Jump to content
270soft Forum

I ran 2,001 simulations of the 2020 election using polling data. Here's what I found (with data and maps!)


Actinguy
 Share

Recommended Posts

 I went back to 270soft's Presidential election simulator -- and this time, I ran the simulation 2,001 times.

Not all in one setting, of course.  But I ran it over several days, during conference calls for sections I didn't have to actively be engaged in.  That kind of thing.

Why 2,001?  That was exactly the number of times I had to run it to remove the outlieriest of the outliers -- for example, exactly one time out of the 2,000 simulations, Biden won Alabama, South Dakota, and Utah.  Likewise, exactly one time out of the 2,000 simulations, Trump won Connecticut, New Jersey, and even Biden's own Delaware.

2,001 simulations was exactly what was needed to drop these outcomes to below 0.1%, and therefore to fall off the session statistics.  You still have some outliers left over.  For example, Biden winning Louisiana (which happened twice out of 2,001 simulations, Biden winning Mississippi, Kansas, Indiana -- and on the flip side, Trump winning New Mexico, Virginia, and Colorado.  So outliers are still present on both sides, but we've at least erased the most egregious perhaps.

Interestingly, despite having been "swing states" for most of my adult life, in 2,001 simulations, Virginia and Colorado went to Trump WAY less often than Texas and Alaska went to Biden.

Approximately 16 of the elections ended in a 269-269 tie.

Out of 2,001 simulations, Pennsylvania was the most frequent bellwether state -- PA backed the ultimate winner 88.6% of the 2,001 simulations.  This was followed by Nevada (88%) and Michigan (87.8%).

The tipping point in nearly 25% of the simulations was Florida.  Tipping point is defined as "The state that gives the election winner 270 electoral votes, when ordering from largest to smallest margin of victory."

Below this chart, I've used the data to generate several maps.
 

image.thumb.png.a4d276bd44e18146dffd7c9f54030f5f.png

For our first map, I've ONLY awarded a state to a party if they won more than 99.9% of the 2,001 simulations.  If a state swung at least twice in 2,001 simulations, then I called it a swing state.   Here we see that Biden is trouncing Trump in this measure alone, though of course not nearly enough to claim the 270 votes needed to win.
image.png.1b28db37658000fff2f35af285c5666c.png

Our next map goes to the other extreme:  If a state went to a given candidate at least 50.1% of the time (as all states did), then it is awarded to that candidate.

Our average result, over 2,001 simulations, was Biden winning 335 to 203.

image.png.1b35052a02883e76a6dcacf7eb51d1e6.png

But now let's get to the real interesting stuff.  The swing states.

How are we defining swing state?

Well, let's see what happens if we say a state could swing 5% in either direction.

That...turned Iowa and one district in Maine into swing states.  We're going to need much more than that to swing this election in Trump's favor.  Especially since Iowa was his to start with.

image.png.5705cf09c38cd863542ca91630c655e0.png

Ok.  We have to go bigger.

What if we say a state could swing TEN points in either direction?

Well, that would put Georgia, North Carolina, and Ohio in play.

But we still have the same problem as before -- we haven't helped Trump, we've HURT him.  Most of those states were previously Trump's.

image.png.7d8f35bdfd4eb39c493dbd52724f1bf5.png


Alright.  What if we say a state could swing FIFTEEN points in either direction?

Now we've bumped Arizona and one district in Nebraska.  

The good news is those were both Biden's. 

The bad news is that it STILL isn't enough.

image.png.6cfd422d20ae55595fc7c53d73e60cc2.png

Alright.  Let's shoot for the stars.  What if we say any given state could swing TWENTY points in either direction.

That swung...Texas.

Whoops.  We made it worse again.
 

Okay...TWENTY-FIVE percent.

What if any state could just completely swing in the other direction by a margin of TWENTY-FIVE percent?

That takes out Florida.  NOW we're talking!

But we're not talking enough.  Even if every single swing state below...with a 25% margin...all go to Trump, Biden STILL wins.

image.png



Jesus.  ...Is the answer THIRTY percent?

...yes, actually.  Thirty percent could do it, as that unlocks Wisconsin.

image.png.0670de5db4e7f49736f3fecf69f47627.png

To come even close to winning, Trump has to win every single swing state above (and we had to define "swing state" as "These 2,001 simulations were potentially wrong in each individual state THIRTY percent of the time -- and every single one of these errors was in Trump's favor) AND he needs to win at least either the spare district in Maine or the spare one in Nebraska.

Impossible?  No.  After all, that's almost exactly what happened in about 16% of our 2,001 simulations.

But extraordinarily unlikely.

 

 

 

 

  • Like 3
  • Upvote 1
Link to comment
Share on other sites

32 minutes ago, Hestia11 said:

The problem is that Michigan and Pennsylvania aren't locks. Especially PA. The polls out of Pennsylvania should be the #1 most worrying thing for the Democratic Party. Michigan looks good, but Pennsylvania is consistently +5 or less. 

I agree they're not locks.  The only "locks" are the states I indicated in the first map, where I only granted a state if Biden was projected to win it more than 99.9% of the 2,001 simulations run.

Biden has consistently led Trump in the polls in PA since before he won the nomination.  In Mid April his lead dropped to less than 1%, but it has since consistently widened again to about 5%.  

