Mid-September Election Update

 EDIT: shortly after this was posted a slew of new data came out invalidating much of what I posted. This will be updated to reflect the new data. My apologies for posting on a Monday when all the new data comes in.

 So, let's get right into it.

Presidential Election

2 way race: Biden + 7.0%

4 way race: Biden +5.8%

Biden does seem to lose around 1.5% when third parties are introduced. He is still pretty far ahead. Unfortunately, most battleground polls are still 2 way, so I can't see the impact on a state level. 

States	Winner/Margin	Z score	% Biden Win	% Trump Win	Electoral Votes if Biden Wins	Electoral Votes if Trump Wins
Minnesota	Biden +8.8%	-2.20	99%	1%	222	326
Wisconsin	Biden +6.8%	-1.70	96%	4%	232	316
Nevada	Biden +6.0%	-1.50	93%	7%	238	306
New Hampshire	Biden +5.5%	-1.38	92%	8%	242	300
Arizona	Biden +4.7%	-1.18	88%	12%	253	296
Pennsylvania	Biden +4.3%	-1.08	86%	14%	273	285
Michigan	Biden +4.2%	-1.05	85%	15%	289	265
Ohio	Biden +2.4%	-0.60	73%	27%	307	249
Florida	Biden +1.6%	-0.40	66%	34%	336	231
Maine CD2	Biden +1.0%	-0.25	60%	40%	337	202
North Carolina	Biden +0.7%	-0.18	57%	43%	352	201
Nebraska CD2	Unknown (0.0%)	0.00	50%	50%	353	186
Georgia	Trump +1.3%	+0.33	37%	63%	369	185
Iowa	Trump 1.7%	+0.43	33%	67%	375	169
Texas	Trump +3.5%	+0.88	19%	81%	413	163
Missouri	Trump +8.0%	+2.00	2%	98%	423	125

Right now keystone state is the key to the election. Assuming polling is accurate and performance across states is going to be correlated, Trump needs to gain 4.3 points to win the election. That gives Biden an 86% chance and Trump a 14% chance.

If I assume third parties take away 1.2 points, that brings the margins down to 3.0, which would still mean that Biden has a 77% chance of victory, with Trump having a 23% chance.

 I found some shifts interesting but some unsurprising. The rust belt rebounded after a dip that gave me 2016 Vietnam war flashbacks, although Pennsylvania still seems to be trending a bit red compared to where it was. Florida dropped almost to a toss up, which is surprising given it was up 5 points not long ago. Arizona on the other hand, which is normally a red leaning swing state, is now very blue. North Carolina is still being the least predictable state other than Florida. 

Now, to recap, as this might be important to readers. I'm assuming the electoral outcomes exist on a normal bell curve, and I'm assuming a margin of error of 4 points in reading the polling results. it varies in practice, but I feel like this is the "safe" assumption to take as most margin of errors in polls vary between 2 and 5 points. If I went with 3, I could be assuming states are more safe than they are, and I'd rather take the side of caution and lowball various states' chances if I have to err one way or another. That gives us an 8 point margin being accurate within about 96%. 68% of probable outcomes will be within 4 points, and 96% within 8. The reason, I give, for example, New Hampshire, with an exact 8 point margin, 98%, is because if Biden is actually up, say, 17%, being 9 points off in that direction, that doesn't change the outcome, Biden wins. For Trump to win, the polling error would have to swing 9 points toward him, causing him to win by 1%. There's only a 2% chance of doing that. 

Now, we saw massive shifts in 2016 toward Trump. Wisconsin was up 6.5% in the polls and it went toward Trump. There was only a 6% chance of that. But then we saw outcomes like that all over the rust belt and midwest. It's possible the polls were wrong, or more likely, in my opinion, the polls just didn't detect how fast public opinion shifted leading up to the election. So just because I'm saying that Biden is winning states by 80%, 90%, 95%, etc, doesn't mean that a similar shift can't happen. It's just going to be unlikely. 

I just wanted to remind people of that since people might be wondering how I figure this out. 

Now, onto the congressional races:

Senate

There's 46 safe seats on both sides, so the battle is over 8 particular seats. Each side needs 4 for a tie or 5 to win control of the senate

States	Winner/Margin	SD	% DEM Win	% GOP Win	Senate seats if DEMs win	Senate seats if GOP wins
Minnesota	Smith (D) +7.6%	-1.90	97%	3%	47	54
Maine	Gideon (D) +4.5%	-1.13	87%	13%	48	53
North Carolina	Cunningham (D) +3.6%	-0.90	82%	18%	49	52
Michigan	Peters (D) +3.5%	-0.88	81%	19%	50	51
Iowa	Greenfield (D) +0.3%	-0.08	53%	47%	51	50
Montana	Daines (R) +2.0%	+0.50	31%	69%	52	49
Georgia	Perdue (R) +4.3%	+1.08	14%	86%	53	48
South Carolina	Graham (R) +8.0%	+2.00	2%	98%	54	47

The senate looks like it favors democrats. Four seats lean heavily democratic, three lean republican, and one is a toss up. Ultimately, it looks like the GOP only has a 19% shot to actually control the senate outright. The democrats on the other hand, have a 53% chance to control it outright if they win in Iowa. That means there's roughly a 28% chance of a split senate, which puts actual effective control up to who wins the presidency. 

House

I'm going to be honest, there's little to no polling data on specific house seats, this is going to be a really crappy barebones forecast, but I'll lay out the house this way:

The democrats have 214 safe seats, and the GOP has 190. There are 31 toss ups. A party needs 218 seats to control the house. That said the democrats only need 4 seats to win, and the GOP needs 28. 

I'm just going to call it for the democrats here. For reference, the democrats only need to win 13% of the seats up for grabs, and the GOP needs to win 90%. Unless a miracle or statistical anomaly happens, the chance of democratic control is very high.

Anyway, I hope you enjoyed this forecast. I'll likely do another one in a few more weeks.