I tried to weed out episodes that came after drastic timeslot moves, as well as episodes after the end of the regular season. In a couple instances (Bob's Burgers and American Dad!) I actually projected what the average would've been in the original timeslot (applying how much its numbers dropped after the same point last year). But I ended up being too lazy to do that for everything.
|Once Upon a Time||2.52||1.90||2.18||2.19||+0%|
|The Voice Mon Fall||4.35||3.48||3.96||3.97||+0%|
|NCIS: Los Angeles||2.73||2.27||2.45||2.43||-1%|
|The Vampire Diaries||1.22||0.94||1.03||1.02||-1%|
|American Idol Thu||3.50||2.25||2.60||2.57||-1%|
|Sunday Night Football||8.56||7.60||8.00||7.90||-1%|
|The Voice Tue Fall||4.15||3.26||3.73||3.68||-1%|
|We Are Men||2.30||1.80||1.93||1.90||-2%|
|The Big Bang Theory||5.50||4.35||4.97||5.07||+2%|
Let's start with the highlights. Above are the 20 averages (15% of the total predictions) that went within 2% of the Likeliest pick. Another 19 were within 5% (30% total), and just over half (66 of 130, or 51%) were less than 10% off.
Compared to last year:11 (9%) were within 2%, 32 (25%) were within 5% and 51 (40%) were off by less than 10%. So by the 2%, 5% and 10% standards, I had somewhat more close picks this year. Also, I had five picks nailed to the hundredth, up from two last year.
Among the specific highlights: for the second straight year, I nailed the Community pick to the hundredth! With Community now gone (at least from broadcast TV), perhaps similar-skewing Brooklyn Nine-Nine (the only perfect newbie prediction) can take up the mantle. New shows are always the toughest, so shows like Brooklyn, Dads and Surviving Jack are particular series worth being proud of. It was also nice to basically nail Revolution's rating in a very drastic timeslot move. And I was once again very close on the two biggest shows on broadcast TV, missing football and Big Bang by just a tenth apiece. (Both were on this list last year as well.)
Worst Picks to the Negative:
|The Michael J. Fox Show||2.88||1.30||2.25||1.17||-48%|
|Once Upon a Time in Wonderland||2.04||1.17||1.65||0.98||-40%|
|Sean Saves the World||2.41||1.10||1.60||1.02||-37%|
|The Crazy Ones||3.45||1.80||2.95||2.08||-29%|
|Welcome to the Family||1.75||1.00||1.30||0.93||-28%|
As usual, most of the worst misses (with a couple big exceptions) were on new shows. One of my biggest predictive failures was the NBC Thursday lineup. Michael J. Fox basically premiered right at what I thought it would average, which already made it a bad pick. But the Fox pick could've been only half as far off if it had held up reasonably. Instead, half the audience was gone by week three, and it dropped sub-1.0 in week eight. It wasn't a good show, but I didn't think it was horrible enough to experience that much post-premiere rejection. Shrug.
I can't really come up with much to say about the rest of the new shows except that they were mostly pretty stupid. Despite a 31% miss, I do think I was still one of the more pessimistic people about Hostages' prospects. It just bombed to a far more extreme degree. I always said I expected Intelligence to do better, and I was right about that, but... perhaps I should've gotten a bit more of a clue about the timeslot difficulty from the plight of Hostages. Throw in The Crazy Ones and Mom and it was definitely not a banner year for the CBS newbie class.
Worst Picks to the Positive:
|Law and Order: SVU||1.63||1.33||1.47||1.79||+22%|
|Friends with Better Lives||2.50||1.27||1.60||1.83||+14%|
|The Carrie Diaries||0.38||0.20||0.27||0.31||+14%|
|Dancing with the Stars Spring||2.29||1.77||2.05||2.33||+14%|
|Dancing with the Stars Fall||2.26||1.60||1.90||2.16||+14%|
|Growing Up Fisher||2.00||1.05||1.47||1.62||+10%|
Worst Returning Show Picks:
|The X Factor Thu||2.65||1.82||2.17||1.68||-23%|
|Law and Order: SVU||1.63||1.33||1.47||1.79||+22%|
|The X Factor Wed||2.77||1.90||2.25||1.77||-21%|
|Beauty and the Beast||0.50||0.28||0.38||0.30||-21%|
|Person of Interest||2.80||2.00||2.45||1.97||-20%|
Like in last year's post, here's a separate "worst" category weeding out the newbies, since newbies are more difficult. I'm usually harder on myself for the returnee misses. As far as mistakes go, Glee was my biggest by a long shot. I really thought the Monteith tribute stuff was going to heavily inflate the numbers for most of the fall, and instead it was purely an isolated one-episode spike, with pretty much the whole rest of the season down 40%+ year-to-year. I can't imagine anybody was all that close to predicting the truly epic disaster the season became in the second half, but this was still very dumb.
Most of the other returnee misses to the negative fall into either the "general Fox malaise" category or the "underachieving after changing timeslots" category (or both in Raising Hope's case). If there's one lesson here, it's that I probably need to be a little more punishing toward shows that change slots. However, a couple of these were a bit exaggerated by other circumstances; Hope didn't have nearly as much original Bones lead-in as I was expecting, and a large portion of Parenthood's miss could be explained by how much Michael J. Fox underachieved.
The two most overachieving returnees were Law and Order: SVU and Supernatural. The SVU spike is a good bit more surprising to me (it's the only series that went more than 0.05 points above the Best prediction) because it had seemed to be fizzling out over the last several years, then suddenly turned it around. At least Supernatural had finished the previous season on a positive note (though I expected it would get damaged by the timeslot change and less compatible lead-in). SVU really came out of nowhere.
Big Picture Stats:
87 of 130 "Likeliest" picks this season were low (66%), which is slightly less than the 70% last year. 38 were too high (29%), and five were nailed to the hundredth.
The median miss size (in other words, halfway down the list of absolute value of %Diff, a measurement of how much I missed by) was 9%. That's down from 13% last year.
The median miss (in other words, halfway down a list of %Diffs from +117% to -48%) was -5%, closer to even than last year's -9%. So I was still too optimistic in general, but I leaned less in that direction than last year.
All of these stats suggest some steady improvement vs. last year's predictions. One thing worth noting is that the average miss size (which counts the big misses much more heavily) was 13%, only a bit improved vs. the 14% last year. The basic takeaway is that I did considerably better in the middle of the pack, but the worst misses were a lot worse this year. I didn't miss any returnees last year as badly as Glee and Raising Hope, and I didn't miss any newbies as badly as MJF and Sleepy Hollow. So hopefully the total brain farts will be reduced next time.
The stat that I was most bothered by last year was how many shows did worse than my worst case scenario: over a third! I tried to allow for a somewhat wider range of possibilities this year, and I cut the worse-than-Worst volume nearly in half (to 18%). That's still a lot higher than I would like, but I'm probably not going to expand the ranges much more. I already hedge my bets enough around here. ;-) If the predictions improve, let's hope this will naturally improve as a result. There were also five shows that did better than the Best Case, but four of those were very close (within 0.05) to the Best. (The aforementioned SVU was the only show that went well above the Best Case.)
For the completists, below are four images with a complete record.