Election Update 10/20: Return to the mean or beginning of the decline?

So, new polls numbers released in the last week seemed to reverse a lot of the trends that Biden had going for him. He was up by double digits, and it looked like he had a commanding lead in the electoral college. Biden is still in a fairly strong position, but looking at realclearpolitics' polling data, Biden appears to be in free fall in some areas of the country, losing several points in key states vs last week. While he still maintains his electoral college lead mostly, the race is much more narrow, and he's only winning by 3-4 points in the electoral college rather than by 6-7 last week. That's not good. Regardless, probability is still on his side, but if he continues like this, like Clinton did in 2016, he might be in for a bad time. He currently is tracking similarly to Hillary in 2016....

National popular vote: Biden +8.6%

4 Way: Biden +7.6%

Biden still seems to have a commanding lead in the popular vote. It's less than last week, but it's still very significant. Regardless of what is happening in the electoral college, Biden is in a very strong position in regard to the popular vote. However, the popular vote isn't what wins elections. The electoral college does. Let's see what he's doing there.

Changes: Minnesota declined a bit to 6.3%, after being up 8-10 points for several weeks. This puts it in swing state status. On the flip side, we finally have a decent Alaska poll and I'm deciding to include it too. Much like Missouri and South Carolina I HIGHLY doubt it will flip and most keep these states outside of "swing" status because the "fundamentals" there lean heavily conservative, but my model is JUST focusing on the statistics, and if it's below 8 or so points, I think it's a good idea to include it. That said, this race appears to be tightening up with more states in the swing category overall. This brings Biden down to 216 starting electoral votes and Trump down to 103.

States	Winner/Margin	Z score	% D Win	% R Win	EV D win	EV R win
Michigan	Biden +7.3%	-1.825	96.6%	3.4%	232	322
Nebraska CD2	Biden +6.5%	-1.625	94.8%	5.2%	233	306
Minnesota	Biden +6.3%	-1.575	94.2%	5.8%	243	305
Wisconsin	Biden +6.2%	-1.550	93.9%	6.1%	253	295
Nevada	Biden +5.2%	-1.300	90.3%	9.7%	259	285
Pennsylvania	Biden +3.8%	-0.950	82.9%	17.1%	279	279
Arizona	Biden +3.1%	-0.775	78.1%	21.9%	290	259
North Carolina	Biden +2.0%	-0.500	69.1%	30.9%	305	248
Iowa	Biden +1.2%	-0.300	61.8%	38.2%	311	233
Georgia	Biden +1.2%	-0.300	61.8%	38.2%	327	227
Florida	Biden +1.0%	-0.250	59.9%	40.1%	356	211
Maine CD2	Biden +0.3%	-0.075	53.0%	47.0%	357	182
Ohio	Trump +0.2%	+0.050	52.0%	48.0%	375	181
Texas	Trump +4.4%	+1.100	13.6%	86.4%	413	163
Alaska	Trump +6.0%	+1.500	6.7%	93.3%	416	125
South Carolina	Trump +6.3%	+1.575	5.8%	94.2%	425	122
Missouri	Trump +7.5%	+1.875	3.0%	97.0%	435	113

Let's unpack this a bit. Yes, Biden still has, on paper, a commanding lead. If the election went perfectly, as the odds predict, Biden will go 357 to Trump's 181. However, it's a very SOFT lead, if that makes sense, at least compared to last week. As we know in 2016, I predicted Hillary would win 272-266. Then Trump overperformed by about 2-3 points statistically. Many of those states, such as Florida, Georgia, North Carolina, etc., are in that range. And they're big states. If Trump overperforms by 2 points, he has 248 electoral votes to Biden's 290. From there, say things get weird like in 2016 where he pulls in Pennsylvania, which is only up 3.8, and either Arizona, Nevada or another rust belt state like Wisconsin. Good night, Biden. A lot of people are acting like this is over, it's all locked up. No, we're 2 weeks from the election. Hillary took a similar path where she fell apart shortly before election day. I do believe there are mitigating factors to this outcome like mail in voting, but if things are off a little bit, the outcome could change a lot. Meanwhile I don't think Biden can really gain much more than he already has. He could win back Ohio which is currently a coin flip, but the odds of him flipping Texas is about the odds of Trump actually winning the election outright. Alaska, South Carolina, and Missouri are only included for formality's sake. 

Regardless, statistically, a Trump victory is not that high yet. Biden has a 82.9% chance of winning and Trump has a 17.1% chance. This might not seem like a lot, but Trump is almost 3x more likely to win than he was last week (17% chance vs 6% chance), so that is a fairly big swing. Still, Biden holds a huge statistical advantage. The nightmare scenario above is only the worst cast scenario. Of course that worst case scenario has happened before so I felt the need to discuss the possibility of it happening again.

Senate forecast

I'm removing the Texas race as it is no longer competitive. We're back to a 46-46 baseline.

States	Winner/Margin	Z score	% D Win	% R Win	Seats D win	Seats R win
Minnesota	Smith + 7.2%	-1.800	96.4%	3.6%	47	54
Iowa	Greenfield +4.8%	-1.200	88.5%	11.5%	48	53
Michigan	Peters +4.3%	-1.075	85.9%	14.1%	49	52
Maine	Gideon +4.2%	-1.050	85.3%	14.7%	50	51
North Carolina	Cunningham +3.2%	-0.800	78.8%	21.2%	51	50
Georgia	Perdue +0.9%	+0.225	41.1%	58.9%	52	49
South Carolina	Graham +2.3%	+0.575	28.3%	71.7%	53	48
Montana	Daines +3.3%	+0.825	20.5%	79.5%	54	47

All in all, the democrats still hold a 78.8% chance of winning the senate, while republicans hold a 14.7% chance. The chance of a tie is 6.5%. This has shifted slightly in the republicans' favor, but is very similar to last week's predictions overall. All in all, the map looks like this.

That said, there have been some interesting changes in the presidential race at least. Biden's lead has started to erode, and while this could be due to a lot of mundane reasons. Bad polling, simply regressing to the mean as Trump recovers from COVID, it is a cause for concern to me. If I did not experience 2016, I would not think much of it, but I remember Hillary followed almost the same exact trajectory. We won't know for another few days or even a week if this is, indeed, a meaningful shift, but it is cause for alarm if you want Biden to win. 

On the flip side, the senate races barely changed at all. A few races got more competitive, like Minnesota and Georgia, but on the flip side, others got less so, such as Graham's seat in South Carolina. Not a lot has fundamentally changed here. 

 That's all folks, see you next week, or in a few days if I feel like doing another one of these and the conditions justify it.