538 model & poll tracker thread

[Crumpets]

If I could offer a counter-point to the "538 overestimates Trump" line, consider: What % probability would you give to the Trump administration pulling something out of a hat like the Comey letter in 2016? For example, Barr arresting Hunter Biden for espionage 10 days before the election (of course, only to release him later). That uncertainty would seem to model such a scenario, even if Nate claims this doesn't model election interference. The venn diagram of interference and black swan events would seem to overlap quite a bit.

[dunceDude]

Model looks like garbage. He shouldn't have been getting in fights on twitter if it was going to look like this. Shouldn't have been fighting when he had no product out himself at all either. Methodologically fine I guess, I won't pretend I read deep into it.

He's kind of the alpha in this field so I expect a lot more frankly. Makes the same mistakes his "genre" of person—thinks themselves clever, avoids mirrors—always makes, meaning he's trying to make this look good for this particular moment when he should just try and be the fkn best. Whatever. Titans fall.

[tjstarling]

I’m not getting too worked up over the 29% chance figure since the model is attempting to forecast 80+ days from now with a great deal of uncertainty. That figure doesn’t seem too unusual. However, Silver is always talking about model overfitting and the ad hoc choices of other empirical models but so many of his choices here seem so, shall we say, ad hoc. It almost feels like he think Trump’s chances are around 30% and he’s tried to build the model to output the results he has in his head.

[S019]

This model to me has a Republican bias, all of the evidence seems to agree that Biden is up in NC and GA and up big in MI/PA, but the model doesn't really reflect that. Also the polls show TX as a pure tossup, but the model doesn't reflect that either.

[Sadader]

The reckoning has begun

https://twitter.com/gelliottmorris/status/1293593884677877761

[Penn_Quaker_Girl]

The reckoning has begun

https://twitter.com/gelliottmorris/status/1293593884677877761

Silver vs. Morris -- now THIS is something that I didn't see coming in 2020. 

[Skye]

https://twitter.com/gelliottmorris/status/1293599611249135627

Yesssss, keep this going.

[WD]

https://twitter.com/gelliottmorris/status/1293599611249135627
Yesssss, keep this going.

FIGHT! FIGHT! FIGHT!

[The Mikado]

I do think that uncertainty is especially high at the moment because we're in a polling drought. I'd expect uncertainty to shrink quite a bit if there's a big wave of polls this weekend (which I think is likely as people want to get their pre-convention snapshots to compare with post-convention).

I do really miss the density of state polls we had in the 00s. I remember in the late 2000s, SUSA literally did a "Bush approval in all 50 states" monthly tracker for a while. No pollster would dare touch something like that now. If we had more actual polls of, say, Mississippi, it'd help 538's model get to 99% for Trump and eliminate some of the uncertainty.

[Alben Barkley]

I do think that uncertainty is especially high at the moment because we're in a polling drought. I'd expect uncertainty to shrink quite a bit if there's a big wave of polls this weekend (which I think is likely as people want to get their pre-convention snapshots to compare with post-convention).

I do really miss the density of state polls we had in the 00s. I remember in the late 2000s, SUSA literally did a "Bush approval in all 50 states" monthly tracker for a while. No pollster would dare touch something like that now. If we had more actual polls of, say, Mississippi, it'd help 538's model get to 99% for Trump and eliminate some of the uncertainty.

Morning Consult at least was doing a Trump approval tracker in all 50 states, but I’m not sure if they’ve updated it in a while.

[Alben Barkley]

If I could offer a counter-point to the "538 overestimates Trump" line, consider: What % probability would you give to the Trump administration pulling something out of a hat like the Comey letter in 2016? For example, Barr arresting Hunter Biden for espionage 10 days before the election (of course, only to release him later). That uncertainty would seem to model such a scenario, even if Nate claims this doesn't model election interference. The venn diagram of interference and black swan events would seem to overlap quite a bit.

Such a thing COULD happen but:

1. That doesn’t mean it’s probable.

2. There is no guarantee voters will react the same way they did in 2016, or if that even would be enough to swing the election given that Biden’s starting from a higher point and that many people in key states will have already voted before any last-minute stunts due to the increased prevalence of early voting this year.

3. Such hypothetical possibilities should not have such an outsized effect on the model, especially when the “uncertainty” cuts both ways AND when Nate claims election interference doesn’t affect the model as you say.

