• Moderators, please send me a PM if you are unable to access mod permissions. Thanks, Habsy.

OT: American Politics

Read it the other day, and for the most part I agreed with a lot of it.....you can't draw a straight line from early voting to election day percentages or final totals....and they have been historically bad predictors of final outcome (when judged in a vacuum, and not with other outside information).....in part because historically early voting = mail in. Which meant a large swath of it was made up of military folks, who vote heavily republican. Etc etc.

The key is using it along with other data, using it on state by state basis, and weighing it based on previous benchmarks in that state.

If you just blindly look at early voting totals & believe it will tell you who is going to win, and by what margin....you're liable to be wrong as often as you're right.

It's a data point for the larger equation versus being a definitive answer of what's to come.
 
Montana said:
538's model is slow on the uptick when it comes to Nevada.....the Clark County registered voter data today all but seals the race, imo.

Nate is also being extremely conservative with his model....far as I can tell a good 10% of his model's belief in a possible Trump win, equates to "maybe all we know about polling & this years data is 100% wrong....all in Trumps direction "

He's intentionally covering his ass for an enormous, unseen, unpredicted (by anyway)...event that contradicts all the data we have (and have had in past elections.)


On the bright side. ......his "30% chance Trump wins" silliness is precisely what Dems need to make sure their remaining voters don't get complacent. Should help ensure their turnout is at full force come election day.


All this bed wetting is going to look pretty funny in hindsight tho, when we see all the final totals.
Click to expand...


538 absurdly (imo) hypothesizing Trump now has a 35% to win as of today....

Nate must have read my post tho, because he has seemingly been walking back his confidence in the model of late...and I did read this this morning......

If Trump loses Nevada, Silver says he wins in only 9 percent of scenarios.
Click to expand...


Realistically, that's about what I'd give Trumps chances of winning on election day, 5-9%.
 
Thanks Montana. I'm trying to crunch the map to see if Trump has any chance of winning. I don't see it but regardless the stock market has priced in a certain amount of fear.

The VIX future contracts (gauge of fear in the market) has climbed more than 40 percent over the last six days. Market managers are paying a lot for protection --- I'm tempted to take the other side of that bet.
 
CH1 said:
Thanks Montana. I'm trying to crunch the map to see if Trump has any chance of winning. I don't see it but regardless the stock market has priced in a certain amount of fear.

The VIX future contracts (gauge of fear in the market) has climbed more than 40 percent over the last six days. Market managers are paying a lot for protection --- I'm tempted to take the other side of that bet.
Click to expand...


Check out Jon Ralston's work breaking down Nevada & early voting there.....he lays it all out with great clarity.

http://www.ktnv.com/news/ralston/the-nevada-early-voting-blog
 
Last edited:
Montana said:
Im always intrigued to hear your views on various political policies & the potential economic impacts....but this prognostication game has never been your strong suit.
Click to expand...
You're funny. Whether he has a 33% chance or a 10% chance he's still a heavy underdog. So when he loses you'll what? Claim you were right?

Too funny.
 
I don’t like making excuses for the model's output. it is what it is. i'm inclined to trust what it's telling us, with the usual polling caveats.

caveats for me are that polling usually understates young and minority voting, due to more uncertainty in their voting habits. this usually underrates dem chances.
 
zeke said:
I don’t like making excuses for the model's output. it is what it is. i'm inclined to trust what it's telling us, with the usual polling caveats.

caveats for me are that polling usually understates young and minority voting, due to more uncertainty in their voting habits. this usually underrates dem chances.
Click to expand...

I don't think Nate's making excuses for his models output....so much as he's awknowledging it's not an infallible oracle of future events. For all the love 538 gets now, it is still basically just an algorithm that fuses polling & economic data.

Nate knows better than anyone for example that early voter returns aren't included....that the discrepancy in the two campaigns GOTV aren't accounted for......

minority voting, due to more uncertainty in their voting habits
Click to expand...

...and that Spanish speaking Americans often aren't included in many phone polls. Which becomes a more and more glaring discrepancy for pollsters as the # of Latinos that get out to vote grows with each election....


His model worked fantastically well in low volatility elections, and was obviously better than anything we used to use to predict elections....it simply (imo) wobbled a bit this time around. He'll take in everything he learned thus go around, adjust accordingly and have an even better model in 20'.


Now maybe I'm completely wrong and the model was right all along....and we'll see a Trump victory, or him win Nevada, or Hillary win by only 5-10 electoral votes. I just don't think that's what we're going to see tho.
 
Last edited:
zeke said:
I don’t like making excuses for the model's output. it is what it is. i'm inclined to trust what it's telling us, with the usual polling caveats.

caveats for me are that polling usually understates young and minority voting, due to more uncertainty in their voting habits. this usually underrates dem chances.
Click to expand...

bigger picture; general election modelling by definition suffers from small sample size so I don't think caveats/tinkering are out of line
 
You must log in or register to reply here.
Back
Top