Oscar 2021 Post-Mortem: Stats Analysis Wrap-Up

I’m here. Not for long, but I’m here. Not unexpectedly (after the major effort I made to get everything done in time for the Oscars), I just could not stop sleeping…  In a number of spells. I got up, did some things, got too tired again in a few hours, went back to sleep, got up again, and so on. Repeated 2-3 times, at least. Got some work done and all of the other things in the meantime, but this week is still very busy.

I do have a bit of time now to put together this stats round-up I’ve been meaning to get to for many hours, then hopefully reply those who have written to me since I went off grid, a few hours after the ceremony – we’ll see if I can get that second part done today as well. If not, over the next few days.

Just one thing before I move on to the stats: I haven’t read what people are saying and I imagine that I’m still very much in the minority here, but for me the main takeaway from Oscar night was still WE HAVE TO BRING BACK THE HOST! (Ever since they moved away from that, the Oscars have just had no personality for me, as I’ve said many times.) The best parts of the night (which weren’t many – and apart from a handful of speeches) happened when somebody acted as a host for a few minutes. (Regina King, Lil Rel Howery, etc.) But this just didn’t happen enough, at least for my liking…

Worst Oscars I’ve ever seen, that’s for sure! Although the setting was brilliant. They just didn’t do anything with it. Most of the decisions were terrible. (I still enjoyed it, of course, because the main thing for me has always been celebrating the movies – which, even if not well, they still did, just about – and making history, which always happens, inevitably.) That’s about it…

Now, about the stats… There were certainly quite a few upsets – no fewer than five categories saw winners that were being predicted by under 30% of the experts, editors, top 24 and all-star top 24 (or whatever they’re called) at Gold Derby, and only one of those upsets came in a short film category!

That said, the only true stats-busting winner was “Fight For You” in song. (One could claim Colette was in that category as well, but not really, given how unreliable the stats are in those categories in general. More on that below.) These are all of the categories where the stats favorite (or at least what I, based on the data available to me and how I read it, had decided was the stats favorite) did not win:

Best Actor

Chadwick Boseman was probably the stats favorite here. Leaving out all things based on both Mulligan and McDormand losing in the other lead category (which did not end up being relevant, and it was always unclear whether they would), he was mainly up against:

• of the previous 26 Best Actor Oscar winners in years with 8 or more Best Picture-nominated movies, only two had failed to be in one (Jeff Bridges – Crazy Heart – and Fredric March – Dr. Jekyll and Mr. Hyde -, who, however, only tied for the win with somebody who was in a Best Picture nominee);
• since 2011, no Oscar Best Actor winner whose movie had been seen by BAFTA (so, not McConaughey in Dallas Buyers Club), and certainly none that was a BAFTA nominee, had failed to win the award there as well (9/9), and BAFTA significantly altered its voting procedure around 2012-2013, making this stat probably rather more meaningful than the average ordinary 9-year matching streak;
• and, finally, losing the Independent Spirit Award (not a very strong stat, particularly in that category, but a reasonably strong one overall
• not many Oscar-winners are nominated there and lose, in any category the two have in common, at least not since 2009, for some reason).

His stats looked mostly beatable to me – although the possibility of Davis (or maybe Day or Kirby) winning complicated things, bringing other, stronger stats into the fold -, only the BAFTA streak looking rather concerning. I guess it’s possible that voters realizing they were about to give a movie not up for Best Picture four Oscars, including both Best Actor and Best Actress (given their SAG wins), might have hurt their final tallies in both categories, so perhaps those stats were more relevant than they seem, even with both losing.

Anthony Hopkins was second-favorite, at worst. His issues were:

• not having won LAFCA/NSFC/SAG/Critics Choice, a stat on 100% (26/26) up until this year, in the SAG era (but a somewhat artificial one, and not one I trusted too much to begin with, despite the lack of exceptions – I’m glad there is now an exception, confirming this);
• not having won SAG or the Globe (an alternative stat, at best, since it also includes SAG), which was going on 17 years, but had witnessed several exceptions before that;
• being potentially the oldest Best Actor winner ever (not a very convincing stat, for the same reason Parasite being a foreign film wasn’t much of a stat either – but one can never be sure).

