Skip

Silver's Picks for the Silver Screen
February 19, 2009 8:34 AM   Subscribe

He predicted a losing season for the White Sox in 2007 and foresaw that the Tampa Bay Rays would be the best team in the American League in 2008, although he wrongly predicted that the Rays would win the World Series. He also predicted Obama's 6-point victory over McCain. Now the stats guru Nate Silver is picking the Oscar winners and predicting an upset win for Taraji P. Henson in the Best Supporting Actress category.
posted by jonp72 (30 comments total) 4 users marked this as a favorite

 
On Olbermann's show last night, Silver said his prediction for Henson was "very probably" a mistake.
posted by roomthreeseventeen at 8:42 AM on February 19, 2009


These are fun and all, but let's get down to business: Does this guy help with lottery numbers?
posted by The Gooch at 8:43 AM on February 19, 2009


See, I'd be curious what the results would be if he applied his calculations to (for instance) last year's Oscars. Eg, how predictable are the Oscars, really?
posted by inigo2 at 8:48 AM on February 19, 2009


See, I'd be curious what the results would be if he applied his calculations to (for instance) last year's Oscars. Eg, how predictable are the Oscars, really?

The Vegas odds generally do about as good of a job as anybody in laying out the chances for each of the nominees, although by their nature they never predict upsets. Silver's list seems fairly similar to the odds this year, except for the upset prediction for best supporting actress.
posted by burnmp3s at 8:58 AM on February 19, 2009


No shout-out for fivethirtyeight.com? Probably the most sane blog during the 2008 election season.
posted by jabberjaw at 9:02 AM on February 19, 2009 [3 favorites]


I predict Jimmy "The Greek" will rise from his grave and smack Silver's wannabe ass and then disappear just like that.
posted by pianomover at 9:05 AM on February 19, 2009


I <3 Nate Silver.
posted by ORthey at 9:08 AM on February 19, 2009


I'd be curious what the results would be if he applied his calculations to (for instance) last year's Oscars.

Isn't that exactly how he predicted this year? By examining past data, including past Oscars?
posted by DU at 9:11 AM on February 19, 2009


But can Silver top the awesome and terrifying power of the Da Colbert Code?
posted by Rhaomi at 9:12 AM on February 19, 2009


Isn't that exactly how he predicted this year? By examining past data, including past Oscars?

Yeah, I'm just impatient and curious how accurate his model is.

burnmp3s, thanks for the info about the Vegas odds, didn't even think about that.
posted by inigo2 at 9:13 AM on February 19, 2009


(...and this year's predictions)
posted by Rhaomi at 9:17 AM on February 19, 2009


inigo2: I would think that he used past data to create a predictive model that he then applies to the current Oscar race. If you turn that model back on the past, of course it's going to do a great job of predicting the past winners.

What generally happens is that something will happen this year that invalidates the model confirming William Goldman's suspicions.
posted by sexymofo at 9:37 AM on February 19, 2009 [1 favorite]


He also has a ridiculous post telling us to leave Timothy Geithner alone.
posted by euphorb at 9:42 AM on February 19, 2009


I think Obama and I would agree that his prediction for the White Sox was very disappointing, but his prediction for Obama was great.

Now I'm asking all Americans to support my proposal to President Obama to send bailout money to go to the White Sox organization, so we can get another few pitchers. I'd also like to see the Cuban Baseball Players Act of 2009 to allow for several other top Cuban players to get White Sox contracts.
posted by Ironmouth at 9:46 AM on February 19, 2009


My years of using PECOTA to inform my roto league baseball draft have taught me three things about Silver's prediction methods:

1. If a player has a lot of similarity to historical players, PECOTA is almost always dead-on;
2. If a player is more unique or has less than two years MLB experience, PECOTA is all over the place;
3. Sometimes PECOTA just hints there's something there without being too specific (e.g. last year saying that Carlos Quentin was a serviceable OF, which was enough for me to draft him in the last round and ride his bat to a fourth place finish).

I think the Henson pick is a mix of #2 and #3. He's only using 30 years of Oscar numbers, so only 180 possible data points in the six main categories. Thus, not a lot of similarity to work with. At the same time, the analysis suggests that there's a pattern to how supporting actress is awarded -- typically, only actresses in best picture nominated movies win. Which may be true, and may suggest a pattern. Whether to go with this pattern is up to the person making the decision. And that's where experience and zeitgeist comes into play. With players, it's easy -- you cross-check against the more conventional predictions systems or against CHONE. I know to never trust PECOTA's predictions about Ichiro, because Ichiro is too unique for PECOTA. I trusted PECOTA on Grant Balfour years ago and paid for it (though PECOTA was right about him in the end once he joined the Rays).

