised.

I am as disappointed about the US electorate as I was about the UK electorate about Brexit but it's the same thing....

- Wanting anything different
- Believing the lies
- Not believing experts

It's the same thing indeed. Everybody who pretends that "the West" is in good shape a quarter of a century after it "won" the battle of ideologies, can not win if it is one man one vote, and likely not even under the ridiculous US state-by-state winner-takes-it-all system. Because it just isn't true. Which you know if you are not in at least the upper half of society AND not putting trust in "experts" of bought and paid for media outlets.

If Trump has good advisers, tonight will be about how everybody lost from the time her husband came into office, laying the groundwork for banking deregulation, prison-industrial complex, "Arab spring" and so much more.

Anyway, I have fast food and alcohol and energy drinks (for me it will be 3am) handy to watch that circus maximus show live as well.

 

[Dr. Keith]

If I were to bet 10 quids on the U.S. having a bad president for the next four years, how much would I win?

Nada.

(That means nothing in English, just to make sure we are clear. Don't want one of those Nova no va moments.)

 

[Brainster]

Neither of you in that particular matter. One for reading that threefourteenwhatever site and panic, and the other one for calling "realism" any confirmation of his wishful thinking

FiveThirtyEight is a highly respected website that does sophisticated analysis of polling data. In 2008, the site's owner, Nate Silver, correctly predicted the presidential results for 49 of 50 states. In 2012, he rang the bell, correctly picking 50 for 50.

Silver at one point had Hillary with an 80% likelihood of winning the presidency; the fact that he now has her hovering around 50% is certainly relevant to the topic at hand of whether Trump can win.

 

[aleCcowaN]

Nada.

(That means nothing in English, just to make sure we are clear. Don't want one of those Nova no va moments.)

Thanks. Yes, I got it. "No problemo" as some say there. Unlike "no problemo" which is pig Spanish (I'm not saying not to use it, we love it), "nada" means exactly "nothing".

 

[aleCcowaN]

FiveThirtyEight is a highly respected website that does sophisticated analysis of polling data. In 2008, the site's owner, Nate Silver, correctly predicted the presidential results for 49 of 50 states. In 2012, he rang the bell, correctly picking 50 for 50.

No **** if he keeps changing the numbers according how the polls change on a daily basis. Did he already predicted the winner? I did and stood in my prediction, for systemic reasons (statistics is just part of its hardware). I also said what could make my prediction fail.

In that site there are many signs that render Silver's systems on plain sight. Go to this page and scroll down until you find the bar figures -very thinly sliced as they are 538 electors- showing Clinton's and Trump's electoral votes. That's the origin of the percentages everybody comments. In Silver's montecarlos there is about a 0.001% chance of Trump getting 530 or more electoral votes and Clinton getting 8 or less. That shows that particular montecarlo using a generator of random numbers and little more, and not logical criteria.

That site is a no-no. It's not because it's so wrong but because it shows that tomorrow the sun will go out and there'll be daylight. Great prediction. I said many times and I repeat it: I don't like sites which suckle the tit of the central limit theorem. If you know what a montecarlo is, and what the central limit theorem let you evaluate, you'll understand. If not, there's a respected media professional who got it before against an unknown gaucho who speaks weird things in a broken English. In that case, we should not continue this part of the conversation because our understandings on this issue don't overlap and there's no common terrain.

EDIT:

And I forgot to add, Silver's montecarlo predicts a 0.6% chance that Trump and Clinton would get both 269 electoral votes each (he's not saying it because it calls his bluff). Go to the "draw your own map" part of CNN predictions and start to explore how to get this particular stalemate. Of course, California is for Clinton as Oklahoma is for Trump, but there are some 22-25 states and congressional districts (both of Maine, one of three in Nebraska) you could play with. Tell me if you can reach that particular combination in any way.

 

[Brainster]

No **** if he keeps changing the numbers according how the polls change on a daily basis. Did he already predicted the winner? I did and stood in my prediction, for systemic reasons (statistics is just part of its hardware). I also said what could make my prediction fail.