His stats looked quite beatable to me as well, although slightly more convincing (in the negative sense) than Boseman’s, but I can see how this could be interpreted differently, too… In any case, I was quite clear after BAFTA that I thought this was going to be a very close race and Hopkins could easily lose. The stats paint the same picture. At best, one is a marginal favorite over the other, depending on which stat(s) one gives more credence to – I honestly have no idea what the correct answer is, if there even is one.

Best Actress

Here, there were three roughly equally stats-valid possibilities, with Davis probably slightly ahead of Mulligan and McDormand. Again, setting aside the stats based on Boseman winning Best Actor, which did not come into play (and, given the above, were never anywhere near guaranteed to – though they of course needed to be kept in mind to some extent, even so, as I explained),

The roadblocks for Davis were:

• having only won SAG (not even also NBR, like Halle Berry), which put her in a-single-precedent-in-the-SAG-era territory (Susan Sarandon, back when there was no AFTRA attached to SAG, which surely was an advantage for Davis there, with her TV background and all that, and back when BAFTA was post-Oscars, there was no Gold Derby Award and so on);
• not having any particularly good excuse for missing the BAFTA nomination (which Mulligan maybe did, and McDormand, the eventual winner, in any case, didn’t need, since she was nominated and even won there), which is a must for all eligible, of course (the point is, while it’s hard to see Mulligan missing with BAFTA, had there been no jury system, it was never that hard to see Davis missing, either way  but of course one can’t be sure);
• losing the Independent Spirit Award while nominated (McDormand broke this anyway, but it was on 11/11 all-time, before this year’s Oscars).

There was some precedent, at least, so I figured she maybe could beat these. I mean, I didn’t really think she would, honestly, which is why I didn’t predict her personally, but objectively it seemed and perhaps still seems like she had the least damaging stats, just about. BAFTA and its juries really did us in this year. Had McDormand won there fair and square, this would have been so much easier to call!…

Mulligan’s issues:

• not winning SAG or the Globe (34-year streak that was just broken by McDormand, but that could be argued to be all-time, since there was no SAG the last time there was an exception).

That was it. A big one, but clearly beatable, since McDormand had the same issue. (And, since I personally predicted Mulligan, I clearly thought it was beatable before the Oscars too, especially given that Davis also had big stats issues.) Any stat that’s based mostly on the votes of AFTRA and the 90 members of the HFPA is probably beatable. Still, of course, pretty strong.

Anyway, like I said, I’m just happy my two favorite groups (BFCA & Film Independent), at least, picked Carey as their winner. 🙂 As did many other groups. Pity that the industry didn’t get it… (In my opinion, of course.)

McDormand’s:

• the same no-SAG/Globe-win stat;
• the stat about losing the ISA;
• and the stat about losing 3/4 of the Globe, Critics Choice, SAG and BAFTA as a nominee, which was on 100% in Best Actress up until now (no winners had lost that many in the BFCA-SAG era), but had been beaten in the other acting categories more than once, so it never looked unbeatable, by any means.

Clearly, it’s not easy to make a case for any of these being in a much better position, stats-wise, than the others. I still think Davis had the easier stats to beat, and she had precedent for beating most of them. But who knows?! Objectively, their cases look about the same. McDormand is hardly even an upset – I didn’t think she would win anymore, I’ll admit it, after she lost SAG and beat nobody strong at BAFTA. But that’s got nothing to do with the stats. I guess AFTRA just keeps messing things up…

By the way, Andra Day was just a bad prediction, according to the stats. (Already over-explained this point, no sense going over the nearly interminable list of her stats problems again.) Even in a year with four different precursor winners, when each of the other three had their issues as well (though far, far fewer).

