Stable Boy (The Road to 270, Take 15)

With eight days to go, Joe Biden maintains a solid and substantial lead in the national polls. He leads in more than enough states to be elected president and I am more confident than ever that this will be the final result. Trump needs something to change the election, and it has not happened yet.  

But tonight I will focus on the unlikely. I will write about what a Trump victory would require and look like. 

More Americans have voted in the 2020 presidential election than voted in the 1956 presidential election. High turnout votes usually favor the Democrats and they almost always favor the candidate leading opinion polls. (The larger the vote total, the more representative the polls are.)

I. Three Ways Trump Can Win.

a. Another Systemic Polling Error.

Trump's victory caught most pollsters off guard last time because they didn't properly appreciate that the white electorate had fragmented into two halves-educated people that hated Trump and voters with less formal education, who found him appealing.  (For demographic purposes the dividing line seems to be a 4 year degree, so maybe in 2021 we should get together and form a United Caucasian College Fund. Pretty sure we could dupe a lot of Fox News Viewers into writing checks for that one.)

For Trump to win a second time, he might need another systemic polling error. Obviously, no one can predict what that would be. It is possible that turnout will be so high that it distorts the polling picture. Maybe Trump will over perform with Black voters or young voters. Maybe senior citizens will excuse his attacks against his apparent indifference to how many old people are dying from COVID. Maybe they will also excuse his repeated insinuations that Joe Biden has dementia or lacks the physical energy to run a vigorous campaign.

If this is the case, we'll probably know early. I think Ohio might be the place to watch. The vote swung from Obama by three in 2012 to Trump by eight in 2016. The polls this time suggest a dead heat. If he wins there early, it could be a sign that his coalition turned out in big numbers again. Biden doesn't need to win Ohio but he will want to hear the words "Too close to call" at some point after the polls close there next Tuesday.

b. Scraping Together 270.

Hillary Clinton won 232 electoral votes last time. Biden leads in all of her states, mostly by substantial margins. If the polls are accurate there, he will hold on to all of those states. He also looks very strong in Michigan and Wisconsin. If he wins those states, then Biden has 259 votes. 

Trump is the heavy or betting favorite in states worth 125 electoral votes. If the election is to be close, he will also win TX (38) and OH (18). Let's also give him IA (6) and Maine's 2nd District (1). 

That puts us at Biden 259, Trump 188. The remaining 91 electoral votes come from five states that Trump carried last time-FL (29), PA (20), GA (16), NC (15) and AZ (11) Biden just needs one of those states. Trump would have to sweep them.

I think that's a tall order. Pennsylvania and Arizona are likely to go for Biden. But a sweep is possible, and it's probably Trump's best chance of winning. 

c. Just One Upset.

As of tonight 538.com says that in 24% of its simulations, Trump wins at least one Clinton state. Some of those scenarios would reflect the kind of systemic polling error we already discussed. But in a really close race, one legitimate upset could put Trump over the top.

Let's back up to Biden being at 259. Instead of Trump sweeping the Big Five, he only wins four.

1. If Biden wins only AZ, Biden is at exactly 270. Trump just needs to win any Clinton state-NH, NV, MN or Maine.  Even Nebraska's 2nd district, where Biden currently polls quite well, would be enough to put Trump in the White House again.. 

2. If Biden wins NC, he is at 274 electoral votes. Trump would need to win one state, but if that state is NH, he would also need NE-2nd.

3. If Biden wins GA, he is at 275 EV. Trump needs NV or MN.

4. If Biden wins PA, he is at 279 EV.  Trump would need to win MN or NV plus NH.

5. If Biden wins FL, he is at 288 EV. Trump would need to sweep MN, NV and NH.

Each of these scenarios is unlikely and most of them can be mooted if Biden wins Iowa. Polls look good for Biden in Iowa this week but the demographics are tough for him. The state is VERY white.

II. Happier Thoughts.

The simple truth is that all of the above scenarios are unlikely. Biden's lead is stable and nothing the Trump team has trotted out to attack him has worked. Most of the states that we look at stand in just about the same place as last week, with on exception...

Florida. The polling in Florida continues to suggest a close race. Five Thirty Eight has Biden winning 66% of the time, by an average of two points. JHK has him winning 75% of the time, by an average of three and a half points. Biden leads the Real Clear Politics average by 1.8 points.

But this week, MASSIVE numbers of voters voted early. And all kinds of political prognisticators have been diving the meaning of the totals of ballots cast by registered Democrats, Republicans and unaffiliated voters. The mail voted favored the Democrats by wide margins. The in-person vote went the the Republicans by a more narrow margin. Polls suggest that the independents have been favoring Biden, but that might change on election day.

