Though I'm posting the season's best and worst lines, this post will not really be about delving into my boneheadedness. (I got plenty of that in the Best Case/Worst Case post!) I will largely be using this post to ramble about a few potential rule changes I've brainstormed over the last couple years. I don't currently plan on implementing any of these, because keeping it simple has worked pretty well. But if any of these get overwhelmingly positive feedback, we'll see.
Compared to Best Case/Worst Case, the beauty of The Question is that I can compare my picks not just with each other (and with past years) but also with the consensus of the commenters. A 20% miss is not a good line no matter how the consensus responds, but it's much more forgivable if the commenters went 2-18 going over/under. So I'm going to start with the very worst mistakes of the season: the 20%+ misses that the majority won on.
| Date | Show(s) | O/U or Spread | Result | Winner | W | L | Miss Size |
| Sept. 17 | Dads + Brooklyn 9-9 | 3.85 | 4.8 | Over | 18 | 1 | 25% |
| Sept. 24 | Agents of SHIELD | 3.75 | 4.7 | Over | 16 | 4 | 25% |
| Oct. 2 | Super Fun Night | 2.35 | 3.2 | Over | 15 | 5 | 36% |
| Oct. 7 | Hart of Dixie + Beauty and the Beast | 0.95 | 0.7 | Under | 17 | 1 | 26% |
| Oct. 8 | The Originals + Supernatural | 1.65 | 2.1 | Over | 16 | 3 | 27% |
| Oct. 13 | The Walking Dead | 6.55 | 8.2 | Over | 16 | 4 | 25% |
| Jan. 2 | Community | 1.65 | 1.3 | Under | 15 | 8 | 21% |
| Jan. 6 | The Bachelor | 2.25 | 2.7 | Over | 16 | 1 | 20% |
| Mar. 9 | Resurrection OR Cosmos | 2.25 | 3.8 | Over | 11 | 10 | 69% |
| Apr. 14 | Friends with Better Lives | 1.45 | 1.8 | Over | 12 | 8 | 24% |
And here are the slightly more forgivable ones, 20%+ misses that the majority lost on:
| Date | Show(s) | O/U or Spread | Result | Winner | W | L | Miss Size |
| Sept. 16 | Sleepy Hollow | 2.45 | 3.5 | Over | 6 | 11 | 43% |
| Oct. 10 | Once Upon a Time in Wonderland | 2.15 | 1.7 | Under | 7 | 14 | 21% |
| Nov. 15 | Bones + Raising Hope | 2.65 | 2.0 | Under | 3 | 11 | 25% |
| Jan. 23 | Rake | 2.35 | 1.7 | Under | 7 | 12 | 28% |
| Jan. 27 | The Following | 2.65 | 2.0 | Under | 4 | 14 | 25% |
| Mar. 19 | The 100 | 0.75 | 0.9 | Over | 8 | 15 | 20% |
So of the 58 over/under games, 16 (28%) were off by 20% or more. 11 (19%) were off by 25% or more. 10 (17%) were games that the majority got right and I missed by 20%+, while eight (14%) were majority wins off by 25%+.
Why am I highlighting 20% and 25%? It's time for Possible Rule Change #1!
Since season one, I've considered adding an additional strategy element that would allow players to alternatively choose "Way Over" or "Way Under." Those guesses would win if the actual result went more than 20% (or more than 25%, or some other amount) over/under the line. A win on that would be worth double the points. But if, say, it didn't go "way over" (even if it went over by less than 20% or 25%), a "way over" guess would be worth a negative point. The "even if it went over" part is negotiable; maybe "over" would be worth zero and "under" be worth negative one. But I would want there to be a significant risk here.
I don't think this would be terribly easy to use in practice, since less than 20% of the games would've been "Way"-eligible that the majority got right. And most of these were in the 20-30% range, meaning many were just a tick or two from not qualifying for a "Way" win. Even a slight improvement in my predictive ability could wipe many of these out. Ultimately, this may end up getting most used at the end of the season by people looking to come from behind in the standings. And I'd feel bad if someone made a huge comeback with this due to me having a bunch of terrible picks to end the season, so it would add to the stress from my end.