Of course he updates continuously. He's not a pundit sticking a wet finger upwards to see which way the wind's blowing, he's a stats guy. As new data come in he crunches it.

In that site there are many signs that render Silver's systems on plain sight. Go to this page and scroll down until you find the bar figures -very thinly sliced as they are 538 electors- showing Clinton's and Trump's electoral votes. That's the origin of the percentages everybody comments. In Silver's montecarlos there is about a 0.001% chance of Trump getting 530 or more electoral votes and Clinton getting 8 or less. That shows that particular montecarlo using a generator of random numbers and little more, and not logical criteria.

I assume that link is something you happened to have in your clipboard; it certainly has nothing to do with Silver's site. Anything could happen; if Silver's wrong about Trump having a 1 in 100,000 chance of getting 530+ EVs, it's probably that he underestimates Trump's chances there. While nobody yet has gotten 530, two presidents in the last 50 years (Nixon with 520 in 1972 and Reagan with 525 in 1984) certainly came came close.

That site is a no-no. It's not because it's so wrong but because it shows that tomorrow the sun will go out and there'll be daylight. Great prediction. I said many times and I repeat it: I don't like sites which suckle the tit of the central limit theorem. If you know what a montecarlo is, and what the central limit theorem let you evaluate, you'll understand. If not, there's a respected media professional who got it before against an unknown gaucho who speaks weird things in a broken English. In that case, we should not continue this part of the conversation because our understandings on this issue don't overlap and there's no common terrain.

I know what a Monte Carlo simulation is, and it works very well in assessing probabilities. I note that for example, Silver gives Trump only a 0.7% chance of winning California even though Hillary is up around 20 percentage points there. That seems pretty reasonable to me, so it's not as if he simply rolled (say) a 10-sided dice and assigned the Golden State to Hillary if it came up 1-6.

And I forgot to add, Silver's montecarlo predicts a 0.6% chance that Trump and Clinton would get both 269 electoral votes each (he's not saying it because it calls his bluff). Go to the "draw your own map" part of CNN predictions and start to explore how to get this particular stalemate. Of course, California is for Clinton as Oklahoma is for Trump, but there are some 22-25 states and congressional districts (both of Maine, one of three in Nebraska) you could play with. Tell me if you can reach that particular combination in any way.

Do you really think Nate Silver is unaware of how the states allocate electoral college votes? Here's a site which gives 97 different potential 269-269 tie possibilities, after allocating California and other safe Democrat states to Hillary and Oklahoma and other solid GOP states to Trump.

In 2000 it could have happened with a vote swing of 5,381 votes--switch Florida to Gore, and Iowa, New Mexico and Wisconsin to Bush and Bush ends up 269-268 (IIRC there was one faithless elector that year).

 

[aleCcowaN]

I assume that link is something you happened to have in your clipboard;

Sorry all of you.

This one: http://projects.fivethirtyeight.com/2016-election-forecast/?ex_cid=rrpromo

 

[Ladewig]

Now may be the best time to bet on Hillary if you have some spare coin.

I made a small bet on Clinton a few weeks ago. I also made a bet that Trump would show for all 3 debates. I think I'll win the first and lost the second. Then I'll be down about $10 (including the vig).

 

[aleCcowaN]

@Brainster
I don't mean any disrespect but I think that, without knowing it, you have provided another shovelful of soil to bury Silver.

Do you really think Nate Silver is unaware of how the states allocate electoral college votes? Here's a site which gives 97 different potential 269-269 tie possibilities, after allocating California and other safe Democrat states to Hillary and Oklahoma and other solid GOP states to Trump.

There are many more involving battleground states and not so likely battleground states. They are about a thousand. Do you know how many ways you have to select 27 from 48 states, 1 federal district and 5 congressional districts? 1,946,939,425,648,112. But there are several winner-takes-it-all districts which get 3 electors: Montana, Alaska, Wyoming, the Dakotas, DC and Vermont. There are 70 ways to allocate those in a way both candidates get the same result for the Electoral College. That means there are high bars in the kind of figure Silver show near the bottom of that page. And there are! but they are not the real thing.

