Monday, June 20, 2016

2015-16 Best Case/Worst Case Predictions Recap


Here's a look back at the highlights and lowlights from the Best Case/Worst Case preview posts, in which I predict the Live+SD 18-49 averages for broadcast series before their premieres.

For these averages, I tried to weed out episodes after unexpected drastic timeslot moves and (as noted on the individual posts) after the end of the regular season. There were more of the unexpected timeslot moves than usual this year, from Brooklyn Nine-Nine to Sleepy Hollow to The Vampire Diaries and The Originals. Weeding out the post-move episodes probably tends to inflate my average, since these shows would've fallen farther and I was already low on some of them even at these numbers.  But it's probably still less impactful than it would be to penalize me for episodes in unanticipated situations, so it is the lesser of two evils.

Best Picks:

NameBestWorstLikeliestActualDiff
Code Black1.750.901.251.26+0%
Life in Pieces2.701.551.981.99+1%
Best Time Ever2.101.001.511.50-1%
Hawaii Five-01.400.951.141.13-1%
Criminal Minds2.051.601.801.79-1%
The Blacklist1.801.051.441.43-1%
CSI: Cyber1.200.750.920.91-1%
The Vampire Diaries0.680.440.550.56+1%
Blindspot2.551.151.851.83-1%
Chicago Fire2.091.461.731.76+2%
Rush Hour1.700.600.910.92+2%
20/20

1.131.11-2%
Second Chance1.900.751.081.10+2%
Mike and Molly1.951.251.651.62-2%
The Middle2.001.601.841.88+2%
Arrow1.150.740.960.94-2%
Bob's Burgers1.180.861.061.04-2%
Hollywood Game Night1.340.871.101.12+2%
Heroes Reborn2.500.851.301.27-2%

Here are the 19 picks that fell within 2% of the mark. Sadly, I did not have a single pick this year that was correct to the hundredth, despite having no less than nine that were only off by 0.01 and quite a few more within 0.02.

The overall story seems to be that I had fewer amazing picks but more solid ones. These are the 19 picks that fell within 2% (14% total), while 44 (or 33%) fell within 5%. That's down from 19% and 35% last year. However, the within-10% metric was improved noticeably, from 52% last year to a new best-ever 58% (or 77 total) this year.

Specific highlights: it's always tough to peg new series, so having three of them at the top of the table is nice. (Maybe should've weeded out a couple of those extra post-Voice episodes in Best Time Ever's case, though.) Being that close on the Blindspot roller-coaster ride was also rather impressive, though I was convinced throughout the fall I was gonna be way high.

Worst Picks to the Negative:

NameBestWorstLikeliestActualDiff
Scream Queens3.201.052.181.10-50%
Minority Report2.100.881.350.71-47%
The Family2.150.831.450.82-44%
The Muppets2.851.162.081.27-39%
Galavant1.330.700.980.64-35%
The Grinder1.800.701.120.75-33%
Blood and Oil2.200.801.390.96-31%
Agent Carter1.450.821.120.80-29%
Crazy Ex-Girlfriend0.650.230.400.29-26%
Fresh Off the Boat2.231.151.771.32-25%
World's Funniest1.050.600.860.65-24%
Grandfathered1.700.751.220.92-24%

Here's everything that was more than 20% weaker than I predicted, most of which can fall into the "LOL Fox" and "LOL ABC" columns. Though Scream Queens was the season's biggest negative miss, in looking back I'm actually more surprised now by Minority Report (the only prediction that was over a tenth below my Worst Case) and by The Family (which still ended up being one of my favorite new shows of the season). They were rejected enormously from the jump, and they seemed like fairly on-brand options from my perspective.

Overall, just seven shows (5% of all predictions) went worse than my Worst Case, which I'm pretty sure is easily the lowest number of those I've had. You can see Minority, Family, Galavant and Agent Carter above. Surprisingly, the other three were all ABC veterans: Castle (at 1.087, technically behind my 1.09 worst case), Nashville (0.94 vs. my worst case 0.95) and Once Upon a Time (1.37 vs. my worst case 1.40). And I actually thought I was pretty down on Nashville in predicting over a 20% drop after it was within a league average drop the previous year! Just goes to show how much ABC underachieved this season...

