Our friend Marshall Flores wasn’t able to devote as much time to walking us through his Stasgasm process as thoroughly as he did last year. But I was really glad to hear that he could spare us a few hours of his time to apply his regression models to the data he gathered for the 2015 nominees. If you missed Marshall’s Statsgasm series in 2014, I’ll link to those detailed posts on the next page. But here are the basic charts that show the final results of his predictive analytics. What do the percentages represent? Those are the probabilities of each possible outcome, based on the available data. The Statsgasm numbers seek to predict the likelihood of each nominee winning. In my own simplest dumb-guy terms: for example, Patricia Arquette has a 96% chance of winning Best Supporting Actress Sunday night; Bennett Miller has a 1.25% chance of winning Best Director (which is barely any better than his chance of winning Best Supporting Actress).
Statsgasm 2015
by Marshall Flores
Best Picture
Best Director
Best Actress
Best Actor
Best Supporting Actress
Best Supporting Actor
Best Original Screenplay
Best Adapted Screenplay
Best Cinematography
Best Editing
Best Foreign Language Film
Best Feature Documentary
Best Animated Feature
Best Score
Best Song
Best Production Design
Best Costumes
Best Sound Editing
Best Sound Mixing
Best Visual Effects
Best Hair & Makeup
What sorcery is this?
Complete explanation from 2013-2014:
Pilot episode of AwardsDaily’s Statsgasm
Statsgasm – Week Two – Regression!
Statsgasm – Week Three
Statsgasm Episode 4: Best Picture Nomination Voting Simulation (Pt. 1)
Statsgasm Episode 4: Best Picture Nomination Voting Simulation (Pt. 2)
Statsgasm Episode 5: the Art of Prediction
Statsgasm: Final Predictions for the 86th Academy Awards, Pt 1
Statsgasm: Final Predictions for the 86th Academy Awards, Pt 2
So far in the 6 Brutally Honest Oscar Ballots on Hollywood Reporter, for Adapted Screenplay, The Imitation Game has 2, American Sniper has 2, Theory Of Everything has 1 and Whiplash has 1. It’s a very close race.
“The theory itself isn’t necessarily “wrong,” but remember theory is more or less a set of guidelines. Application of said theory to a real-world situation is a totally different matter. Sometimes a few things get lost in translation, often times it’s messy from A to B.”
Yeah, you’re right. 🙂 Very good explanation!
Yeah, I know. That’s what I don’t get. That’s why I said the theory had to be wrong here
The theory itself isn’t necessarily “wrong,” but remember theory is more or less a set of guidelines. Application of said theory to a real-world situation is a totally different matter. Sometimes a few things get lost in translation, often times it’s messy from A to B.
🙂
“When I talk about “academic” I’m talking in the context of the statistical theory (regression analysis) that forms the backbone of the Statsgasm prediction models.”
Yeah, I know. That’s what I don’t get. That’s why I said the theory had to be wrong here. I just can’t imagine any theory that ignores a crucial element like that being correct…
I was wondering when you would finally show up Claudiu. My money was around 50 comments, so I think I would’ve won that bet. 🙂
“There’s actually no academic basis for me to even consider that at all”
There’s a difference between “academic” and “practical” basis. Practically yes, precursor ineligibility is an important factor and, as stated, I do factor that in a few of the models. When I talk about “academic” I’m talking in the context of the statistical theory (regression analysis) that forms the backbone of the Statsgasm prediction models.
As I’ve always said, predicting Oscars is as much art as it is science. Building applied prediction models depends heavily on the model builder’s knowledge and judgement. There really is no standard nor right way to do things, which is why if you compare Statsgasm to, say, Ben Zauzmer’s regression models, we may end up with mostly the same predictions but the actual percentages may be quite different.
Great news! Makes a very interesting season even more interesting.