For what it's worth, the betting markets moved towards Trump as the week progressed. I still favor Biden, but it will be close.

Most of the other states have been stable this week. 

State	Elect Votes	2016 Margin	Predictit Price	538 Margin	538 %
New Hampshire	4	0.37%	77 (None)	10.2 (-0.1)	87 (None)
Nevada	6	2.42%	75 (None)	7.9 (+0.7)	91 (+3)
Minnesota	6	1.51%	76 (+1)	7.9 (-0.7)	92 (None)
Michigan	16	-0.22	69 (None)	8.4 (+0.4)	94 (+2)
Pennsylvania	20	-0.72%	62 (-6)	5.1 (-1.0)	84 (-3)
Wisconsin	10	-0.76%	66 (-2)	6.7 (None)	89 (+2)
Nebraska-2	1	-2.00%	N/a	4.7 (-0.5)	78 (-1)
Florida	29	-1.19%	46 (-6)	2.3 (-0.9)	66 (-5)
Arizona	11	-3.50%	58 (+1)	2.5 (-0.6)	66 (-3)
North Carolina	15	-3.66%	51 (None)	2.(-0.5)	65 (-3)
Georgia	16	-5.10%	42 (+1)	-0.4 (-0.3)	47 (-3)
Ohio	18	-8.07%	31 (-6)	-1 (-0.9)	42(-7)
Iowa	6	-9.41%	39 (+4)	-0.1 (-0.9)	49 (+6)
Maine-2	1	-10.29%	N/a	0.2 (-0.2)	52 (-1)
Texas	38	-8.98%	31 (+4)	-2.4 (+0.2)	33 (None)
Alaska	3	-14.73%	14 (-2)	-7.7 (-0.8)	18 (-4)
South Carolina	3	-14.27%	10 (-1)	-8.3 (-0.4)	9 (-2)
Montana	3	-20.23%	11 (+1)	-8.8 (+1.1)	12 (+1)

III. The Forecasts.

1. FiveThirtyEight. Biden is at 87 percent, down one from a week ago.

2. The Economist. Biden is at 95 percent, up from 92 percent a week ago. And their nerd is getting chesty on Twitter. He seems to think he beat Nate Silver to being aggressive here. He has Biden winning 355 Electoral Votes-that's a sweep of the Big Five, plus Georgia and Iowa.

3. Sabato Crystal Ball. Biden 290, Trump 163, Toss-up 85. Unchanged.

4. 270 to Win. Biden 290, Trump 163, Toss-up 85. Unchanged.

This is two weeks in a row that Sabato and 270 have it exactly the same.  Biden wins WI/MI/PA plus AZ. FL, GA, NC and IA are toss-ups. I'm beginning to think they may have it "right", if you think cautiously.

5. JHK Forecasts. Biden at 88.3%, Unchanged from last week. (I told you things were stable.)  They have Biden winning 352 votes. That's the Sabato 290+FL, NC and GA.  

6. Plural Vote. . Biden at 69.4%. That's down 3.7% from last week. By average outcome of each competition, their map is Biden 352, Trump 186. But their most common outcome is Biden 308, Trump 232. I think that's the Sabato 290 plus Ohio. 

That's quirky. I do not see Biden winning OH but losing both FL and NC.

7. Betting Markets. Predictit has Biden up 60%-40%. That's down 1.9 percent from last week. The international markets listed on Real Clear Politics favor Biden 64.5% to 35.5%.

8. Spider Stumbled.

No news is good news. Did I mention there was a debate last week? No one cared. But it did reduce the amount of polling for a couple days and some right-leaning pollsters like Trafalgar and Rasmussen might be crowding out the field slightly at the moment. But the people I trust mush (Harry Enten, Dave Wasserman, Nate Silver, Nate Cohn) all seem pretty confident that Biden's lead is solid. There are no warning signs in the district-level polls. 

Last week I moved Biden down from 91% to 89%.  Today I put him back at 91%. I think Biden's chances of winning are higher than ever. I think his chances of winning a blowout might be diminishing. Trump got good polls in Texas today and I think it's fair to conclude that Florida will likely be close.

I'm keeping my map unchanged from last week. Biden 351, Trump 187. As for the popular vote:

Candidate	Now	Change	10/19	10/12
Biden	52.44%	0.08	52.36	52.55
Trump	45.03%	0.01	45.02	45.19
Jorgensen	1.72%	-0.04	1.76	1.66
Hawkins	0.62%	-0.03	0.65	0.65
Other	0.19%	-0.01	-0.2	0.19

I'll be obsessively looking at the Florida data, because I can't help myself. But I like what I see, generally. And I want this to be over.