By the way, at the moment of writing this paragraph, Silver's bar graph shows a 0.45% probability for a tie, not 0.6%. This value varies wildly, that's why there are so many "interesting" calculations in that page ("map exactly the same as in 2012") but not "tied in 269 votes" which calls unnecessary attention on Silver's methods.

Anything could happen; if Silver's wrong about Trump having a 1 in 100,000 chance of getting 530+ EVs, it's probably that he underestimates Trump's chances there. While nobody yet has gotten 530, two presidents in the last 50 years (Nixon with 520 in 1972 and Reagan with 525 in 1984) certainly came came close.

He only underestimates his readers, the same way both candidates underestimate their voters. I can't understand why this has to become a matter of a country of +320 million people with many living Nobel laureates and just those two candidates to hold the nuclear codes of the country having 60% of the military power of the whole planet, and Silver happens to be regarded as close to the quality of those laureates and far from the quality of those candidates. This is not baseball. Everybody knows what a home run is and who managed to get it because they're outstanding players. Silver's is alchemy and silver tongue. You don't really know what's behind his "home runs".

But anyway: there is just one way to get 536 electors for Trump and 2 electors for Clinton. This way implies Trump winning DC and Clinton winning 2 of 3 Nebraskan electoral districts but losing the whole state. But this result seems to appear now and them -remember Silver's concoction changes every few hours-. Of course, Trump getting 535 and Clinton 3 is far more probable as Clinton winning just DC is not so far-fetched as the previous option.

What I am saying is that Silver's methods seem to show some kind of smoothing, possibly to convey more credibility. But that's not the only flag there...

I note that for example, Silver gives Trump only a 0.7% chance of winning California even though Hillary is up around 20 percentage points there. That seems pretty reasonable to me, so it's not as if he simply rolled (say) a 10-sided dice and assigned the Golden State to Hillary if it came up 1-6.

That's when you have to stop trying to figure out what I've done and how I come to my conclusions. Your conclusion seems totally unreasonable. If the base is polls that asked "if the election were today, who of these candidates would you vote?" as most do, there's no reason for a standard deviation of some 8% in California which allows that 0,7% you quote. Trump's chances in California yesterday were very near to nada.

For me this shows that Silver's algorithms include some way to deal with polls including Johnson and Stein and polls excluding them -and all in the middle, like offering "other candidate" as an option, or avoiding that yet registering "other" as a reply, or even avoiding to record those kind of answers (polls telling Clinton 50%, Trump 47% do that -3% replied no idea or hanged up angrily the phone-)-. They also include ways to handle polls asking registered voters and likely voters, and the multiple methods to define a likely voter.

Lots of tweaks like that drive standard deviations to the roof, and the statistical prediction becomes just a complicated way to say "but we don't really know". Of course, if Silver's gets 0.7% chances of Trump winning California, that doesn't make sense but that doesn't affect the overall estimation much, does it?

I know what a Monte Carlo simulation is, and it works very well in assessing probabilities.

I will take the text after the colon to be more self-affirmation than actual experience with it.

I inquired you about both montecarlos and the central limit theorem. The second one is fundamental to assess this question and it's almost impossible to grasp it using Wikipedia, so I eventually use it to flag posters. This was not the case here because your post is articulate, written in good faith and deals with quite the pertinent questions. But anyway, you should have to have some deal of knowledge and experience in this specific field to understand why I'm positive Silver's site is to polling and prediction as Urban Dictionary is to lexicography and education.

 

[aleCcowaN]

This figure is a good example of a visual induced inference

I has the typical heading image people like. 55.5% versus 44.5%, the kind of information loved by people who heavily bet on horses without knowing if a horse is a ruminant or not.

Below, what anchors that heading telling visually that the method behind is serious and thoroughly. Because, you know, statistics is a matter of "eyes".