My observations and objections (that Marshall hasn’t already addressed): it makes ZERO sense for American Sniper to be as well positioned to win BP as The Grand Budapest Hotel… And I’m not sure the model takes into account the editing snub for BP. Is Birdman such a big favorite IN SPITE of the editing snub, Marshall? Keaton the favorite over Redmayne – I guess it can be argued, but I don’t think it’s correct. Redmayne’s wins are more important than Birdman’s extra nominations. The BP winners don’t sweep as often as they used to anymore. I don’t understand why Mr. Turner is so close for Cinematography, but maybe I’m missing something (probably way too much weight given to the critics’ wins – can’t think of anything else). I’m very much in agreement with the percentages for Editing, and with the fact that Birdman is a favorite over Sniper for Sound Mixing (Whiplash above both there I disagree with, at least stats-wise, but I agree it’s a big threat).
But, other than that, awesome work! Like having a review of the stats situation without having to actually waste time to do the research (which I’ve already done anyway), despite the few things I disagree with.
“There’s actually no academic basis for me to even consider that at all”
Then the theory is clearly wrong on that one! To consider a precursor win when your main opponent was ineligible as equally important to one where both were nominated is just wrong. I’m sure you can come up with counter-arguments, but I’m equally sure that those are also wrong… (Which doesn’t mean I’m right – there might be arguments I haven’t anticipated.)
“Bit surprised that Sniper is considered the movie with almost no chance to win editing.”
Not me. Makes perfect sense, stats-wise.
“And no matter what happened I still think that the AMPAS will not split DIrection and Picture for the third time on a row.”
Truth is, it’s happened before with the preferential (in the 1930s, I think it was).
“This doesn’t change my view that American Sniper is going to take down Picture, Actor, Adapted Screenplay, Film Editing and both Sound cathegories.”
So, have you bet money on that? 🙂 I’m thinking no… Probably not a gambling man, given the fact you’re willing to make such bold predictions without specifying you’re going out on a limb…
“BAFTA has always influenced the AMPAS voters”
And the guilds never have, right?…
Why do you have Keaton so far ahead, considering the SAG and BAFTA?
Already explained that the BAFTA is not in the Best Actor model due to some statistical complications. SAG is the strongest individual predictor, but there are others things Keaton has that Redmayne doesn’t, e.g. Birdman has twice as many nominations as Theory of Everything.
And try remember that although I am Dr. Frank-N-Furter to Statsgasm, do keep in mind that Statsgasm’s predictions are not necessarily my *own* predictions. Having watched only nineteen 2014 films last year (that’s what happens when you become an Oscar watcher emeritus), I have disqualified myself from making any predictions this year.
Marshall,
Why do you have Keaton so far ahead, considering the SAG and BAFTA?
What you can’t predict are the inevitable upsets – mine (from these stats) are Boyhood for BP and TGBH for Best Score and Birdman for Screenplay. Those aren’t even big upsets – just minor ones.
The Visual effects graph is pretty bad. Dawn of the Planet of the Apes swept VES, and VES has a better track record of matching the winner than BAFTA does.
BAFTA is a predictor, but it isn’t the one of the most weight. Having the most nominations in that category has been the best predictor since 2000. And VES and BAFTA have had the same track record with Oscar ever since VES came onto the scene in 2002, both missing twice (VES missed on Spiderman and Golden Compass, BAFTA missing on Spiderman and Hugo).
“Here’s a stat:
If GRAND BUDAPEST HOTEL wins Best Picture, Ralph Fiennes will become Hollywood’s first “fourbie”– i.e., an actor who has appeared in four Best Picture winners, SCHINDLER’S LIST, THE ENGLISH PATIENT and THE HURT LOCKER being the others.”
Try again: http://en.wikipedia.org/wiki/List_of_actors_who_have_appeared_in_multiple_Best_Picture_Academy_Award_winners
The Visual effects graph is pretty bad. Dawn of the Planet of the Apes swept VES, and VES has a better track record of matching the winner than BAFTA does.
Seeing a bit of pundit chatter leaning back to Linklater for director. Even Tom O’Neill switched back. Wishful thinking or are those in the know hearing something?
And Her didn’t even have the benefit of being the Bet Picture frontrunner.