So, I don't trust Silver's prediction. The fact that even he doesn't trust it suggests that he knows what I know -- past performance and patterns formed by cross data from the past can only hint at, not predict, the future.
posted by dw at 10:29 AM on February 19, 2009




Jews with numbers: 1,454
Racist old white men talking about "hustle": 0
posted by silby at 10:33 AM on February 19, 2009


I had a friend who spent a lot of time building a system for betting. All summer, he bet on baseball - and won nearly $4000, off a $300 original stake. Pretty good, right? But then he tried the same system on football and lost all the money he'd won on baseball. Why?

Well, it turns out baseball has a lot of data. They have a very long season (mostly during the summer, when conditions are less likely to be extreme) and during the season they play all the time. Its not uncommon for a team to play two games a day. Because its not a contact sport, the risk of injury is (for professional sports) sort of low. Plus the format of the game naturall allows for statistical measurement, because batters go to the plate one at a time. With so much steady data, its easier to predict the outcome.

Football on the other hand has very low data input: teams play about once a week, for about a quarter of a year, and - most importantly - teams always run the risk of random injuries because of the physical nature of the game. Because of this, fluke conditions (like unreasonable weather or injury) are much less likely to average out than they do in baseball. Furthermore, football doesn't lend itself as easily to good statistical measurement because most of the players are not touching the ball and are not involved in every play, and thus the success or failure of the play is only sometimes their fault even when they are on the field all the time. The more people playing at the same time, the harder it is to quantify numerically their role in achieving the desired goal.

My point being: so what if Nate Silver was right about Obama and baseball? Its easy to be right about those things. There were about a million polls done every day for about two years on Obama, and baseball probably has the most anally compiled set of statistics known to man. The Oscars, on the other hand, are not based on any set of data; people are not going to vote for Slumdog Millionaire because they liked Gladiator because they are completely different movies, and people didn't vote for Gladiator because they liked Schindler's List. Trying to handicap that contest would be like trying to handicap a football team where every player only played for one year before retiring - just pointless.

And even if you could extrapolate some general theories on Oscar trends based on past wins you have to account for the fact that the voting system is so complicated because of the way they split the votes (actors vote for acting awards, directors for directing awards, etc.) that a significant portion of the time what wins might not have been very popular overall, but just popular enough with one segment of the voting bloc to get a plurality. So its not just that the data you have is incomplete, its also inconsistent - and because the actual tallies are not made public, and the whole thing is manipulated by money, marketing and campaigning - there is no reason to assume that any form of statistical analysis is going to be a better predictor than just going with your gut feeling. After all, your gut is probably more sensitive to the marketing than the numbers are, and the marketing is perhaps the biggest factor.

I like Nate Silver, but at this point, he's just whoring himself out for tv time. An actual mathematician should know: don't bet on what you don't know and can't know.
posted by Kiablokirk at 10:58 AM on February 19, 2009


An actual mathematician should know: don't bet on what you don't know and can't know.

An actual star knows that any publicity is good publicity.
posted by Brandon Blatcher at 11:06 AM on February 19, 2009


Slumdog Millionaire was the absolute biggest let-down of all the Oscar movies this year.

That movie lost all seriousness halfway through and then turned into some sort of cross between Bollywood and the Spanish TV the Simpsons watch.

Easily one of the most cynical and biggest train wrecks ever.
posted by Lipstick Thespian at 11:21 AM on February 19, 2009 [1 favorite]


That movie lost all seriousness halfway through and then turned into some sort of cross between Bollywood and the Spanish TV the Simpsons watch.

Shut up and dance!
posted by Cool Papa Bell at 11:32 AM on February 19, 2009


Trying to handicap that contest would be like trying to handicap a football team where every player only played for one year before retiring - just pointless.

It depends on what your goals are. If you want to beat the casino by having a big enough edge over educated guesses to make up for odds that are purposely shifted in the house's favor, then you are right, statistical analysis probably isn't enough. But if the goal is to to make the best possible guess given the data available, Silver's method is fine.

So what if some applications of statistical analysis aren't much better than guessing? The point is that doing that kind of analysis well is extremely difficult, and Silver is better at it than the vast majority of people. The polling data and baseball stats that he uses are for the most part publicly available for anyone else to use, but not many people would know how to use it properly. Personally, I would much rather see people like Silver who actually understands statistics and probability doing these kinds of data-based predictions than someone who takes the Redskins Rule or other such nonsense seriously.
posted by burnmp3s at 11:56 AM on February 19, 2009


Though I've been impressed with his work with numbers, I feel the way that Oscars are selected make them a crap shoot to guess based on past behavior. Every year there is some headline story about how a certain victory (or sweep or shutout) is special because no combination of win, losses, or past behavior could have predicted it. This year will be no different.

