Simulating 2024 based on my recent election predictions

 So back in 2020, I did this cool exercise where I took the data and simulated the 2020 election based on the probabilities I had acquired. It wasnt a perfect exercise. Given how biased the polling results were, virtually every result was a Biden win, but it still did an okay job at giving a reasonable statistical range of results. It seemed to be a bit more varied in 2016 and 2012, with some 2016 results showing clinton's weakness in the rust belt costing her the election. So I do wanna see some idea of where this goes with 2024. As such, I'm going to repeat the strategy based on my last prediction. Basically, I'll take this map, with the safe states and use a random number generator in order to predict whether the result goes Biden or Trump. And then I'll post like 20 of these and that should give us a good idea of the range of outcomes. 

Without further ado.

Trial 1: 276-262 Biden

With the first go, we already are getting a Biden win. So Biden can still win. Especially if the number gods favor a sweep in the rust belt. 

Trial 2: 287-251 Trump

 And just like that, we're back to Trump winning. The joys of probability.

Trial 3: 272-266 Biden

And we're back to flipping for Biden. I mean, it can happen. The states are competitive enough that it can easily go either way if randomness were applied to the results.

Trial 4: 300-238 Trump

Here the sun belt in the SW goes Biden and the rest of the swing states practically go Trump. Another possible outcome in my opinion. Nothing presented is out of line with probability in my opinion. 

Trial 5: 278-260 Trump

A more narrow outcome, but also nothing particularly unexpected.

Trial 6: 306-232 Trump

This time we literally got the 2016 results/the results I predicted this time on the nose. 

Trial 7: 306-232 Trump

Another 306-232 one but with different states. Interesting how these combos can work out.

Trial 8: 296-242 Trump

Not sure why I keep getting AZ blue but it seems to be happening every prediction. And no it's not a methodology problem, it's random probability. I'm literally looking at my spreadsheet with a random number generator open like "okay if it goes 23 and below it's Biden, 24 and above it's trump" and it keeps going like 11 and stuff on the AZ one. It's weird but it keeps happening. Otherwise nothing out of the ordinary. 

Trial 9: 295-243 Trump

AZ went blue again. it's like a meme at this point.

Trial 10: 317-221 Trump

The most pro trump result yet

Trial 11: 272-266 Biden

Finally another Biden one!

Trial 12: 283-255 Trump

Yeah Biden only seems to win when the odds just go more in his direction than they should. Most of these are going trump. 

Trial 13: 312-226 Trump

 Trial 14: 295-243 Trump

 Randomly blue Arizona strikes again.

Trial 15: 316-222 Trump

Trial 16: 292-246 Trump

Trial 17: 287-251 Trump

Not quite the same result as #2, but close.

Trial 18: 282-256 Trump

This was probably the weirdest one. Trump took Colorado but Biden won Texas somehow, but still lost because the rust belt went red. The crazy stuff that happens with statistics.

I don't think that this kind of model is necessarily accurate as far as results go. Here I treated each state as an independent trial, whereas as we know maps tend to go in trends. Maybe the trend doesnt follow perfectly and there is random variance, but I highly doubt something like this would EVER happen. If texas goes, it's likely that GA, NC, and AZ would all go blue overwhelmingly. And Florida would arguably be "in play" by that point. 

Still, as a statistical thing, interesting outcome.

Trial 19: 275-263 Biden

YAY ANOTHER BIDEN FINALLY!

They seem rare (roughly 20% by this model) but they seem to happen on occasion. 

Trial 20: 272-266 Biden

Another weird Biden win. It can happen if Trump is unlucky on those rust belt states.

So out of the 20 planned trials, we have 5 Biden, and 15 Trump. About 25% Biden, 75% Trump. Not all that different than the 70-30 odds. I would've expected 6 Biden wins and 14 Trump wins if probability worked perfectly. But it almost never does (outside of trial 6), so yeah. Generally speaking, Trump wins can be narrow, or they can be substantial. His ceiling was 317 electoral votes. However, it is possible for Trump to lose. Especially if he suffers bad luck on the rolls so to speak. Biden was able to pull off some wins, although his wins were all narrow, with his best one only being 276 electoral votes. Still, a win is a win, and I'd be happy if he won by the skin of his teeth with 270 if it came down to it. 

As I said, these results are weird. I do think in reality trends happen, if one state goes a certain way, other similar states likely will to, and this doesnt account for that. So it isn't perfect. But still, these kinds of models do kind of show the weird ways that probability can swing elections. I know I've had some people saying "polls aren't everything", but in a sense they kinda are. The thing is, they're not 100% accurate. They have margins of error. And in my model, I use a 4% margin of error to estimate probabilities and this is what that looks like in practice. 

Still, I'm not done. I feel like doing more trials so, let's keep going.

Trial 21: 290-248 Trump

Trial 22: 287-251 Trump

Trial 23: 271-267 Trump

Narrow, but a win is a win.

Trial 24: 307-231 Trump

Trial 25: 327-211- Biden

The very rare "Biden runs the board" scenario. It can happen. It's not very likely. But it can. 

Trial 26: 276-262 Biden

A more modest Biden win. But a win is a win.

Trial 27: 319-219 Trump

Trump wins everything he's expected to and manages to win virginia. Fun fact. The number that came up for Colorado is 84. Biden has an 84% chance of taking Colorado. Meaning he JUST held onto it in this hypothetical model. Trump almost swept here.

Trial 28: 313-225 Trump

This one largely followed the "wave", but it stopped at Minnesota. This could happen if Trump overperforms a bit in 2024.

Trial 29: 275-263 Trump

Trial 30: 297-241 Trump

So at 30 trials we got 23 Trump and 7 Biden. Sticking with the rough 25% success rate for Biden here. 

And I still have some time so I'm gonna keep going.

Trial 31: 298-240 Trump

Trial 32: 290-248 Trump

Trial 33: 286-252 Trump

Trial 34: 312-266 Trump

I feel like I've seen some of these already. You know when you're starting to get enough samples when this happens. I'll stop at 40.

Trial 35: 291-247 Trump

Getting a bit groundhog day at this point

Trial 36: 274-264 Biden

Well that's a Biden win i didn't see before...

Trial 37: 288-250 Trump

Trial 38: 282-256 Biden

Trial 39: 296-242 Trump

Havent seen this specific one before. Kinda weird. 

Trial 40: 272-266 Biden

And yeah, I'm calling it here. At this point, we got 30 Trump wins and 10 Biden wins. So exactly a 25% chance of Biden to win, and 75% chance Trump will win. This is close to his statistical odds according to the trend model, and it seems about right.

Basically Trump is favored, but if Biden can flip a lot of 70-30 type situations in his favor, he can walk away with a win. Generally in this model the ones that exceed an 80% chance in one direction don't flip much. Not that it can't happen, but given this it seems relatively rare. Texas did flip twice though. So it can happen. But yeah, all in all, most of Biden's wins are narrow. he ran the board once, but most of the time he barely wins here. Trump is the definite favorite, with him running the board far more often than Biden. Again, that's gonna happen. I mean, trial 6 is basically what the model predicts is the most likely outcome. 

And yeah. That's where we're at right now. Biden isn't SCREWED, but his path to victory is narrow. Trump's path is wide and he's the clear frontrunner right now.