Worst Picks to the Positive:

NameBestWorstLikeliestActualDiff
The X-Files2.951.302.103.17+51%
Little Big Shots2.200.751.492.17+46%
Lucifer2.250.711.051.47+40%
Superstore1.600.801.111.49+34%
Reign0.390.150.240.31+29%
CFB on Fox

0.770.98+27%
Dr. Ken1.250.750.911.11+22%
American Idol Thu2.401.331.832.23+22%
Telenovela1.200.550.790.95+20%
Chicago Med1.950.851.441.71+19%
Law and Order: SVU1.551.131.371.61+18%
The 1000.680.300.410.48+17%
American Crime1.020.600.800.93+16%
The Mysteries of Laura1.180.750.961.11+16%

Here's everything that was more than 15% bigger than predicted. Usually I feel like the biggest "mistakes" are in the negative picks, but I'm actually a lot more frustrated by many of these this year:
  • Two of the three biggest misses were Fox's winter Monday dramas; I actually thought I was fairly high on The X-Files, but the 6.1 after the football game was such an explosion that that pick was doomed from the start.
  • Superstore and Telenovela are frustrating because I kinda forced myself to weigh those late December "originals" that had been available on demand and were essentially treated as repeats. I don't think I would've ever gotten all the way to a 1.49 with Superstore, but part of me definitely wanted to go higher than 1.11 after seeing the post-Voice preview ratings.
  • Reign is kind of a case of small raw numbers leading to huge percent differences, but still should've accounted for the network's magical Friday mojo with female-skewing dramas a little more. (Especially since I have made this mistake about 10,000 times in the past.)
  • Chicago Med is another case where I got tripped up by some early indicators that were not actually relevant: namely, Chicago Fire being down 30%ish in its first two post-Voice episodes of the season. By the spring, Med was really only doing a tiny bit better than I had expected, but it did a ton better in the winter episodes. It had trouble pulling the trigger on it actually beating the year-ago comedies by triple digits, but it went and did it.
  • And the last annoying one is American Idol. I acknowledged in the post that it was a pretty hard one to piece together because of the season being so much shorter, but I don't think I did a very good job of that. Still, it was definitely stronger than I expected; I was thinking it'd break 2.0 for less than half of the season, but it stayed there for all but three weeks.
Just two shows, The X-Files and Law and Order: SVU, were better than my best case. Some day I am going to learn on SVU, I swear it...

A few other big-picture stats:

This year, 83 of 133 picks (62%) were too high. That's a lot worse than last year's 54% but better than the 70% and 66% in the previous seasons.

The median miss was -3.5%. This is halfway down the list of all the Diffs, basically a measurement of optimism/pessimism. Like the last stat, this is worse than last year's -1% but better than -5% and -9%. One interesting note is that in these last two years, this metric has lined up almost exactly with how far I was off on the league average. I expected -10%ish each of the last two years, and it was a point more than that last year and over three points more this year. So if I really wanted to spin it, I could say that I'm being too optimistic on individual shows, but rather the environment as a whole. If that makes any sense. As I said last week, it may be time to give up on expecting -10% league averages.

The median miss size (a measurement of accuracy) was 9.1%. Looks like this is my best ever, improving on last year's 9.7% and just barely on 9.3% from two years ago. Might be a bit inflated by the timeslot moves, though.

The average miss size, where the outliers count more, was also a career best at 12.0%. This is a considerable improvement on 14.9% last year, when the enormous Empire miss threw things off. (Though I said year it would've been 13.2% without Empire, so... still an improvement!)

Overall, this seems to go down as a year of improvement. I felt as the season was happening like I was doing worse, probably because there were fewer truly awesome picks and I tend to dwell on the bad ones. But it seems to be a more meat-and-potatoes year, with improved consistency in the mid-section of the list rather than anything flashy. Thanks for reading!

Here's a spreadsheet with the full record.

No comments:

Post a Comment

© SpottedRatings.com 2009-2022. All Rights Reserved.