Biden will need to win one more Trump state to win office. He is currently favored in Pennsylvania, North Carolina, Arizona and Florida. 

There has been exc

A Fortnight's Sleep (The Road to 270, Take 14)

My first map of 2020.

Two weeks from tonight I will sit down at this desk and type out my final election prediction for 2020. Then I will go to bed and be hit with a wave of nervous anticipation that I have not felt since my last Santa Claus Christmas. 

For now, Joe Biden leads by nine or ten points. He has leads in the pivotal states and has a few routes to running up the score. Ohio and Georgia, two states that Biden does not need to win, are true coin-flips. 

There is only one debate remaining. Perhaps most importantly, President Trump seems to be incapable of changing his course. He lives in the Fox News Bubble and he seems determined to campaign there as well. The election is more likely to be ugly and scary than it is to be close.

But this project is a commitment, so I crunched the numbers again. There was some modest movement towards Donald Trump in some of the betting markets for individual states. But the polling data remains pretty stable. Biden's numbers improved slightly in Arizona  and North Carolina, but remained where they were last week in most other places.

 I.  The Taxonomy Down the Stretch.

Last week I introduced a new way of thinking about the election, dividing all of the remotely competitive states into six categories. I'll walk through each of those tiers here.  The chart this week includes changes from last week for the prediction market prices and the 538.com forecast. 

A. The Nightmare scenario: The States Biden Defends.

There is no sign that NH, MN or NV have become competitive in the polling data. Nate Silver has all three of these states trending towards Biden. It must be noted that the betting markets for these states have drifted slightly toward Trump.  I think that is mostly profit-taking. People sold their shares in the 80s to buy more competitive states in order to collect more on election night.  

B. The Tight Scenario: The Rust Belt Hillary Killer States.

Very little movement in these states. Trump got one good poll from a Republican pollster called Trafalgar that had him up by one.  The other polls this week had Biden winning by at least eight points.

Pennsylvania, Wisconsin and Nebraska-2nd congressional district are basically unchanged from last week.

There was some Trump movement in the prediction markets here, but less pronounced than in the long-shot states.

C. The Good Scenario: Sun Belt States.

Here is a certified quirk of October. Florida's in-person early voting began today. About 50 of the 67 counties released data on the party affiliation of ballots returned. It was just about a draw, with the Republicans netting about 4K votes. That margin was more than made up for in today's vote by mail numbers. Most importantly, Miami-Dade did not release numbers by party affiliation but will do so tomorrow.  Today, the betting markets moved to Trump by about seven points, based on this data. But I expect we will see a correction tomorrow when the Miami numbers are released. 

Arizona and North Carolina have both trended towards Biden this week, but not dramatically.

I still like Biden to run the table up to this point. That would get him up to 334 Electoral Votes.

D. The Great Scenario: The Map Expands.

Georgia and Ohio remain true toss-ups. For most of the week Nate favored Biden slightly in GA and Trump slightly in OH. But neither is more than a 52/48 proposition. We are expecting a high-quality New York Times poll of Ohio tomorrow. I expect it will be very close, and that is consistent with my overall theme-if Ohio is close, Joe Biden is winning the election.

I am starting to think of Ohio and Iowa as a 7-10 split for Biden. He probably can't win both so he might as well focus on winning one of them. With two competitive senate races, Georgia is the logical place to put more resources. To be sporting, I'm moving Georgia into Biden's column tonight. But it is a true coin toss.

Trump remains a modest favorite in Iowa.

E. The Dream Scenario: A Blue Texas

Fun little data point here. In both the Economist and FiveThirtyEight models, the most common EV total for Joe Biden is 413.  That means running the table in all of the above states plus adding Texas.

But don't get carried away. It's "the mode" but it only happens 3.8% of the time. There are a lot of combinations in a race that has 56 jurisdictions A lot has to go right for Biden to get to this point. I think it's unlikely. 

This week, Texas moved slightly to the left in the models and slightly to the right on Predictit.com. The fundamentals are this-Trump is the favorite but Biden doesn't need to win there.

F. The Blow Out Scenario.

Biden has remote shots of winning Alaska, South Carolina and Montana.  You can probably add Kansas to that list, but none of these states will decide the election and the movement this week was modestly towards Donald Trump. 

All margins and prices are for Biden/Democrat. Changes from last week are shown in parenthesis.)

