538 Model Megathread

[Maxwell]

yeeeeeep - which is why people keep talking up these states makes no sense.

[Nym90]

Based on polling, Clinton's odds in Alaska should probably be significantly higher, but remember that Alaska polling is awful.

[Illiniwek]

It was very easy for people like Joe Scarborough to hit the panic button and freak out this past week. When Hillary solidly wins they will say "hey she won the ground game and came through in the end," rather than "ooops we were wrong to freak out."

[Meclazine for Israel]

Things have turned decidedly bad for Trump on 5-38.

She is encroaching 300 EC votes.

[riceowl]

Welp, on polls only, he led exactly 1 day the entire cycle. Bless.

[elcorazon]

It was very easy for people like Joe Scarborough to hit the panic button and freak out this past week. When Hillary solidly wins they will say "hey she won the ground game and came through in the end," rather than "ooops we were wrong to freak out."

I don't think Joe Scarborough freaks out at a possible Trump victory. I suspect he's more worried about Hillary winning. Guy makes my skin crawl

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

He has ME-2 back over to Clinton, barely.

[ProudModerate2]

Is it just me, or have you noticed that in just one or two days, trump's chances went from 35%-ish to what is now about 28%-ish.
That is a huge drop in such a small amount of time.

[muon2]

Is it just me, or have you noticed that in just one or two days, trump's chances went from 35%-ish to what is now about 28%-ish.
That is a huge drop in such a small amount of time.

It's not so much a huge drop as it is a slight change in the slope of the projecting line. Think about a laser pointer aimed at a spot across a room 40 feet away. As the pointer is walked towards the wall a slight shift in the hand holding it can move the spot by a number of feet. Towards the end the point won't move as much because the pointer is much closer to the wall.

The trendline last week was very favorable for Trump but there was a slight shift back away over the weekend. That moves the projection point back away from last week's projection, just like a shift of the light on the wall. It's too close to the election to shift all the way back to where Clinton was a three weeks ago, but it resets the final point back about a week.

[eric82oslo]

Ugh now I'm in the awkward position of hoping Clinton loses North Carolina just to keep everyone from going OMG HE DID IT AGAIN! 50/50 STATES!!!!!!!1!

Don't worry, Clinton will win Arizona.

[riceowl]

Nate be having wet dreams about Alaska tonight

[Erich Maria Remarque]

Guys, you were so wrong...

[True Federalist (진정한 연방 주의자)]

So it looks like that of the aggregators, Nate turned out to be the least wrong. Why that was is sure to provide the fodder for a lot of articles on that site.

[emailking]

So it looks like that of the aggregators, Nate turned out to be the least wrong. Why that was is sure to provide the fodder for a lot of articles on that site.

I'm not sure how you can say it was even wrong though. Was he really supposed to have an algorithm that that produced >50% chance of a Trump victory when almost all of the polls showed a modest Clinton win? Predicting a significant chance of a Trump win, to me, seems to be about as right as he could have been.

[Admiral Kizaru]

Semi related but Sam Wang (Princeton Model) fulfils his promise to eat a bug (if Hillary didn't win).

https://www.youtube.com/watch?v=O9Xksz3i2mg

[Beet]

Semi related but Sam Wang (Princeton Model) fulfils his promise to eat a bug (if Hillary didn't win).

https://www.youtube.com/watch?v=O9Xksz3i2mg

Dude, he could have just said it was a joke like Amy Schumer about leaving the country.

[Antonio the Sixth]

Bump. If anyone's interested, here's a map of how wrong 538 was by State.

I took the polls-only forecast on Nov. 6 (before the massive herding that took place in the final two days). It had Hillary winning by 2.9 points nationally, which means it wasn't that far off in the aggregate. In individual States, however:

(Blue States are those where T***p did better, red are where Clinton outperformed. Increments are of 2 percentage points.)

[jaichind]

The 538 error map seems highly correlated to the trend map by county.

Trend:

[Antonio the Sixth]

That makes a lot of sense. I think 538 might use past election results as a baseline for how nationwide margins translate into statewide ones - thus, it tends to underestimate the extent to which States are trending away from their prior results.

The notable exceptions here are UT, ID, KS, AK, GA and NC (T***p overperformed but still trended D) and most of the Northeast and Hawaii (where the opposite occurred).

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

The lesson to be learned here is to never trust Nate Silver ever again.

[jaichind]

The lesson to be learned here is to never trust Nate Silver ever again.

Why? He came closer the pretty much all other models.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

Because I was very close to giving Florida to Trump in my final map, which would have meant getting one less state wrong. My trust in his model was the reason I kept it in Hillary's column.

[The Other Castro]

Because I was very close to giving Florida to Trump in my final map, which would have meant getting one less state wrong. My trust in his model was the reason I kept it in Hillary's column.

He used the same polls that others used, but was much more cautious than many other projectors at the time. The model was heavily dependent on polls, and if the polls aren't good then the model isn't going to be as accurate. For what he had to work with, he did relatively well.

[Slander and/or Libel]

I feel like Silver came closer to the others not because he detected something about the electorate that they didn't, but because his assumption of huge error bars made everything go closer to 50-50. He was more right, but only trivially so.

[Attorney General, LGC Speaker, and Former PPT Dwarven Dragon]

I don't think taking polls at face value is the way to predict elections anymore. In 2018 I will predict things with the expectation that whatever general trend is happening in the last couple weeks will expand significantly beyond what the polls and pundits say it will reach, which has happened in each of the last three election cycles.