Japan's Weekly Blu-ray & CD Rankings for Mar 18 - 25

symbv said:
That's the same thing: Temporal BD+DVD Sales Ranking == Preliminary vol.1 Sales Ranking.

One reason I hesitated to put it for this week is the glaring omission of Tamako and MaoYuu. This indeed makes any Temporal Ranking misleading as we know full well that Tamako has not sold just 600 or so discs the past week. I am thinking of putting it only after the full list is out -- particularly in a week where there are tons of titles released. What do you think?

Kaioshin_Sama said:
4 completely different anime produced by Sunrise in the top 20. They really seem to know how to get their stuff off the shelves regardless of genre lately.

Blu-ray
*1, 81,782 81,782 MS Gundam UC vol.6 - limited edition
*4, 21,848 21,848 MS Gundam UC vol.6 - regular edition
*6, 18,429 18,429 Love Live! School Idol Project vol.1
*7, 17,145 17,145 Kyoukaisenjou no Horizon II vol.7

DVD
*1, 44,035 *,*44,035 MS Gundam UC vol.6

CD Single
69, *1,014 20,084 Love Live! School Idol Project insert song "Susume -> Tomorrow / Start:Dash!!"

*2, 53,429 53,429 Lyrical Nanoha: The Movie 2nd As - deluxe limited edition
*3, 23,528 23,528 Girls und Panzer vol.3

Meddigo said:
Does Love Live include an event ticket or any other similar incentive? I don't really see the later volumes dominating the charts like I usually do for a consistent seller. Then again, it's probably too early to tell if there will be a bigger drop than normal.

EDIT: Ah, I see that it's a discounted vol 1. Will be interesting to see if there's a big drop then.

Meddigo said:
Does Love Live include an event ticket or any other similar incentive? I don't really see the later volumes dominating the charts like I usually do for a consistent seller. Then again, it's probably too early to tell if there will be a bigger drop than normal.

EDIT: Ah, I see that it's a discounted vol 1. Will be interesting to see if there's a big drop then.

Higashi_no_Kaze said:
I don't know, Tamako Market is one of the more enjoyable KyoAni titles for me, I'm kinda sad to see it flop like that. I just hope that doesn't mean we'll get more forced drama again from them in the future? I really like KyoAni more when they are doing lighthearted stuff.

symbv said:
^ Don't forget Sunrise is a subsidiary of Bandai Namco and all these anime have Bandai Visual as the primary producer - and marketing, investment, and collaboration strategies are mainly the role of the producer. So the kampai must also be taking place in Bandai too...

Nachtwandler_21 said:
And Sunrise studio 8 get their proft once more.

bippo said:

8 can go die in a fire. 5 all the way.

Nachtwandler_21 said:
And I really hope that KyoAni compare Chuu2Koi and Tamako sales and would make right decisions about doing smth more serious next instead of pure moe.

ThangLong said:
However, have we ever gotten actual data on Amazon sales?

Dalek-baka said:
speaking of it, is there some kind of estimate on how Minami-ke or gdgd Fairies will do?

Nachtwandler_21 said:
And I really hope that KyoAni compare Chuu2Koi and Tamako sales and would make right decisions about doing smth more serious next instead of pure moe.

Meddigo said:
I'd say that Tamako Market flopped for the same reason that Nichijou did. There wasn't enough of a pull on otaku for them to buy it. There wasn't enough moe. It tried to step outside that range and as with Munto/Nichijou it will have some of the worst sales for a KyoAni show.

jmal said:

Nope. Only rankings.

symbv said:
One thing that Amazon Stalker finds it difficult to model on is the Amazon sales data are not available

jmal said:
In general it tries to estimate the actual Oricon first week sales, not just the Amazon sales. That is the goal, and it is built on the fundamental assumption that as a gigantic retailer, Amazon will be representative of sales at other retailers.

symbv said:
I am actually quite impressed by whoever develops the Stalker model to use just the changes of position on Sales ranking in Amazon on one side and the actual Oricon sales data (past and present) on the other hand (and a few minor factors like Nico reservations for tweaking purpose) to provide a running estimate of how the sales will go on the whole (not just Amazon) which turns out to have quite meaningful accuracy

Meddigo said:
I'd say that Tamako Market flopped for the same reason that Munto/Nichijou did. There wasn't enough of a pull on otaku for them to buy it. There wasn't enough moe. It tried to step outside that range and as with Munto/Nichijou it will have some of the worst sales for a KyoAni show.

Progeusz said:
@Higashi_no_Kaze, what are you talking about? KyoAni's best selling show, K-ON! had even less drama than Tamako Market. Hyouka had just small amounts.

