Why is Nate Silver looked to?

[mollymongold]

I've never heard of him until here. why are his analysis and predictions so widely read? Has he proven himself? Just wondering.

[elkston]

1. TROLL

[mollymongold]

6. hat the hell? Im just asking why people love Nate Silver. What have his analysis shown.

[DrToast]

2. I like his site, but he's a bit overrated IMO

[BlooInBloo]

3. HAHAHAAH!!!!!

[rockymountaindem]

4. I think it's because he runs so many computer simulations based on polls

If you look at his website, you'll see that he simulates the election 10,000 times every day using updated polls. That's about as credible as you can get in the political forecasting game these days.

[mollymongold]

8. what does simulating mean?

[gravity]

18. It is the best estimate of the election based on polls

[RichardRay]

34. He uses a technique called 'Monte Carlo' simulation.

It's a method in which as many cases as can be defined are given a range of possible outcomes defined by a particular probability for each. The cases are then varied through those ranges in as many combinations as it takes to get good coverage - for the electoral votes for president he uses 10,000 iterations. The results are tabulated as to the outcome being explored; that's his 'Win Percentage' - it's the per cent of the 10,000 cases in the simulation in which each candidate wins. You can also see a bit more detail in which he breaks out the results for things like 'Obama Landslide', 'Elector College Tie', 'McCain Wins When Losing Pennsylvania', etc., etc.

Some people don't like that kind of modeling as it doesn't give a pat answer (even if some folks read it that way). It's a way of talking about likelihoods in an uncertain universe. As with any modeling methodology it can spit out nonsense (Garbage In, Garbage Out), but he seems to use it effectively.

[gblady]

5. interesting article on DU....

from earlier today....may give ya some clues...
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=132x7706356

[mollymongold]

12. thank you!

[BrklynLiberal]

24. Interesting article. Thanks for the link.

[Every Man A King]

7. His modeling correctly predicted

North Carolina and Indiana primaries within a point or 2.

[mollymongold]

9. ooo ok.

[Every Man A King]

13. Yes its only 2 results

But, it just shows that his modeling got more and more accurate as the primary season went on and it had more data to run with.

[WeDidIt]

10. Of all the prognosticators during the primaries

Nate Silver is the ONLY ONE who got each and every call spot on.

[mollymongold]

19. ooo really? wow. thank you for not having a STANK attitude.

[scheming daemons]

11. Never heard of him until here? In your less-than-a-day of being here?

Oh wait..... you'll say "I've been lurking forever but only signed up for an account today..."


:eyes:

[mollymongold]

16. why does everyone have an attitude about me asking who he is?

i just asked how good his tract record is, don't take it so personally. gawd.

[alcibiades_mystery]

20. tract record, eh?

oy.

Guess what? People have been working on the campaign and discussing here for years. You showed up today calling into question a usually accurate poll. Do you walk in to cocktail parties and shit on the living room floor? If so, expect people to call you an asshole. That's what you're doing here. Have some respect for our community, at least.

[BlooInBloo]

28. It's funny watching all of the credulous in this thread tho.

[cameozalaznick]

37. That's a bit harsh don't you think?

I would hardly call asking a simple question the equivalent of defecting on the floor.

[targetpractice]

39. Sheez...

When I joined DU, I asked a bunch of questions and got plenty of "Welcome to DU" along with helpful answers. Seems times have changed.

Why do you feel the need to be so mean-spirited and nasty?

Nate Silver is shaking up the world of polling and there is plenty to learn about him since he's new on the scene. It's not surprising that someone would ask questions about the buzz he's getting.

[arKansasJHawk]

40. Oy, indeed ...

"Shit on the living room floor?"

Really? That's what asking a question gets you? Way to demonstrate the progressive ideal of keeping an open mind, you worthless jerk. Apparently, when anyone new shows up at YOUR cocktail party, you fucking meet them at the door and beat them over the head with a baseball bat until they're unconscious.

"Respect for the community" indeed.

[Jennicut]

31. Sorry about the attitudes, a lot of us are on edge regarding the freakin election.

I would never have heard of Nate or so many other things if I didn't spend a few hours on this board each day. After the election, I need a serious break. Welcome to DU!

[mollymongold]

35. people need to calm down. and...

thank you!

[Hamlette]

36. some think you should do some research yourself

when I first heard of him, for instance, I went to his site and spent a couple of hours looking around there for information about who he is and the methodology he uses. Then I made up my own mind.

That way I don't have to rely on the opinion of strangers on the Internet.

[kstewart33]

14. Welcome to DU. Ignore the snarky responses.

Everyone is abit twitchy lately.

[alcibiades_mystery]

15. Because he knows more than Freeper Trolls

Of course, that's not hard.