I think ilone can conclude after Oscar night last year that American Hustle wasn’t nearly as big of a BP threat as some were predicting. Huge underperformance. Sony’s spectacular inability to close BP continues. “Coffee’s for closers only.”
Of course it’s fine for you to go your way as well. I was just curious as to what you were using to predict gbh. I just find it fascinating that , even when the stats were entirely against it, the majority picked Hustle last year, and now with the stats showing us the winner once again, the majority is going another way yet again. And Her didn’t even have the benefit of being the Bet Picture frontrunner.
Actually, I do factor in ineligibility by performing a manual penalization to the predictor in question (in this case the WGA). There’s actually no academic basis for me to even consider that at all, but I felt it needs to be applied in a few cases. I do this for both screenplay categories.
If you don’t agree with Statsgasm’s prediction for Grand Budapest, that’s fine. Take the pundits’ word for it – there’s a strong consensus among them for Budapest over Birdman as well. 🙂 Seriously though, there’s nothing stopping you from going your own way with predictions. Go with what you feel is right.
*Segue into rant alert*
My goal for Statsgasm has never been to go 21/21, nor is it to convince anyone that stats in general are the Alpha and the Omega to Oscar predicting – I have never believed that. My goal has always been to showcase the potential value of what (hopefully) careful analytics can offer in a field dominated by narratives and personalities. Stats and history have very clear and obvious limitations. But some people forget that they have utility as well.
Predicting the Oscars correctly is something I have never taken too seriously in the 14 years I was an active Oscar watcher (despite performing consistently above average). I won’t lie, going 21/21 would certainly help establish Statsgasm’s “legitimacy” among the many vocal numberphobes in the Oscarverse. But in the end predictions are just the top layer of a rather deep and tasty English trifle – it’s only one of the many interesting fables stats can tell with regards to the Oscars. There are many other enigmas out there that deserve further analysis. All it takes is someone with a love for both movies and math and the time/courage to endure thousands of hours of getting their hands dirty through experimentation (as I did).
Apologies for any spelling grammar problems. I’m texting this with frozen hands.
Hmm, it cut me off for some reason. Anyways, the last part was Tinker Tailor Soldier Spy. All this is to say, BAFTA is rather unreliable for screenplays and Birdman is certainly in the lead.
You’re not taking into account the fact that Birdman was ineligible for the wga, where it would’ve if win if it had been (golden globe & critics choice has led to victory there for the past decade at least). Also, BAFTA is terrible at predicting the screenplay categories. They had Hustle last year, and The Artist and In Bruges in recent years. Plus, they’ve had Philomena, Silver Linings Playbook, and Tinkr Tailor Sold
Impeccable work! My God.
Sure hopes that MR. TURNER “upset” in Cinematography materializes! (Though I’m not predicting it — going with consensus on that one hehe)
And by the way, I’ve switched my prediction in Best Actor back to Michael Keaton; like the model predicts, the pair of WGA winners are in fact winning tomorrow night and I better hush before I give you guys then whole enchilada and those Blu-rays!
I can already break the news to you guys that BOYHOOD has won Best Picture and Best Director though other than that zipping it!
Otherwise thanks to Marshall again and have a happy Academy Awards eve, everyone!
“Previous acting nominations are a factor in the Supp. Actor model, and while Norton does get a boost with his 2, Duvall got an even bigger boost with his 6. That’s the only reason why he’s the predicted runner-up.”
Roger that. Thanks for your response (and the explanation in this case), Marshall.
Not sure how Statsgasm was off the mark on many when it was 16/20 overall and 9/10 in the techs. That’s not too shabby for a statistical version of Pinocchio.
As for Budapest, its WGA and BAFTA wins (and lest we forget it beat Birdman head-to-head and the BAFTAs) propel it to pole position. Yes, Birdman is the BP favorite but let’s not forget Budapest has a wide base of support as well.
Previous acting nominations are a factor in the Supp. Actor model, and while Norton does get a boost with his 2, Duvall got an even bigger boost with his 6. That’s the only reason why he’s the predicted runner-up.
