I just spent my weekend building electoral simulators...(ways Harris can win)

 So, now that I've gone full "excel" with my election charts, I keep trying to build on them to improve them. I even toyed with adding a simulator that would allow me to simulate election outcomes. First, I built a simple simulator which allowed me to do a random trial one at a time, and it was kind of effective. However, despite Harris having a 23% chance of winning the presidency, I tried to add a modifer to simulate the "wave" function of elections and how over/underperformance is systemic. While this did produce a number of Harris outcomes that I would expect to be more in line with the actual probabilities, it didn't actually give me USEFUL data. Once I started analyzing the trends I started realizing it was simulating extreme outcomes happening far too often. Like, it would make Florida or Maine, which have a 1% chance of flipping, flip on a magnitude of 10-15%. So that didn't work. 

I also tried to make a simulation spit out hundreds or even thousands of models like 538 did, but the data was unwieldy, hard to read, and it gave my spreadsheet a heart attack where it just freaked out and kept freezing from too much data. So...that's out. 

As such, I'm going to go back to implementing at best a simple simulator for now.

Essentially, it runs random numbers, which are then compared to the probabilities in my original chart, and if the probability threshold is exceeded, then the states flip. Basically, it lets me run electoral college trials. I also added a senate version. 

Now, again, in terms of sheer probability, i do not trust the outcomes. Each state is treated as a separate trial and the odds of a state flipping is separate for the others, so it systemically underrepresents the underdog. Like, for the record, I'd say I'm getting 95% Trump wins, 4% Harris wins. And 1% Ties. 

So, with all that said, what does this data tell me?

Well, it tells me that Harris's electoral path is relatively narrow, but not TOO narrow. There are multiple viable paths to the presidency, and that said, I'd like to discuss them here.

First, we have the vanilla "by the chart" path:

Ya know. NE2, WI, MI, PA. Boom. Straight path to 270.

Now, this path is kind of precarious. We have NO room for error. If we lose even one electoral vote, it'll throw the system into a tie and a possible constitutional crisis. 

And that CAN happen. It does appear maybe 1% of the time in my attempted simulations. 

Yeah. This is that weird nightmare scenario where we grasp defeat out of the jaws of victory by losing Omaha, Nebraska. It can happen. Something to worry about.

Now, that said, there are alternate paths to the white house. Georgia does come up relatively often in them, either by gaining Georgia and getting a straight 286 electoral votes, or maybe losing Michigan and NE2 and getting a straight 270 that way. It does happen. A few examples of that:

So yeah if the rust belt falls or even Omaha falls, we need Georgia. 

It's also slightly possible to pick up Arizona, although I haven't seen many funky electoral maps including that one. Here's a hypothetical I came up with in case we lose Wisconsin and NE2 and win with Arizona though. It can happen. Probability is weird like that.

Yeah, these are the scenarios randomness can come up with.

Now, I doubt many of these are probable. I mean, my impression is that the rust belt is probably gonna come as a trio, with the tipping point for PA not far beyond Wisconsin and Michigan. However, the rust belt is weird these days. Polling is inconsistent and while WI is normally the most secure for democrats, it really doesn't show up as such in reality, and weird scenarios like above are possible. it is very well possible we lose one state and need to pick up another in the south.

Still, I really do think we should be investing in the rust belt first and foremost. We can invest in the sun belt too, but I do see it as the harder path probabilistically, and my normal election thinking is on the wavelength that the sun belt is just an entirely different animal, and while the rust belt states are likely to go the same way, the sun belt states aren't particularly likely to follow the rust belt. Still, we should keep the 2020 electoral map in mind and know that it is possible to flip Georgia and Arizona. We've done it before. Maybe we can again. Still, I do think that's the secondary priority. 

Right now, Nevada, North Carolina, and Florida are off the table. Switching to Harris made them beyond that 8 point threshold where stuff just doesnt flip very often naturally. That's one of the consequences of switching to Harris so far.

Speaking of which, I did run the same kind of simulator on Biden, and it's weird. Biden had more points of attack in the few scenarios where he did good. he could flip NC/AZ/NV/FL more often. However, he also put MN/VA/ME/NH in risk. Harris locked down the states on our side of the map, but she's also locked out of some sun belt states. This leads to a more rigid map with fewer paths to victory, but it does greatly improve her probability of winning that one most likely path. As such, I tend to favor Harris over Biden. I mean, at least now we're not losing Virginia and a lot of states that were in play that could work against Biden are now locked down for Harris. As I said, it's a more rigid map where there's fewer paths to victory. This is worse if you believe election outcomes are truly random with each state being separate than another. I mean, in the simulator, Biden could pull off some zany wins, but he also lost due to a lot of states that wouldnt be in play for Harris flipping to Trump way too much. 

So yeah. That's where we're at.

Also, another weird map I did want to bring up that I saw happen. 

This one is a long shot, I don't see this actually happening, but the fact that it popped up once was another artifact of randomness. Here we lost NM, WI, and NE2, only to randomly pick up GA and get 270 THAT way. It CAN happen.

But yeah. I'm gonna keep fooling around with these simulators. I doubt I'll integrate them into my forecasts in any SIGNIFICANT way at this point, but I did want to discuss them since I did the random trial thing before and now I built a machine that does it automatically at a push of a button.