Election update 10/23: Changes around the edges

So I've been watching the polls all week and things have changed a little bit, not a ton, but enough where I feel the need to do another calculation. At this point we're 11 days out from the election and based on 2016 data this is the make or break phase for Trump. Around now is where Hillary started tanking, which ultimately paved the way for a Trump win. If Trump can't pull off a similar trend in the next several days, he's screwed. And given his mediocre debate performance last night, I don't know if he can. That said, let's get into it. 

Popular vote: Biden +7.9%

4 way: Biden +8.0%

 Biden went down slightly in the popular vote, although given the third party average that was to be expected. Keep in mind there doesn't really seem to be a statistically significant difference between the two. It's just noise. One week there's a 1-2 point difference, then it disappears. 

For the chart, I'm adding Kansas and Montana now. I don't expect either of them to turn but they're getting near or within that 8 point window I look at so I might as well include them. 

Biden starts out with 216, Trump with 94. 

States	Winner/Margin	Z score	% D Win	% R Win	EV D win	EV R win
Michigan	Biden +7.8%	-1.950	97.4%	2.6%	232	322
Nebraska CD2	Biden +6.5%	-1.625	94.8%	5.2%	233	306
Minnesota	Biden +6.0%	-1.500	93.4%	6.7%	243	305
Nevada	Biden +5.2%	-1.300	90.3%	9.7%	249	295
Pennsylvania	Biden +5.1%	-1.275	89.9%	10.1%	269	289
Wisconsin	Biden +4.6%	-1.150	87.5%	12.5%	279	269
Arizona	Biden +2.4%	-0.600	72.6%	27.4%	290	259
North Carolina	Biden +1.5%	-0.375	64.7%	35.3%	305	248
Florida	Biden +1.5%	-0.375	64.7%	35.3%	334	233
Iowa	Biden +0.8%	-0.200	57.9%	42.1%	340	204
Georgia	Biden +0.8%	-0.200	57.9%	42.1%	356	198
Maine CD2	Biden +0.3%	-0.075	53.0%	47.0%	357	182
Ohio	Trump +0.6%	+0.150	44.0%	56.0%	375	181
Texas	Trump +4.0%	+1.000	15.9%	84.1%	413	163
Alaska	Trump +6.0%	+1.500	6.7%	93.4%	416	125
South Carolina	Trump +6.3%	+1.575	5.8%	94.2%	425	122
Kansas	Trump +7.0%	+1.750	4.0%	96.0%	431	113
Missouri	Trump +7.5%	+1.875	3.0%	97.0%	441	107
Montana	Trump +8.0%	+2.000	2.3%	97.7%	444	97

Oof. I'm becoming a bit concerned about Biden's chances here. The blue wall is showing significant erosion over the past few weeks. While Pennsylvania is a little stronger now, now Wisconsin is weakening as well. And most of those more "swing state" swing states? They're declining for Biden, with most being within 2 points now. The best projection I have still has the most likely outcome still at 357-181, but it won't take much to shift things.

This particular projection is weird, as it is the first I've seen a 269-269 tie explicitly mentioned. These kind of happened in previous forecasts if you swapped states around, but were distant possibilities not really worth bringing up in the model. However, this one I'm forced to address it head on. This scenario happens if Biden wins Michigan, ME2, Minnesota, Nevada, and Pennsylvania, but loses Wisconsin and everything else. The odds of it happening in this model is about 2.4%, assuming Trump overperforms by at least 4.6 points and less than 5.1 points.

What happens here? Well, apparently the house will elect the president and the senate the vice president. But the house will be doing their elections with each state voting as a bloc for a single vote. This skews it more toward the republicans than it would otherwise be and could be a total nightmare if it happens. I'd like to think of this being a distant possibility. I hope it's a distant possibility. 

Beyond that, Trump currently has a 10.1% chance of winning the election, Biden a 87.5% chance, and there's a 2.4% chance of a tie. Biden still remains strong on paper, much like Hillary did in 2016, but I'm gonna be honest, I kind of worry this might be another 2016. Biden is narrowing A LOT and remember on election day many of those very swing states swung back to Trump by like a point when they were previously pro Hillary. Then Trump swept those and laid into Hillary's blue wall. It can happen again. Just saying. 

Senate Forecast 

For the senate I'm making a few changes. Arizona is competitive again. I'm adding Kansas for consistency's sake despite iffy polling. And I'm also gonna add the Georgia special election. I previously declined to add this one given that it leaned conservative and was being chosen by ranked choice voting, but I'm including it and counting votes by party affiliation*. I'm not sure how accurate this will be, but we're trying it. It's within range. All in all, this leaves the democrats with 45 safe seats and the republicans with 44.

States	Winner/Margin	Z score	% D Win	% R Win	Seats D win	Seats R win
Minnesota	Smith +5.8%	-1.450	92.6%	7.4%	46	55
Arizona	Kelly +5.6%	-1.400	91.9%	8.1%	47	54
Michigan	Peters 5.2%	-1.300	90.3%	9.7%	48	53
Maine	Gideon +4.2%	-1.050	85.6%	14.7%	49	52
Iowa	Greenfield +1.8%	-0.450	67.4%	32.6%	50	51
North Carolina	Cunningham +1.7%	-0.425	66.5%	33.5%	51	50
Georgia 1	Perdue +0.6%	+0.150	44.0%	56.0%	52	49
South Carolina	Graham +2.3%	+0.575	28.2%	71.8%	53	48
Georgia 2	GOP +2.5%*	+0.625	26.6%	73.4%	54	47
Montana	Daines +3.3%	+0.825	20.5%	79.5%	55	46
Kansas	Marshall +4.0%	+1.000	15.9%	84.1%	56	45

Ooof. Yeah, those North Carolina and Iowa races. Ouch. This is gonna affect the odds. Even adding in the additional republican seats, it doesn't help. The democrats need to win 6 swing seats to win outright control. It has a relative lock on 4 of them but 2 are starting to get a little more iffy. And then the other 5, one is a toss up and the other 4 are lean republican. All in all, I'm still going to say that democrats have a 66.5% chance of winning the senate, republicans having a 32.6% chance of winning. A 50-50 split has a 0.9% chance of happening under this model. This is a major change from a few days ago. Democrats are weakening. This could turn into 2016 again, hold onto your butts. Regardless, a 51-49 spread with democratic control is still the most likely outcome.

Next forecast is expected Tuesday. Given we're this close to the election I want to pump these out every few days. I might wait if no major changes happen though. I won't want to make these if we don't see major changes.