Interesting. Like last year, many of these are off the mark, but they’re still neat to look at. The DGA award alone gives Inarritu the lead here and, as has already been established, Birdman has the clear lead for Original Screenplay. I do hope it’s right about Keaton. He certainly does have a chance. That one’s going to be a nail-biter right to the end.
I’m pretty happy with your paragraph too. Could not have said it better myself – well, actually I would have used the fucks/shits/bitches without blinking. 🙂
Correction: “Thanks to Marshall for the taking his time” — Thanks to Marshall for taking his time.
Thanks to Marshall for the taking his time. Plus, that kick-ass title “Statgasm,” too.
—
[Sort of random]
I thought, with the former on top, Redmayne and Keaton were most likely breathing on each other’s necks in a close race in Best Actor category, but according to Statgasm, Keaton seems the one to lose, not to mention a vast difference between both of their forecasted performances (62.31% vs 25.68%). What a surprise indeed. I’m still predicting Redmayne, though.
Facepalm while typing this paragraph. Princess Kaguya, the best of the group, alas, has got but 5.30%, akin to Song of the Sea, while Dragon 2 [here’s your sign . . .] is taking a lead . . . . . . . .
I love what I’ve seen in Best Cinematography regardless of what’s happening on Sunday (Monday) in reality.
Birdman vs Mr. Turner.
Mr. Turner surely deserves wider recognition; so, I’m glad that it seems to have a chance to make it happen.
Quite a bit of surprise to have spotted Robert Duvall rather than Edward Norton on the first runner-up spot. At the end of the day, though, “Just Keep” rules: Glad to see that happening.
To be honest, I’m surprised by Statsgasm’s results as well. In general I predict the transformative performance over the flashy one, and I’ve found that’s been a pretty reliable rule of thumb. So despite not having seen either Birdman nor Theory of Everything, I would say Redmayne is certainly the safer bet based on experience.
That said, as I indicated last year the BAFTA is actually not a factor in Statsgasm’s Best Actor model – oddly enough, Redmayne’s percentage would have been *worse* if I had included it.
The BAFTA is a factor in Statsgasm’s Director model, but want to know the primary reason Linklater has a lead? Boyhood’s editing nomination.
Still waiting for the BIG BAD PREDICTIONS CHART! Where is it???????????????????
I officially categorized the Statsgasm prediction in Original Screenplay as a wash last year. Hustle’s 1% lead over Her was well within the model’s margin of error.
It’s that and its BAFTA win
New York Post: ABC banking on diversity to justify bump in ad rates
http://nypost.com/2015/02/19/abc-banking-on-diversity-to-justify-bump-in-oscar-ad-rates/
What’s more, ABC has been doing more than giving lip service to diversity: Just look at its sitcoms – ”Blackish,” ”Cristela,” ”Fresh Off the Boat,” not to mention Shonda Rimes’ Thursday night dramas: ”Scandal,” etc.
Let’s see how long it takes Hollywood to realize that diversity actually can lead to more $$$$$.
Nothing at the Oscars would please me more than Linklater, Inarittu, Keaton and Anderson each taking an award home. All are very derserving. Also happy to see Dick Pope’s chances for Cinematography. His work in Mr. Turner was outstanding – as was the film.
Where is the big bad predictions chart??
All white, blacks for song and a holocaust film. The Academy is hiding behind black presenters. About a third of the presenters will be black as opposed to one twenty fifth of the Academy membership.
Here’s a stat:
If GRAND BUDAPEST HOTEL wins Best Picture, Ralph Fiennes will become Hollywood’s first “fourbie”– i.e., an actor who has appeared in four Best Picture winners, SCHINDLER’S LIST, THE ENGLISH PATIENT and THE HURT LOCKER being the others.
These are the only threebies I know of, though I’m willing to be corrected:
Clark Gable
Jack Hawkins
Shirley Maclaine
Dustin Hoffmann
Meryl Streep
Jack Nicholson
John Cazale
Joe Spinell (of course)
Plus Robert Redford gets honorary membership for directing ORDINARY PEOPLE.
Ryan – I think the stat regarding Cinematography (older voters prefering Mr Turner / younger, Birdman) may just apply here on AD.
