Posts: 3,303
Threads: 0
Joined: May 2010
Reputation:
52
fivethirtyeight losing favor since not towing the line
03-31-2014, 08:42 PM
Nate Silver's new fivethirtyeight site has run some articles that basically arent towing the line. And shocker. Hes losing favor with the media.
Shocking.
Still bet on his election numbers any day of the week. I can't believe how many Republicans last time actually thought Romney would win up until the last moment. Insane amounts of bias out there. This election it might be time to bet against liberals who wont listen to Nate.
Posts: 1,820
Threads: 0
Joined: Apr 2012
Reputation:
22
fivethirtyeight losing favor since not towing the line
03-31-2014, 08:51 PM
Election season is all about gambling arbs.
First, find the actual bookie odds so you can see what the real odds are. Next:
Find your democrat friends and bet on the republican at long odds. (I got Romney at 5:1)
Find your republican friends and bet on the democrat at long odds. (I got Obama at 1:2)
No matter who wins, you win.
Check out my occasionally updated travel thread -
The Wroclaw Gambit II: Dzięki Bogu - as I prepare to emigrate to Poland.
Posts: 1,536
Threads: 0
Joined: Nov 2010
Reputation:
21
fivethirtyeight losing favor since not towing the line
04-01-2014, 10:36 AM
Silver seems like a smart guy, and also quite young. It looks like he's playing the long game, which is very admirable. Ja, I know the rule is that you need to tell people what they want to hear, not the truth, but this only works in the very short term. When the next elections roll around and the tide goes out everyone can see that you are not wearing any swimming trunks if you have not told the truth.
The interesting thing about top statisticians like Silver getting involved in political forecasting is imagining what "democracy" will look like when the result can be reliably predicted to 0.1% a couple of weeks from an election?
Posts: 341
Threads: 0
Joined: Jan 2013
Reputation:
4
fivethirtyeight losing favor since not towing the line
04-01-2014, 10:52 AM
It's toeing the line. It means to adhere to a rule or doctrine, as in keeping your toes on or behind the starting line in a race.
Posts: 140
Threads: 0
Joined: Oct 2013
fivethirtyeight losing favor since not towing the line
04-01-2014, 10:57 AM
Let me put a plug in for Nate Silver's book: The Signal and the Noise. Excellent read on forecasting and probabilities (and why watching media talking heads, whether on the right or left, is a complete waste of time).
Posts: 454
Threads: 0
Joined: Sep 2012
Reputation:
75
fivethirtyeight losing favor since not towing the line
04-01-2014, 07:03 PM
Nate Silver is a fraud. Always has been, always will be. His political statistical analyses fail the most basic mathematical tests. The only reason he's popular is because the liberals who drove his career are idiots who have no idea how math works.
Silver has exactly two credits to his name: the 2008 election and the 2012 election. Neither of them work to his benefit because his model didn't predict anything that other, traditional pollsters didn't find out. McCain was a terrible candidate who lagged behind Obama for most of the campaign; same with Romney. Silver's model, at best, told us what Ipsos, Zogby, Cook etc. already had.
Everybody seems to forget how badly FiveThirtyEight fucked up both the midterm races in 2010 and the British parliamentary elections in the same year. Silver laughably predicted that the Liberal Democrats would win over a hundred seats, apparently unaware of the fact that polling people about who they think would make the best Prime Minister is useless in a Westminster system where the executive branch is drawn from the legislature, and come election day, everyone votes for their incumbent MP out of narcissism. "All those other MPs are wankers, but MY MP is awesome!"
As for the 2010 midterms, they exposed the glaring flaw in Silver's methodology: it only works when one candidate has a decisive lead in the polls. The model completely collapses in tight races and is worse than useless in three- or four-person races, which is how Silver managed to screw up the gubernatorial races in Maine, Rhode Island, Connecticut, Illinois, Minnesota and Florida as well as the Senate races in Florida, Colorado, Alaska and Nevada.
Don't even get me started on the idiotic premises underlying Silver's model. For example, weighing polls based on their historical reliability? Completely ignores the nature of risk in prediction markets as well as the potential for black swans. I'd love to watch Nassim Taleb rip this guy apart. Not to mention that a good amount of the data that Silver spits out is completely unfalsifiable (e.g. Romney has a 30 percent chance to win based on the electoral map or something like that). What exactly does that tell us? Nothing.
Basically, Silver's methodology would fail a high school precalc class and his model is worthless. It's useless when races aren't close, because he's not telling us anything new, and it's useless when races are close because the slightest amount of bias or bad polling ruins his predictions. He's a flash-in-the-pan, yesterday's news, and I imagine in five years or so we'll all be shaking our heads wondering how we took this charlatan seriously.
Tl;dr: If Nate Silver gets cannibalized by his leftist readers, nothing of value will be lost.