It’s just absolutely absurd that they acknowledge that Biden is doing better than Hillary at her absolute peak, yet are giving him the same odds as Hillary at her low point after the Comey letter. Period. There’s no getting around it.

[wbrocks67]

I do think that uncertainty is especially high at the moment because we're in a polling drought. I'd expect uncertainty to shrink quite a bit if there's a big wave of polls this weekend (which I think is likely as people want to get their pre-convention snapshots to compare with post-convention).

I do really miss the density of state polls we had in the 00s. I remember in the late 2000s, SUSA literally did a "Bush approval in all 50 states" monthly tracker for a while. No pollster would dare touch something like that now. If we had more actual polls of, say, Mississippi, it'd help 538's model get to 99% for Trump and eliminate some of the uncertainty.

I think the problem though is that the model for whatever reason is still outputting that the electorate is going to be 2-4% more R leaning than polls are suggesting right now

[TiltsAreUnderrated]

20% Polling
30% Voting patterns
50% BS

5% data
50% pain
And 100% reason to regard with disdain

N.B. I don't think rating the election as Lean Biden is unreasonable, but the MS/AR numbers made me LOL.

[Crumpets]

[DabbingSanta]

538 is not overestimating Trump, everyone else is overestimating Biden. Same thing happened last time.

[Lief 🗽]

If I could offer a counter-point to the "538 overestimates Trump" line, consider: What % probability would you give to the Trump administration pulling something out of a hat like the Comey letter in 2016? For example, Barr arresting Hunter Biden for espionage 10 days before the election (of course, only to release him later). That uncertainty would seem to model such a scenario, even if Nate claims this doesn't model election interference. The venn diagram of interference and black swan events would seem to overlap quite a bit.

I guess this is reasonable but if that’s your outlook, then these sort of models are basically pointless. What value is added beyond simply averaging the polls here? We know that crazy things could happen; we don’t need a million lines of code to come up with a formula telling us that. If his view really is that things are so uncertain this far out that even a months’ long, consistent 8-10 point lead only results in Biden being narrowly favored, then why would anyone even pay attention to his model before like the last week of October.

[Starry Eyed Jagaloon]

538 is not overestimating Trump, everyone else is overestimating Biden. Same thing happened last time.

Even if that was true, you'd expect to see better Trump percentages in the swing states instead of these absurd numbers in safe states. Some states really do have a 0% chance of flipping.

[President Johnson]

Too optimistic for Trump, especially for the popular vote. His chance here is not higher than 1%. In all fairness, I like JHK forecast much better. This guy is doing a great job.

[ElectionsGuy]

538 is not overestimating Trump, everyone else is overestimating Biden. Same thing happened last time.

Think about this for a second, the guy who gave Trump the most realistic chance out of anybody in 2016, is now getting disdain by liberals for giving Trump a better chance than the other models. Wake up call much? And he STILL has overestimated Democrats since 2012. Silver's models have a lot of flaws and I sh**t on him a lot but he is a lot better than most of these other guys. He's going to probably take crap for the next few months then get praised when his model most likely performs best.

[Penn_Quaker_Girl]

I will never...EVER get tired of that meme template, Crumpets.  Well done!

[ProgressiveModerate]

538 is not overestimating Trump, everyone else is overestimating Biden. Same thing happened last time.

Think about this for a second, the guy who gave Trump the most realistic chance out of anybody in 2016, is now getting disdain by liberals for giving Trump a better chance than the other models. Wake up call much? And he STILL has overestimated Democrats since 2012. Silver's models have a lot of flaws and I sh**t on him a lot but he is a lot better than most of these other guys. He's going to probably take crap for the next few months then get praised when his model most likely performs best.

But there adjusting the fundementals in their model heavily in Trump's favor beacuse of 2016, which is pretty baseless.

[Horus]

The model looks a shade too Biden friendly but unlike most, Silver realizes Minnesota is more gop than Michigan.

[ProgressiveModerate]

The model looks a shade too Biden friendly but unlike most, Silver realizes Minnesota is more gop than Michigan.

I agree. My model has MN as slightly more likely to go to Trump than MI as well. But MN with a 3/10 chance of going to Trump seems a little high.

[Fmr. Gov. NickG]

Regardless of what you think about the model’s approach to uncertainty about the future, even giving Biden a 93% chance to win today seems low.

[Horatii]

Were they really that scared to just give an actual specific prediction/forecast?