Bitecofer model 99.5% Biden wins

[DaleCooper]

I saw this earlier and didn't even bother reading it once I saw that 99% thing. This is worse even than the Huffington Post viral image of the arbitrary Hillary 97% chance of winning. This isn't a legitimate model, this is her prediction being presented as a model. She's clearly branding herself as a Nostradamus of elections and I'd respect her more if she just wrote an article saying, "Here's why my data indicates Biden will win" and explained why she's so confident in it.

[Donerail]

How did someone this dumb get a PhD?

Many dumb people get PhDs.

[ProudModerate2]

LOL.
I wish, but it's not that high.

[redjohn]

This is the dumbest thing I've ever read. 99.5% for an election that could clearly go either way, before any of the debates and a month-and-a-half out?

It really couldn't "clearly" go either way. Trump theoretically could still win, but his path is VERY narrow. And a month and a half almost certainly isn't enough time for things to change enough to make a difference. Not when people have been so entrenched in their opinions on this president and election for so long, and early voting has already begun in some states; I honestly doubt it matters much what happens at all from this point. 99.5 may be too high, but 90 or so? Sounds about right to me.

As for the debates, probably the only election the debates even arguably had anything close to a decisive effect on was 1960. I do not expect that to suddenly change this year. Certainly not in Trump's favor; he has never won a debate one-on-one, while Biden has never lost one.

Clearly means Trump has a very obvious path to victory so yes, it could go either way. You're describing it as narrow, which is true, but that doesn't mean he doesn't have an obvious path to victory. It wasn't so clear in 2016. Hold PA+FL+AZ, or hold PA+FL+WI.

[ProgressiveModerate]

I generally respect her, but the uncertainly in her model is way too high, and the polling errors, and the fact that polling errors tend to largely be off in one party's favor nationally. When I turned the uncertainty down in my model to "2", I got a 99.2% chance of Biden winning, but the map looks like this:

[emailking]

This is the dumbest thing I've ever read. 99.5% for an election that could clearly go either way, before any of the debates and a month-and-a-half out?

It really couldn't "clearly" go either way. Trump theoretically could still win, but his path is VERY narrow. And a month and a half almost certainly isn't enough time for things to change enough to make a difference. Not when people have been so entrenched in their opinions on this president and election for so long, and early voting has already begun in some states; I honestly doubt it matters much what happens at all from this point. 99.5 may be too high, but 90 or so? Sounds about right to me.

As for the debates, probably the only election the debates even arguably had anything close to a decisive effect on was 1960. I do not expect that to suddenly change this year. Certainly not in Trump's favor; he has never won a debate one-on-one, while Biden has never lost one.

Clearly means Trump has a very obvious path to victory so yes, it could go either way. You're describing it as narrow, which is true, but that doesn't mean he doesn't have an obvious path to victory. It wasn't so clear in 2016. Hold PA+FL+AZ, or hold PA+FL+WI.

Well, 99.5% also means it could go either way, just saying.

[ProgressiveModerate]

This is the dumbest thing I've ever read. 99.5% for an election that could clearly go either way, before any of the debates and a month-and-a-half out?

It really couldn't "clearly" go either way. Trump theoretically could still win, but his path is VERY narrow. And a month and a half almost certainly isn't enough time for things to change enough to make a difference. Not when people have been so entrenched in their opinions on this president and election for so long, and early voting has already begun in some states; I honestly doubt it matters much what happens at all from this point. 99.5 may be too high, but 90 or so? Sounds about right to me.

As for the debates, probably the only election the debates even arguably had anything close to a decisive effect on was 1960. I do not expect that to suddenly change this year. Certainly not in Trump's favor; he has never won a debate one-on-one, while Biden has never lost one.

Clearly means Trump has a very obvious path to victory so yes, it could go either way. You're describing it as narrow, which is true, but that doesn't mean he doesn't have an obvious path to victory. It wasn't so clear in 2016. Hold PA+FL+AZ, or hold PA+FL+WI.

Well, 99.5% also means it could go either way, just saying.

Just based on how much the race can shift in a month or two, 99.5% seems too high. The race only needs to tighten by 3 points, and polls be slightly off in Trump's favor for him to win, which while is less likely to happen than not, isn't an extremely tall order.

[Fmr. Gov. NickG]

I guess she has made a decent career as a pundit, but she certainly wouldn’t be respected as an scholar.  As far as I can tell, she has zero academic publications, and it really isn’t a surprise she’s never been offered a tenure-track job.

[Orser67]

Good on Niskanen for taking it down.

Even before the 99.5% thing, I thought Bitecofer had some interesting insights, but is/was waaaay too overconfident.