James Coburn was the only kind-of precedent, but even he was in a movie rather popular with critics that year, which had won an acting prize (even if not for Coburn) from NYFCC & NSFC and made the top 2 with LAFCA (in the same category). Also made the top 3 in Best Film with the NSFC & NYFCC and the top 3 in directing with the latter. It also had another Oscar nomination. (Nick Nolte, the aforementioned acting winner, in lead.) Coburn also had a SAG nomination – BAFTA only had four acting nominees and took place long after the Oscars, in those days. And he was in supporting, where it’s probably easier for such major stats upsets to happen, anyway. (Or was, prior to 2005. One can’t really get away with such anti-stats predictions anymore, above the line.) History suggests it.

It was nice to see most (though not all) of the pundits realized Day was just not a good call by the end – even if one thought they would want to make history, it was clear Davis, the legend, would be the one they would pick, not Day. (Another thing I argued for, earlier on.) I maintain Day was probably in fifth place. Maybe not a terribly distant fifth, I don’t know, but fifth nonetheless. The evidence is all there. The evidence to the contrary… not so much.

Best Adapted Screenplay

The only category where I, personally, went against the stats and got it right… Nomadland was the stats favorite over The Father because the latter, while eligible, had failed to be nominated for the Scripter, as well as the Gold Derby Award. There were precedents for these stats being beaten. (Individually, not together, of course, but that’s the case with most strong stats.) The main reasons I thought they would indeed be broken this year:

• Not being nominated for those two also meant Nomadland didn’t actually beat it in a final vote for either; it was probably not nominated (especially in the case of Gold Derby) mostly because not enough people saw it in time, therefore Nomadland only truly beat it at the Critics Choice, but lost to it at BAFTA (and I suspect, and have for a while, it would have lost at WGA too, had they been eligible there, which wouldn’t have affected its being the Best Picture stats front-runner anyway), which is a much better precursor for the screenplay win, it’s been established;
• In the 9 years in which I’ve run my preferential ballot simulation, the winners of that simulation (The Social Network, Zero Dark Thirty, Her, Birdman, Mad Max: Fury Road, Moonlight, Call Me By Your Name, The Favourite and Parasite) have always ended up with at least one Oscar win, often upsetting in at least one category in the process (Zero Dark Thirty tied for the sound win, Mad Max: Fury Road was not expected to win six, this much I remember, and beat The Revenant in a couple of tech categories the latter was a favorite in, Moonlight upset in picture, The Favourite upset in Best Actress, even though it also surprisingly lost in a number of categories, and Parasite upset in directing – and, from many people’s perspectives, picture), and, this year, screenplay seemed the easiest category for it to win (I of course knew it could maybe win actor too, like I said, but had decided to not predict Hopkins there – I believed Boseman’s narrative would prevail, somehow);
• Nomadland just didn’t make sense as an Oscar winner in this category, as many have said, it did not feel written enough (the old “most vs. best” rule for predicting Oscar wins) and I never bought that it would win just because it was winning picture (screenplay is not the category one “drags along” that way, most of the time – bizarrely, it ended up being Best Actress, which, however, was even easier to win, surely, requiring fewer extra votes gathered and a lower final percentage of the vote) – shout out to john smith, wherever he is! I bet he would have been arguing hard alongside me that The Father was the clear favorite for screenplay (stats or no stats), due to this and other reasons! (Well, I wouldn’t have said “clear” – but he might have.) This also played a part in my decision. He was right about Get Out and others. He was the screenplay wizard…

Best Cinematography

This was another marginal upset. Mank was always clearly in it. Its only issues were BAFTA’s 8-year matching streak in this category (since 2013, so this looked reasonably strong, despite the still-small sample) and having lost the Critics Choice (only 1/11 Oscar winners since the BFCA had introduced that category too had done so). These were clearly both well in the “beatable” range, especially since Nomadland wasn’t a 100% valid winner either, stats-wise, not having won any ADG prizes (13 of the last 15 cinematography winners at the Oscars had).

