>99% accuracy in the last decade
>/pol/ will insist he gets at least 20% of states wrong
>>79771589
>implying there have been at least 10 elections in the last decade.
>>79771589
What happened to that "super accurate" model that some college used to predict a Sanders victory?
>>79771688
>I literally don't know how many states are in the USA
>or 4th grade mathematics
You really don't want to keep posting in this thread, familia
>>79771589
His record of accuracy regarding Trump has been pretty terrible, specifically. Yet, you would have me believe he suddenly got his shit together.
>>79771903
that's not really a good argument. he's saying there hasn't been at least 10 presidential elections in the last 10 years. there hasn't even been 5. and yet you say 538 has a 99% accuracy percentage.
i definitely would not put any money on any odds from that
>>79771589
Use their polls + model
>>79772004
People like this will disregard the fact he predicted 99/100 of the states correct in the last 2 elections
>>79772240
the last two elections aren't even remotely comparable to the one that is happening now, and yet you think they're good sources of reference for how this one will go?
>>79771589
>haven't had a single debate you retard
>>79771589
Last decade was easy to predict. This one no one fucking knows and he's been way wrong.
Trying to predict an election before we even know VPs is insane.
>>79773097
>>79773227
>>79772838
>>79772583
>>79772240
>>79772125
>>79772004
>what is R
>does regression modeling exist?
You're retarded.
And yes, it does!
>>79773570
Nigger
>>79771589
Pols are only accurate until the last few weeks
>>79772838
trump is awful in debates though, how does that help him?
I remain firmly in the "Trump has no chance" camp, however Brexit gives me hope. Stranger things have happened.
>>79771589
/pol/ is very in denial about this subject. I'd love for Trump to win just as much as the next guy, buy I don't think it's going to happen unfortunately.
inb4 shill/cuck
>>79773570
>does regression modeling exist?
>You're retarded.
It's funny because you need at least 10 trials for a regression model to mean anything
>>79774880
>trump is awful in debates
>domineered every single republican debate
>approval spiked after every debate
Stop lying Shlomo
91% accuracy during primaries. Plurality of the probability is enough to say "we got it right" even if it was a 30-40% chance.
A lot of the general election polls are suffering from bias though and are not taking representative samples (they are sampling republicans a lot less than democrats).
Furthermore, betting markets (which have similar odds to FiveThirtyEIght) failed miserably in Brexit, with 3-1 odds for Remain. In reality the odds were more like 1-1.
Real odds if the election were held tomorrow for Trump vs. Clinton? 2-1 in Clinton's favor, accounting for the bias using my napkin math.
/pol/ has its own bias where they don't realize the demographic obstacles that Trump has, and this may lose him the election (not that Trump is really any different, Trump is actually doing relatively average for a Republican with minorities, despite his rhetoric).
I have some serious doubt that his prediction is correct, mostly because there should be a significant shame factor in this election that Silver didn't have to take into account when Obongo competed against Romney or McCain. I think Trump will win the white vote with a big margin, I also think he will attract a lot more minority vote and women than most people think.