[tagimaucia]

Aren't her main "insights" just like... objectively wrong empirically?

[pbrower2a]

Nearly as significant as the margin is the time in which to close it, and while it is conceivable that someone running a statewide race can lose if supposedly ahead by 20% against a generic opponent if the election is a year away (it is less than 90% certainty, and I remember  stories that said in 2009 that  Blanche Lincoln was a sure thing for winning re-election the next year... ha! ha!) such certainty either disappears if things start going wrong -- like a well-funded opponent or a wave year against that pol's Party. To be sure any lead at any time is an advantage, but with time an opponent is known, patterns of campaign speeches are set, and the campaign funds start rolling in one way or the other.

Based on Nate Silver's data on winning Senate seats:

    My estimates for the results for intermediate probabilities based on near-linear differentials  for three and two months out for a Senate campaign, which I see as a good analogue for a statewide campaign for a State's electoral votes:.
 
Time to election  |1 point|5 points||10 points|20 points|
one day............. |...64%|....95%|.....99.7%|.99.999%|
one week........... |...60%|....89%|.......98%|...99.97%|
one month......... |...57%|....81%|.......95%|.....99.7%|
three months......|...55%|....72%|........87%|......98%|
six months..........|...53%|....66%|.......79%|.......93%|
one year.............|....52%|...59%|.......67%|.......81%|

(It may not be perfect, because a Senate nominee does not "triage" himself out of a race, although a Presidential campaign may be swift to 'triage" itself out of meaningful campaigning in states that it sees as sure losses and put in more appearances and advertising funds into hot races where such might make a difference. This probably understates the reality of Presidential races because of a triage effect. The Biden campaign is unlikely to spend much in Indiana, and the Trump campaign isn't going to put much effort into winning Massachusetts.   


lead  likelihood, three months and two months

0   50    50  10  87  91
1   55    56  11  88  92
2   59    64  12  89  93
3   64    69  13  90  94
4   69    73  14  91  95
5   72    77  15  92  96
6   76    78  16  93  97
7   80    80  17  94   97
8   83    84  18  95  97
9   85    87  19  96  97

I might come up with estimates for one month, except that Nate Silver has his algorithm for estimating chances of Biden and Trump and I lack such an algorithm and computing power. Check Nate Silver for his estimates. Basically if Trump is behind by 7% in any state he has to be very desperate to seek to appear there or for his campaign to go on any advertising blitz in that state. So if Biden is up by 7% in a state on October 3, then he has about a 93% chance of winning that state.

Polling has been remarkably stable, mostly because the character, agenda, and behavior of the two nominees have been stable.  What is unstable is the time remaining. Biden is playing a run-out-the-clock strategy at this point more than a blow-out-Trump strategy. If Biden wins some surprises, then such is more a consequence of failures of Trump efforts in some quixotic efforts that cause him to ignore trouble elsewhere. An analogy in American football is for a team down twelve points in the fourth quarter is that the quarterback throws a Hail Mary pass into a nickel defense that results in an interception that some defensive back runs back for a touchdown.

So let's suppose that the Trump campaign sees this on October 3: 

AZ  GA  FL   IA   MI   MN  MO  NC  NH  OH  PA  TX  WI
-5   +2  -3   +2 -9  -10   +5   0    -6   +2  -7  +3   -5

OK, so Trump is behind by 5 in Arizona, 3 in Florida, and 5 in Wisconsin. He must win all three, and to do so he might have to take some chances of losing North Carolina and some other states. Pennsylvania still seems within range, but New Hampshire won't make much of a difference. If everything goes right for Trump, then this result might happen:

 AZ  GA  FL   IA   MI   MN  MO  NC  NH  OH  PA  TX  WI
+1  +2  -3   +2 -14  -15   +3   +1  -6   +1  +1  +3  +1

... and Trump wins! He gets to "keep America Great", whatever that means.

On the other hand, what happens if something goes awry? Maybe this. Democrats make some adjustments  to thwart the effort with their own ad blitz, and Trump loses a grip on some of the states that Trump was close in. This is the result:

 AZ  GA  FL   IA   MI   MN  MO  NC  NH  OH  PA  TX  WI
-3   -2  -5    -2   -14  -15   -1   -3   -10   -2  -5  -1   -4

Well, he did better in Arizona, Pennsylvania, and Wisconsin than he might have expected, but not well enough to win. On the other hand, he also let some things slide... which explains the loss that nobody expected in Missouri.  That is about as lucky as a bare Trump win, and it is a 423-EV win or even bigger (he said some disparaging things about American Indians, which went badly in Alaska and Montana).