One other stat I discovered right after the ceremony makes Mank seem like an even better stats alternative to Nomadland in this category (which was probably always its best chance at winning a second Oscar, given the ASC result) than it already did: no lone nominations leader (as in not involved in a tie for the most nominations) at the Oscars since 1971 has failed to win at least two Oscars! (Before that year, there were 6 exceptions – so not many, anyway.)

Best Live Action Short

Feeling Through and The Present were maybe the stats co-favorites, having no issues, with Two Distant Strangers close by (its only “issues” being the length of its title – although it’s only three words anyway, plus this stat has obviously been overcome before, even recently – and the fact it didn’t have at least two other award wins listed on IMDb, but just the one – but I knew this might change post-Oscars – I’m curious to see if it does -, so this wasn’t even much of a valid stat, just a potentially valid one).

Best Documentary Short

This is probably the second-biggest stats upset, just about. All winners since 2003 had had at least a 7.4 score on IMDb. (Colette is on 7.2, so very close, anyway.) The 40 minute rule has also gone down again, but this was never a tremendously strong stat (again, none of the shorts stats I look at are, they’re just for orientation when I can’t decide otherwise – or should be, anyway, even if I sometimes have a tendency to overvalue them, purely out of habit, given that in the other categories stats actually do matter quite a bit) and it’s been beaten before, rather recently.

Best Original Song

“Fight For You” winning here was definitely the biggest stats upset, as I said. First of all, “Speak Now” had zero stats issues. More importantly, there were all kinds of rather strong stats going against the H.E.R. song:

• not winning at Critics Choice after being nominated for and losing the Globe (no exceptions in the BFCA era and, in fact, for the last 29 years;
• not winning at least one critics award for song (this was on an 8-year streak – there were exceptions before that, of course… no real 100%-all-time stats to be found for song, either);
• not winning the Globe and not being from an animated movie (only one exception since 1991).

This is one of those below-the-line wins stats just can’t explain. There are one or two every year. Can’t be helped. There just isn’t enough data, there aren’t enough strong precursors and there aren’t enough strong stats in these categories, like I’ve always said…

(For song, specifically, there is no corresponding BAFTA category, as well as no guild award and, therefore, there are no industry stats or clues for what is likely to win the Oscar. Only non-industry precursors, which makes the stats here automatically less reliable than even those that exist for other below-the-line categories.)

Bonus:

Stats broken by Nomadland in winning Best Picture:

• not having either the SAG Ensemble nomination or two or more SAG acting nominations.

That’s it. That’s the strongest stat based on the SAG Ensemble snub it was facing (any others that include that would just be “doubles”, basically) and it had no other snubs or losses (all season) that normally disqualify a movie from winning Best Picture. (The Artios “snub”, I guess, but that’s for the same thing and I don’t count those twice – anymore. Plus, its stat is pretty weak, anyway. Barely good enough to be mentioned. I’ll probably take it out of my table altogether, now that there’s yet another exception.) Fabulous run!

Evidently, all of the others had several such issues. Yes, even Promising Young Woman. In the critics phase, and not only. (Let alone Trial and Minari and the rest.) All of those stats held. The list would be far too long… Oh, and it was, it seems, indeed not premature to call a lock that Trial wouldn’t win Best Picture, even as early as just after Critics Choice. Further proof that, as much as people love to ignore them, their stats are quite solid, especially in Best Picture, where the stat I based the call on remains on 100% all-time…

I did fairly poorly overall with my predictions (even if I made a significant profit in Oscar bets, as usual) – I got killed by the shorts, 1/3, and only called one “upset” (The Father in screenplay). Didn’t find the path to calling any of the others. Plus that late switch to Trial in editing… Oh well. 16/23 (same as the stats-only picks), 15/20 if we don’t count the shorts. Not terrible, but not good. Congrats to all of those who did better! And I’ll see you all again in October! 🙂

And with that, I’m semi-officially on break – apart from reading the post-Oscars articles and comments (which I haven’t yet had time to do) and posting a few comments and replies of my own… I would normally stay on longer, but this year there just isn’t time. There’s still too much going on.