Obviously you always want the margin to be as wide as possible, and as a former Pennsylvanian myself I know how "wanna be Southerners" we can be at times.  It is not at all impossible for Trump to win PA.  It's not impossible for him to win any state that isn't marked in blue in the first map -- and indeed, he even won three of those blue ones in exactly one out of 2,001 simulations.  The possibility exists.  

But the simulation suggests that there's only about an 18% chance of it happening.  Not impossible!  But 18%.  Other states are closer than that.  

Link to comment
Share on other sites

3 hours ago, Actinguy said:

 I went back to 270soft's Presidential election simulator -- and this time, I ran the simulation 2,001 times.

Not all in one setting, of course.  But I ran it over several days, during conference calls for sections I didn't have to actively be engaged in.  That kind of thing.

Why 2,001?  That was exactly the number of times I had to run it to remove the outlieriest of the outliers -- for example, exactly one time out of the 2,000 simulations, Biden won Alabama, South Dakota, and Utah.  Likewise, exactly one time out of the 2,000 simulations, Trump won Connecticut, New Jersey, and even Biden's own Delaware.

2,001 simulations was exactly what was needed to drop these outcomes to below 0.1%, and therefore to fall off the session statistics.  You still have some outliers left over.  For example, Biden winning Louisiana (which happened twice out of 2,001 simulations, Biden winning Mississippi, Kansas, Indiana -- and on the flip side, Trump winning New Mexico, Virginia, and Colorado.  So outliers are still present on both sides, but we've at least erased the most egregious perhaps.

Interestingly, despite having been "swing states" for most of my adult life, in 2,001 simulations, Virginia and Colorado went to Trump WAY less often than Texas and Alaska went to Biden.

Approximately 16 of the elections ended in a 269-269 tie.

Out of 2,001 simulations, Pennsylvania was the most frequent bellwether state -- PA backed the ultimate winner 88.6% of the 2,001 simulations.  This was followed by Nevada (88%) and Michigan (87.8%).

The tipping point in nearly 25% of the simulations was Florida.  Tipping point is defined as "The state that gives the election winner 270 electoral votes, when ordering from largest to smallest margin of victory."

Below this chart, I've used the data to generate several maps.
 

image.thumb.png.a4d276bd44e18146dffd7c9f54030f5f.png

For our first map, I've ONLY awarded a state to a party if they won more than 99.9% of the 2,001 simulations.  If a state swung at least twice in 2,001 simulations, then I called it a swing state.   Here we see that Biden is trouncing Trump in this measure alone, though of course not nearly enough to claim the 270 votes needed to win.
image.png.1b28db37658000fff2f35af285c5666c.png

Our next map goes to the other extreme:  If a state went to a given candidate at least 50.1% of the time (as all states did), then it is awarded to that candidate.

Our average result, over 2,001 simulations, was Biden winning 335 to 203.

image.png.1b35052a02883e76a6dcacf7eb51d1e6.png

But now let's get to the real interesting stuff.  The swing states.

How are we defining swing state?

Well, let's see what happens if we say a state could swing 5% in either direction.

That...turned Iowa and one district in Maine into swing states.  We're going to need much more than that to swing this election in Trump's favor.  Especially since Iowa was his to start with.

image.png.5705cf09c38cd863542ca91630c655e0.png

Ok.  We have to go bigger.

What if we say a state could swing TEN points in either direction?

Well, that would put Georgia, North Carolina, and Ohio in play.

But we still have the same problem as before -- we haven't helped Trump, we've HURT him.  Most of those states were previously Trump's.

image.png.7d8f35bdfd4eb39c493dbd52724f1bf5.png


Alright.  What if we say a state could swing FIFTEEN points in either direction?

Now we've bumped Arizona and one district in Nebraska.  

The good news is those were both Biden's. 

The bad news is that it STILL isn't enough.

image.png.6cfd422d20ae55595fc7c53d73e60cc2.png

Alright.  Let's shoot for the stars.  What if we say any given state could swing TWENTY points in either direction.

That swung...Texas.

Whoops.  We made it worse again.
 

Okay...TWENTY-FIVE percent.

What if any state could just completely swing in the other direction by a margin of TWENTY-FIVE percent?

That takes out Florida.  NOW we're talking!

But we're not talking enough.  Even if every single swing state below...with a 25% margin...all go to Trump, Biden STILL wins.

image.png



Jesus.  ...Is the answer THIRTY percent?

...yes, actually.  Thirty percent could do it, as that unlocks Wisconsin.

image.png.0670de5db4e7f49736f3fecf69f47627.png

To come even close to winning, Trump has to win every single swing state above (and we had to define "swing state" as "These 2,001 simulations were potentially wrong in each individual state THIRTY percent of the time -- and every single one of these errors was in Trump's favor) AND he needs to win at least either the spare district in Maine or the spare one in Nebraska.

Impossible?  No.  After all, that's almost exactly what happened in about 16% of our 2,001 simulations.

But extraordinarily unlikely.

 

 

 

 

The average is the same as my forecast map, I think.

  • Like 1
Link to comment
Share on other sites

2 minutes ago, ThePotatoWalrus said:

That's my current thought as well, though I'd give Trump the Maine district

Oh yeah I think my forecast map does have that district for Trump. 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

 Share

×
×
  • Create New...