Or I guess, after readign all the comments, that's just what Kiablokirk said.

Also, only wusses play the Oscar picking game with only the big 5 categories.
posted by MCMikeNamara at 12:25 PM on February 19, 2009


Slumdog Millionaire was the absolute biggest let-down of all the Oscar movies this year.

That movie lost all seriousness halfway through and then turned into some sort of cross between Bollywood and the Spanish TV the Simpsons watch.

Easily one of the most cynical and biggest train wrecks ever.


You know what'll be great about the end of the Oscar season? People who normally don't appreciate film will stop complaining about Slumdog Millionaire.
posted by Pope Gustafson I at 12:33 PM on February 19, 2009 [1 favorite]


Slumdog Millionaire was the absolute biggest let-down of all the Oscar movies this year.

That movie lost all seriousness halfway through and then turned into some sort of cross between Bollywood and the Spanish TV the Simpsons watch.

Easily one of the most cynical and biggest train wrecks ever.


AMEN AMEN AMEN WOULD FAVORITE AGAIN A++++++++++++++
posted by spicynuts at 3:03 PM on February 19, 2009


Pope Gustafson 1 - is there someone I can talk to from your organization about customer service?

I've watched over 2,000 movies, did a bunch of reviews on Netflix like you told me to, but I still haven't gotten my personalized Film Appreciation badge from you guys.

Also - did your people get my letter to you asking for your autographed 8 x 10 glossy? I have the frame picked out for it and everything.
posted by Lipstick Thespian at 3:28 PM on February 19, 2009


It's the fact that I do appreciate film that makes me so bitter and vociferous about Slumdog M's hype.
Indeed if it goes away empty-handed - as it should - then perhaps I'll ease up a little.
posted by Flashman at 3:35 PM on February 19, 2009


Slumdog Millionaire was the absolute biggest let-down of all the Oscar movies this year.

That's only because we all had low expectations of Benjamin Boring Button going in, so there was a lot less to be let down from.
posted by dw at 6:28 PM on February 19, 2009


Dear Lipstick Thespian,

Thank you very much for your kind words in your latest letter. Out of all the tons of fanmail I receive, yours is by far the best: every time I get one of your missives, I immediately stop what I’m doing just to read your latest pearls of wisdom.

Regarding your place in our organization, while I’ll certainly send your well-meaning complaints along to customer service, I’m afraid they’ll be of no use. You did indeed fill your quota of films (and in record time, too!), but your reviews were… well… poor, to say the least.

Take your last review: You called Slumdog Millionaire,

“Easily one of the most cynical and biggest train wrecks ever.”

It’s this kind of hyperbole and lack of proportion that many of the higher-ups just can’t stand. There are many films that would qualify for that position far more easily: The Greatest Show on Earth, Cleopatra, Caligula, Ishtar, Battlefield Earth, Star Wars: The Phantom Menace, and, most recently, Mamma Mia. All of these are truly terrible films, famous for their sheer audacity in shittiness released with the horribly smug view that people will actually watch this crap. Do you really think that Slumdog Millionaire is that bad? Be honest, now….

Moreover, it seems to me that you’ve missed the point of our society. Our goal is to treat the film as it is meant to be treated: to be enjoyed. All of our little tricks – Leaving Our Cultural Baggage Outside the Door, the “Don’t Sweat It” mantra, Give It a Try, Let the Art Speak to You – (all things that you are, of course, very well aware) are designed to maximize that enjoyment. To go in expecting a “Great Film” will inevitably make you frustrated and disappointed that the film didn’t meet with your exacting standards.

Don’t fret, though! With a few horrible films under your belt, and a tiny bit of mental adjusting, you’ll have a much better sense of what’s bad and what isn’t, and that badge should be yours in no time!

Now, about those glossies. I’m afraid I haven’t given out glossies in over forty years. I generally send oil paintings the size of Guernica dealing with me in the nude. If you’re really interested, I suppose something could be arranged….

Regards,
Pope Gustafson I
posted by Pope Gustafson I at 2:05 AM on February 20, 2009


Nate Silver chose 4 of 6. To recap:
Yep:
Supporting Actor -
Heath Ledger, 85.8%

Lead Actress -
Kate Winslet, 67.6%

Best Director -
Danny Boyle, Slumdog Millionaire, 99.7%

Best Picture
Slumdog Millionaire, 99.0%


Nope:
Supporting Actress -
Taraji P. Henson, 51.0%
instead, Penélope Cruz, 24.6%

Lead Actor -
Mickey Rourke, 71.1%
instead, Sean Penn, 19.0%
posted by Pronoiac at 9:06 PM on February 22, 2009


« Older Tiny Art Director   |   I'm going to check my Facebook... Newer »


This thread has been archived and is closed to new comments



Post