And now, for the highlights. These are the 21 lines (36%) that went within 0.15 of the line:
| Date | Show(s) | O/U or Spread | Result | Winner | W | L |
| Sept. 9 | The Million Second Quiz | 1.55 | 1.7 | Over | 8 | 6 |
| Sept. 18 | Survivor | 2.75 | 2.6 | Under | 1 | 19 |
| Sept. 25 | Revolution | 1.75 | 1.8 | Over | 4 | 16 |
| Sept. 27 | Hawaii Five-0 | 1.45 | 1.6 | Over | 6 | 13 |
| Oct. 15 | The Biggest Loser | 2.25 | 2.3 | Over | 11 | 6 |
| Oct. 17 | Reign | 0.75 | 0.8 | Over | 15 | 5 |
| Nov. 4 | Mike and Molly | 2.55 | 2.6 | Over | 11 | 7 |
| Nov. 18 | Almost Human | 2.25 | 2.3 | Over | 7 | 8 |
| Jan. 7 | Intelligence | 2.45 | 2.4 | Under | 5 | 13 |
| Jan. 8 | Chicago PD | 1.85 | 2.0 | Over | 2 | 17 |
| Jan. 10 | Enlisted | 0.65 | 0.7 | Over | 16 | 5 |
| Jan. 13 | The Blacklist | 2.65 | 2.5 | Under | 6 | 13 |
| Feb. 24 | The Voice + The Blacklist | 7.95 | 7.8 | Under | 16 | 8 |
| Feb. 28 | Hannibal | 1.05 | 1.1 | Over | 16 | 5 |
| Mar. 10 | Believe | 2.65 | 2.7 | Over | 10 | 13 |
| Mar. 16 | Believe + Crisis | 2.95 | 3.1 | Over | 12 | 9 |
| Mar. 19 | The 100 | 0.75 | 0.9 | Over | 8 | 15 |
| Mar. 27 | Surviving Jack | 1.45 | 1.3 | Under | 10 | 9 |
| Mar. 30 | The Walking Dead | 7.95 | 8.0 | Over | 10 | 5 |
| Apr. 17 | Scandal | 3.35 | 3.4 | Over | 17 | 3 |
| Apr. 24 | Black Box OR Bad Teacher | 1.95 | 2.1 | Over | 7 | 15 |
Getting a win when the line is 0.15 or 0.05 points off: is it luck or skill? Certainly, size matters; I think a Walking Dead line of 7.95 that lands on 8.0 is pretty much a total crapshoot. Yet Reign and Enlisted were two series for which I had the exact correct rating in mind, and I have a feeling a sizable majority would've won whether I'd gone +0.05 or -0.05. (That's why I hate doing low-rated Questions!) So maybe that means there was some degree of skill involved, or maybe I was just lucky to be that close on those occasions.
Possible Rule Change #2: Make a win vs. the really close lines (within 0.15, or less than the "margin of error") worth less point-wise (maybe 0.8 or 0.9) than a win vs. a "bad" line for which there's more "skill" involved in beating it (1.2 or 1.1).
This is in the same ballpark as change #1, though it's less strategic on the player's part. We'll come back to this.
One other thing that has bothered me is the games when there's a heavy, heavy majority on one side. Actually putting together the numbers, it happened less often than it seemed at times:
| Majority% | Games | Wins | Losses |
| 60+ | 83% | 509 | 504 |
| 70+ | 49% | 318 | 272 |
| 80+ | 24% | 145 | 141 |
| 90+ | 8% | 53 | 41 |
On a certain level, this is unavoidable, because I don't actively try to split the votes. That's a condescending approach, would dumb down the whole "preview" aspect for non-players, and it would probably see me get burned much more often. So I put the line close to my actual prediction, but leaning toward where I perceive the consensus to be. If I have an off-consensus opinion, huge majorities will happen.
I often make mistakes in this approach, but far from always. The 80%+ majorities barely broke 50% (the 51% above is exactly equal to the win percentage in all games), and the 90% majorities won only three of their five games. I don't really believe it is "group-think"; I just feel like people who are on the fence would rather ensure that they won't lose a point vs. the rest of the pack.
So what if I give the coin-flippers an incentive to go the other route? Possible Rule Change #3: Make a win as part of the majority worth less point-wise (maybe 0.8 or 0.9) than a win as part of the minority (1.2 or 1.1).
This is a good idea in theory, but there's a major problem that sticks out to me: it gives even more power to later entrants, who already have the advantage of knowing what previous people guessed. If anyone should have the advantage, it should be earlier entrants who have the courage to speak out with less info. (Or... they just happened to come across the post first.) It would actually be fairer to go the opposite route. Possible Rule Change #4: Each player gets a very tiny deduction from a win (like a hundredth of a point) for each previous person who has entered.