:rofl:

[RichardRay]

17. He's a careful statistician with a transparent methodology who is

willing to engage in real discussions with those who visit his web site. He's willing to recalibrate his models in the face of well supported criticism. His web site provides a good, clear quick summary, but if you want to know more just keep digging, there's always some more content under every big picture graphic. His uses of Monte Carlo simulation is not something I see other places and it provides good predictive results.

He's personally left-leaning, but his statistics are straight ahead; I don't believe he colors the outcomes either blue or red.

Aside from that he's just another bozo on the bus :-).

[Muttocracy]

27. hey, learned about Monte Carlo simulation way back in school...

...now I can't say I never heard of it again :)

[Sinti]

21. He is a professional statistician for sports betting

It's this expertise, and the way he takes an amalgamation of various polls, weighted for relevance given the particulars of each poll, and comes back with what has historically been the correct outcome that makes him kind of special in this silly season. He was highly accurate in the primaries. He lives by his numbers and doesn't have a dog in the fight, if you will.

Welcome to DU.

[On the Road]

22. He is a Highly Regarded Expert on Sabermetrics

the art of statistical prediction in baseball. Silver decided to try his hand at political analysis with fivethirtyeight.com. His discussions of statistical issues seem knowledgable, and his conclusions tend to support those of DUers. Hence he is highly regarded here. Being a newcomer, he may be looked on with suspicion by more veteran political pollsters, although it is very possible his results will be equal or superior to other sites. We will find out on Tuesday nights.

[Every Man A King]

23. You can read more here

http://www.newsweek.com/id/140469

[BrklynLiberal]

41. Interesting article.

If this were a movie, Silver would have either been taken out by now, or be working for the bookies.

[KT2000]

25. Thanks for asking

I had not kept up with what he is about either - and was wondering myself.

[WallStreetNobody]

26. Because he does have very sound methodology which

has been endorsed by many pollsters, statisticians, and political pundits. And he does have a reputation in sabermetrics and other types of baseball statistical analysis prior to applying his analysis on polls. But you are right that he hasn't proven himself in this regard yet, I don't know if he did during the primaries, does anyone know if he was accurate on those? If he turns out to be very wrong on Tuesday his reputation will be shot. But than again so would most pollsters'.

[Caution]

29. He is one of the top baseball statisticians. While this sounds silly, these guys are amongst the

best stats people in the country. He's particularly well known for running simulations of the entire season and showing an incredible amount of accuracy.

[msongs]

30. isnt his claim to fame using data for sports like baseball - before getting into politics? nt

[RL3AO]

33. Yep.

Hes one of (if not the) best baseball statisticians in the country.

[krawhitham]

32. PECOTA

http://en.wikipedia.org/wiki/PECOTA

[hlthe2b]

38. He's made his name in sports statistics and predictions...

But has taken on the conventional "wisdom" in polling methods/stats with some success. I generally like his approach... we'll see how he ends up, results wise.


Because he does not come from traditional stats, polling circles, he's getting quite a bit of snide blowback from some key pollsters he has deigned to question (e.g., zogby).... Interesting guy.

[fujiyama]

42. Another good article in NY Mag

http://nymag.com/news/features/51170/

It explains how he got into this. His claim to fame was predicting NC and IN within two points. He takes a lot of factors into account other than just the polls including demographics and trends over the years. This guy strikes me as really sharp.

And he really doesn't inject his bias into his simulations. Basically it's a numbers game. For example, it's worth noting that he did have Obama losing the number of simulations at one or two points in the summer. And he does what he can to explain his methodology.

Besides the simulations, his website has a lot of great insight into the campaign. I believe either himself or the other guy who works with him traveled to many of the campaign offices. They got a great idea of the ground game. Check out www.fivethirtyeight.com out and read his FAQ as well.

[Chan790]

43. Beyond his predictive successes in politics...

he is a baseball statistician and writer for Baseball Prospectus specializing in predictive analysis (Sabrmetrics).

He invented a complex algorithm called PECOTA which has allowed him to predict the final standings in a baseball season and individual statistics of players with a greater accuracy than anybody had done so before (over the past six seasons) and to do it earlier in a season. More impressively, it has gotten more accurate every season and this year predicted the final standings of both leagues with a error margin of less than 5 wins. It even correctly predicted Tampa Bay winning the AL (the only model to do so) when most professional analysts predicted TB to finish last in the AL East division.

Using a similar mathematical approach, weighting strategy and a simulation allowing him to generate 10,000 mock elections a day based on input data (polling data, some anecdotal assumption, etc.) he has built a model which theoretically should generate accurate probability % for a number of results. The actual effectiveness of this model is only measurable in retrospect, i.e. after the election. However, based on his past successes, he has earned the benefit of the doubt for the time being.