State	Elect Votes	2016 Margin	Predictit Price	538 Margin	538 %
New Hampshire	4	0.37%	77 (-3)	10.3 (+1)	87 (+3)
Nevada	6	2.42%	75 (-5)	7.1 (None)	88 (+1)
Minnesota	6	1.51%	75 (-5)	8.6 (+0.2)	92 (+1)
Michigan	16	-0.22	69 (-7)	8 (-3.9)	92 (None)
Pennsylvania	20	-0.72%	68 (-2)	6.1 (-0.4)	87 (None)
Wisconsin	10	-0.76%	68 (-5)	6.7 (+0.2)	87 (+2)
Nebraska-2	1	-2.00%	N/a	5.2 (+0.4)	79 (+3)
Florida	29	-1.19%	52 (-6)	3.2 (-0.6)	71 (None)
Arizona	11	-3.50%	57 (+3)	3.1 (+0.2)	69 (+2)
North Carolina	15	-3.66%	51 (-5)	2.6 (+0.5)	68 (+5)
Georgia	16	-5.10%	41 (-2)	-0.1 (+0.5)	50 (+4)
Ohio	18	-8.07%	37 (-5)	-0.1 (-0.6)	49(-4)
Iowa	6	-9.41%	35 (-6)	-1.2 (-0.3)	43 (-2)
Maine-2	1	-10.29%	N/a	0.4 (+1.1)	52 (+5)
Texas	38	-8.98%	27 (-3)	-2.6 (+0.6)	33 (+2)
Alaska	3	-14.73%	17 (-3)	-6.9 (-0.1)	22 (-1)
South Carolina	3	-14.27%	11 (-5)	-7.9 (-1.2)	11 (-5)
Montana	3	-20.23%	10 (-3)	-9.9 (+0.4)	11 (None)

II. The Forecasts.

1. FiveThirtyEight. Biden is at 88 percent, up slightly from 87 percent a week ago. 

2. The Economist. The most aggressive forecast in the field inches up again-moving Biden up to 92% from 91% a week ago. He seems to be determined to stay one step ahead of Nate Silver so he can yell "First!" when the networks call Florida or Pennsylvania on Election Night. His tally right now is Biden 350, Trump 188, which means that like me he moved Georgia into Biden's column this week. 

3. Sabato Crystal Ball. Biden 290, Trump 163, Toss-up 85. No changes from last week. 

4. 270toWin. Biden 290, Trump 163. Toss-up 85.  No changes from last week.

I think Sabato and 270 are both being too conservative with their toss-ups. Biden leads in FL and NC. Trump leads in IA. OH and GA are the only true toss-ups at the moment.

5. JHK Forecasts. Biden 88.3%, up slightly from 87.7%  They have Biden slightly favored in Georgia and Iowa with Ohio as close to a dead-heat as possible. They give Trump the edge there, 50.1% to 49.9%. That makes 357 Biden, 181 Trump. 

They are a little bullish on Biden in Texas, with him winning there 38.8% of the time. 

6. Plural Vote. Biden has a 73.1% chance of of winning, down slightly from 74% last week.

But they have Biden 332, Trump 206, which is up from 323 to 215 last week.  And they are even more bullish on Texas, giving Biden a 44% chance of winning there.

Right now this forecast feels the shakiest. They give Biden the lowest chance of winning the election but the best chance of achieving a blowout. 

7. Betting Markets. Predictit.com gives Biden 61.9%, up slightly from last week's 61.1%. That is consistent with other betting markets, that give Biden a slightly better than 3 to 2 chance of winning.

There was state-level movement toward Trump this week but that hasn't moved the big board. It also didn't change any states in their overall forecast, which remains Biden 334, Trump 204.

8. Spider Stumbled.

I remain confident but the state-level betting markets have me slightly more concerned than I was the last two weeks. I will move Biden from 91% down to 89%.  He was at 91% two weeks in a row.

He didn't have a bad polling week, but he did have a stable polling week, after several consecutive weeks of gains. 

But I do move Georgia into his column-by an inch.  That leaves my official prediction at Biden 351, Trump 187. 

And as I started last week, I will take a crack at the popular vote totals again. 

Biden 52.36%  (-0.19)

Trump 45.02%  (+0.17)

Jorgenson (Libertarian) 1.76%  (+0.1)

Hawkins (Green Party) 0.65%

Others: 0.2%  (+0.1)

Looking at all of the date-polls, forecasts, betting markets, there is simply no reason to doubt that Joe Biden remains on a path to 270 and beyond.  We got this week and then next week. Then it will be Monday. Then it will be game day.

Keep the faith.