To wrap this up, here's a quick look at how the standings would've been altered if change #2 (making picks on very close lines worth less) or change #3 (rewarding minority picks) were implemented. I understand that actually implementing these might have changed some picks, but I mean this just as a basic look at how much these things would matter. To illustrate it, I'm using the more extreme versions. A close pick is worth 0.8, a non-close pick worth 1.2. A minority pick is worth 1.2, a majority pick worth 0.8.
| W | L | D | Pts | Close | Close-pts | Minority | Min-pts | |
| War Is the H-Word | 37 | 25 | 1 | 43.3 | 10-11 | 46.7 | 14-10 | 41.5 |
| Billie_Dawn | 36 | 24 | 1 | 42.1 | 11-9 | 44.9 | 13-12 | 40.1 |
| ABC hater | 32 | 27 | 1 | 38.0 | 11-9 | 40.0 | 8-11 | 34.8 |
| Jack Sargeson | 31 | 31 | 1 | 37.3 | 9-12 | 39.9 | 11-17 | 35.5 |
| Dima | 31 | 31 | 1 | 37.3 | 9-12 | 39.9 | 12-16 | 35.9 |
| Chris L | 31 | 30 | 1 | 37.2 | 9-11 | 39.8 | 8-14 | 34.2 |
| James "One" Shade | 30 | 30 | 1 | 36.1 | 9-11 | 38.5 | 10-18 | 34.1 |
| omabin | 28 | 30 | 1 | 33.9 | 11-9 | 35.1 | 8-16 | 31.5 |
| DavidK | 28 | 21 | 1 | 33.0 | 11-6 | 34.2 | 6-8 | 29.8 |
| David Howell | 27 | 27 | 1 | 32.5 | 7-13 | 35.1 | 15-19 | 33.1 |
| RJ | 26 | 24 | 1 | 31.1 | 8-11 | 33.1 | 10-11 | 29.9 |
| SamECircle | 26 | 23 | 1 | 31.0 | 10-8 | 32.2 | 12-14 | 30.6 |
| CarShark | 21 | 35 | 1 | 26.7 | 8-12 | 27.7 | 5-19 | 24.5 |
| Survivor Fan | 21 | 23 | 0 | 25.4 | 8-3 | 26.4 | 4-11 | 22.8 |
| Melissa Katardo | 16 | 13 | 0 | 18.9 | 2-6 | 21.3 | 2-3 | 16.5 |
| Silvio | 16 | 12 | 1 | 18.9 | 6-7 | 19.7 | 8-11 | 18.9 |
| Igwell Predicts | 15 | 16 | 0 | 18.1 | 4-4 | 19.5 | 3-9 | 16.3 |
| Nick | 13 | 18 | 1 | 16.2 | 4-7 | 17.2 | 7-10 | 16.4 |
| Luke | 12 | 9 | 0 | 14.1 | 3-3 | 15.3 | 1-3 | 12.1 |
| John Dorian | 12 | 9 | 0 | 14.1 | 4-3 | 14.9 | 7-6 | 14.5 |
| Eric McInnis | 10 | 14 | 1 | 12.5 | 5-6 | 12.5 | 4-9 | 12.1 |
| ruckfules55 | 10 | 12 | 1 | 12.3 | 3-5 | 13.1 | 3-7 | 11.5 |
| Magnum Dias | 9 | 9 | 0 | 10.8 | 1-2 | 12.2 | 0-3 | 9.0 |
| Patrick Ausgewahlt | 8 | 13 | 1 | 10.2 | 4-4 | 10.2 | 8-9 | 11.8 |
| NBC Fan | 8 | 9 | 1 | 9.8 | 5-3 | 9.4 | 5-5 | 10.2 |
| Kissan Capers | 8 | 7 | 0 | 9.5 | 4-1 | 9.5 | 6-6 | 10.3 |
| Oliver | 8 | 4 | 1 | 9.3 | 2-1 | 10.1 | 3-4 | 8.9 |
| J S | 7 | 4 | 0 | 8.1 | 3-1 | 8.3 | 6-3 | 9.1 |
| Kitsune | 6 | 7 | 0 | 7.3 | 1-1 | 8.1 | 2-4 | 6.9 |
| SonOfTheBronx | 6 | 4 | 0 | 7.0 | 2-2 | 7.4 | 3-3 | 7.0 |
| It depends on what fails | 5 | 13 | 1 | 6.9 | 3-4 | 6.7 | 3-9 | 7.1 |
| roare93 | 6 | 2 | 0 | 6.8 | 1-0 | 7.6 | 1-2 | 6.0 |
| Ben | 5 | 3 | 0 | 5.8 | 3-1 | 5.6 | 5-3 | 6.8 |
Looking at the records in close situations, it seems that there's not a whole lot of "skill" in beating the really close lines. From the top to the bottom of the standings, pretty much everyone with a decent sample size is relatively close to 50% in those situations. What this means is that while it may be an OK idea in theory, it also has minimal effect on the standings.