International Cinephile Society also voted Mr Turner for best cinematography: http://icsfilm.org/our-yearly-awards/2015-ics-award-winners/
Birdman is the easier vote, esp. if you haven’t bothered to see Mr Turner.
Having said that, however, it could very well apply to AMPAS, too. I’m obviously searching for hope under every rock.
Worth noting that these stats are based on concrete data, btw, not ephemeral buzz. For example, Whiplash surely does have a better chance at winning Best Adapted Screenplay than a mere 1.78%, but there are so few (if any) identifiable precedents to support such a win. We’re attuned to the buzz it has received, and to the fact that its competition is apparently weak, but statistical evidence cannot be. These charts form good guidelines for us, but also interesting indicators of when and how a frontrunner can emerge in spite of not being the statistical favourite.
I’m very happy with what I just wrote there, it’s defs one of the best paragraphs I’ve written here in some time. The only thing that’s missing from it is a few fucks, shits and bitches.
This doesn’t change my view that American Sniper is going to take down Picture, Actor, Adapted Screenplay, Film Editing and both Sound cathegories. Birdman seems in advantage, but I think that the events on the last months only launch American Sniper to the upset. And I do hate the film, I’d be happy if I’m wrong.
OMG, poor Roger Deakins, 1¨% shot on this poll. Bit surprised that Sniper is considered the movie with almost no chance to win editing.
And no matter what happened I still think that the AMPAS will not split DIrection and Picture for the third time on a row.
Surely, Whiplash has a stronger chance of winning for adapted than a paltry 1,78 %!
I would say it’s in a too-close-to-call race with TIG.
OOT – a bit support for Boyhood (at the end)
https://www.youtube.com/watch?v=sPE3R8twHrw
Glenn, see best adapted screenplay.
I’d take these winners- trade a Linklater win for a Keaton win any day!!
I still don’t feel a split coming- it’s all for boyhood or boyhood I think, but if it’s going to split, I hope it’s this way
The most shocking statistic ….. not a single win anticipated for The Imitation Game. A film I resisted for a long time but when I watched I was absolutely blown away. Such an important film about the man and what he did …. for the world. I see his statue in the Gay Village in Manchester in a very different light these days!
Birdman is Oscar bait beginning with it’s title. Putting ‘or’ instead of a colon gives it an aura of gravitas. Check it out.
Birdman: The Unexpected Virtue of Ignorance vs. Birdman or (The Unexpected Virtue of Ignorance)
Certainly the latter form seems more artistic, you know, like “Dr. Strangelove: or How I Learned to Stop Worrying and Love the Bomb.”
Just recently, I saw the 1967 Yugoslavian Film “Love Affair, or The Case of the Missing Switchboard Operator,” it was fantastic!
Marshall, I sure hope you’re right about Linklater winning. I say that because even though I’d rather see Iñárritu win, I’d rather see Boyhood win Best Picture, but Birdman still win a big award.
Are you so sure about Interstellar winning Best Visual Effects because it has the most nominations of any film in that category?
Oh, I always love these. One thing that surprises me is how far ahead it predicts Keaton despite losing the SAG and BAFTA to Eddie, who also won a Globe and has a lot going for him in his fellow Best Picture nominee.
Mr. Turner is beautiful, but I’ll be pissed if Birdman goes home empty-handed, and I’m not even that invested this year.
Good to see you back here, Marshall.
The jaw-dropper here is Mr. Turner in a near-dead heat on Cinematography. Lot of people assume Lubeski is a near-lock.
The jaw-dropper here is Mr. Turner in a near-dead heat on Cinematography. Lot of people assume Lubeski is a near-lock.
One of the most interesting things on Dr Rob’s Awards Daily Sim Ballot: Birdman/Lubezki scored higher with voters younger than 56…but AD voters older than 56 swerved to Turner/Pope… so that bodes well for older Academy members to prefer the more stately and literally picturesque work of Dick Pope.
one brutally honest Oscar voter said Birdman’s cinematography gave her a headache
I wouldn’t mind these winners one bit.