The minority adjustment makes a bit more difference, as there are clearly some players who tend more toward the contrarian lifestyle than others. It would not have crowned a new champion or anything, but it does a bit more reshuffling than the other one. This seems like something that could have some value if added in after the fact, but again, the idea of further empowering the later players kinda bugs me.
Looking Ahead
I had always planned on ending the summer mini-season with the premiere of Utopia. This worked better logistically when Utopia was rumored to be premiering six weeks before the new season. However, since two weeks before premiere week is a lot less interesting than usual this year (sorry Hell's Kitchen and The Biggest Loser), I think I will still do it that way. Utopia's premiere in 16 days (September 7) will mark the end of the summer, and then there will be another brief hiatus before season three kicks off eight days later on September 15.
And again, though I've talked a lot about ideas to change the game, I don't think any of these above ideas enrich the game enough to justify the drawbacks and the added complication. The simplicity is part of what makes this work. Please let me know if you disagree or have other ideas!
3 comments:
I've been looking forward to this post and it didn't disappoint.
On the changes, I think all work in theory, but the added element of gamesmanship takes away from some a bit. Here are my thoughts, feel free to disregard them.
The way over/under option: Makes sense that it rewards people who can pick these out, but it sounds like it would be used too much at the end of the season as a ladder climbing method rather than when you actually consider the line bad. Also so few lines are over 25% off, that this would be such a sparingly used option guessers may even forget about it when its relevant.
Also the added complexity in guessing would probably discourage new guessers from joining, I would think any added complexity should come purely from the scoring, and not through further options from us.
The close line option: This looks like it balances out anyway, few are more than 2 points off 50:50 here, only David Howell moved up from this, so unless this year was unusually fair on close lines this seems like addressing a non-problem.
The minority option: This was something I thought about during the season, but maybe thats because I would have benefited from it more than most. It makes sense to reward these guesses, but I think it would just lead to all marginal guesses going against the crowd which sounds like it would actually change peoples predictions which doesn't sound good.
Also, I think rewarding these guesses is already within the game, you get a point and everyone else doesn't. The only two people who got minority calls right more than wrong (of regular-ish players) are the top two, so it seems like getting these calls right are required to do well anyway. Plus getting one right when going against the crowd felt like a reward in itself, at least for me.
The first come option: If there was any way to check if people got here and waited to see responses rather than take an immediate guess then I would say this option makes sense, but I think it would just punish those with busy schedules or those in different time-zones etc.
So basically, i don't think any would work in practice, but are interesting to see the results of here. Maybe keep them for this post for another year.
Finally on the scoring, I've been meaning to ask, whats the reason for getting 0.1 points for each guess regardless of result? It led to a position in one mid-season table where I led by 0.1 points purely through having played one more game but got it wrong, when it felt like that should put me slightly behind if anything.
I like the idea of way under/way over in theory (it is my favourite of the four possible ideas) but look at the lines it would've applied to last season, I don't think I could say honestly that I would've been confident enough to go way under/way over on any of them in practice.
In regards to possibly holding back, if I'm really torn on the line (with multiple pros and cons floating around) I might watch a few guesses come in first but 95% of the time I'd go on gut instinct anyway and a lot of the time I purposefully try to be one of the first to guess. Once the deadline has passed/ratings are in I often look to see whether I was in the majority or minority.
Looking forward I'll try not to be so silly with The Walking Dead. Each time it comes up I'm sure my mind-frame is "SURELY it must stop breaking records sometime??"!
I'm inclined to agree with that last point, and it's got me thinking about how to better implement a "margin adjustment" without making things too complicated. This has led to...
POSSIBLE RULE CHANGE #5: Bonus points for correct guesses and penalties for incorrect guesses, equal to the margin between the line and the result after finals, up to a maximum of 0.5 points.
I like that I was in the minority for a majority of picks! I think that Survivor premiere Question (where I was the one dissenter) inspired me to be a contrarian... though I got stung by the Biggest Loser one, I *really* wanted that one to go under (and even more so in retrospect given how that season ended epitomised my concerns about the show). Not sure I would want the rule to be implemented, everyone would play differently and I'm not sure it'd be for the better.
Post a Comment