ThangLong said:
Unless Stalker stated it on their website, I would doubt that Amazon is not providing Stalker actual sales number under a non-disclosure agreement. There is clearly a benefit for Amazon if someone can actually predict the sale (it helps with stocking issues but most importantly, contract negotiation)

ThangLong said:

So there are two huge pitfalls for Stalker. I can see why it has led them to grossly underestimated series like Love Live! (almost a 100% underestimation). The store-front effect can explain the first, but not second pitfall.

ThangLong said:
After saying all that, I still believe Stalker is a pretty reliable tool if you only want to compare the relative rankings for your series to your friend's. It can give you bragging right if your series is predicted to rank higher in sales. But as far as the actual number go, I will use it with reservations.

ThangLong said:
I'm thinking it's a good idea for symbv to continue posting Stalker ranking for upcoming sales, but maybe with accompanying numbers as well xD, if you don't mind. A lot of people might like to see how Stalker performs vs. actual Oricon sales.

jmal said:

...There's simply no way this is the case.

Why would a multinational megacorporation need to farm out sales prediction work to some guy on the internet? They can just pay a statistics PhD to crank out an algorithm as crude as Stalker's i a couple hours. And they most certainly don't need to disclose it on the internet for everyone to see.

All the information on Stalker is by definition information Amazon already has. They will have sales prediction models that are vastly more sophisticated and accurate than anything Stalker can throw together.

Yes, Stalker is a fan-generated website, full stop. There is nothing official about it.

jimal said:


Right, that's what we've been saying. Stalker is a useful tool, but you can't rely on it 100%, and you have to mix its estimates with a rough sense of how titles are doing outide of Amazon, plus a general understanding of sales history and how things work. Every week in this thread we see people who are mislead/confused by Stalker estimates.

symbv said:
^ Do you think you may have let K-On color you too much about how you see the Tamako show? For me I really see it as a deliberate attempt by the K-On staff to walk out, not wholeheartedly maybe but still with determination, of the K-On shadow they created. Some character design has K-On vibe but that's about it. We have many adults with good air time; K-On has almost none. We have romance and boys; K-On has none. We have neighborhood and community as one key theme, not in K-On which is mostly about school life and the K-On club. K-On has pop band music; here we have oldies. Not saying Tamako is not to be criticized but I don't see how the staff tries to ride on K-On's success besides the reputation they got from previously working on K-On.

jmal said:
I find the second calculation more useful, because it gives a measurement specifically calculating the magnitude of the difference between estimation and reality.

sybmv said:
100% underestimation? How do you get this number? The Stalker projection is around 55% of the actual sales, so shouldn't it be 45% underestimate?

Or our understanding of the word "underestimate" is different?

sybmv said:
Sorry, what is the second pitfall you are talking about?

1) I see a problem for their goal of trying to predict the overall number (Oricon) by predicting a reasonably large, but unreliable representative retailer(Amazon)

2) (Amazon) using not statistically equivalent method (comparing Amazon ranking with actual sales number is not statistically equivalent because they are apples and oranges. You have to compare number to number, or rankings to rankings)

sybmv said:

With what accompanying numbers? If the BD are not on sale yet, there are no sales data to accompany I think?

symbv said:

- SAO v6
- Gintama Enchousen v2
- ToLoveRu Darkness v4
- Little Busters! v4
- Vivired Operation v1
- Haganai2 v1
- Hidamari4 v4
- Minamike4 v1
- Jormungand2 v4
- Sakuraso v3
- OniAi v4
- Senran Kagura v1
- Magi 4 (DVD)
- Amnesia v1 (DVD)
- AKB0048 2nd v1
- OreShura v2
- Sasami-san v1

ThangLong said:
I'm not a Grammatician, but I think both ways are accurate.
-You can say: Stalker underestimates 42.4% of actual sales, which uses the forumla (a-b)/a
or
-Salker makes an exact 73.8% underestimation to its prediction, which uses the formula (a-b)/b. For this, I only used almost 100% as a hyperbole.

ThangLong said:
I will quote myself here, and change it a bit so that it's more clear.

ThangLong said:

The data I was talking about is Stalker's data. Just as you have provided a table with Stalker prediction ranking for next week earlier, you can add numbers next to those anime xD

jmal said:

You're looking for a complicated answer to a simple question, I think. Stalker's formula is largely predictable. To match up against data its being fed from Amazon, it would need to make adjustments that fall outside the predictable point accumulation formula, and people would notice this.

jmal said:

Stalker can achieve what it's trying to do without the actual Amazon data, it's just a bit more effort to work out initially. Instead of correlating Amazon sales data to Oricon data, it can correlate hourly/daily/median/average ranking data to Oricon data. The effect is basically the same - a relationship between Amazon rankings and Oricon data.

symbv said:
^ OK, I see what you mean. If we look at just the 4 girls I can see the automatic association made with K-On. By the way you said you found Tamako "bland and distasteful", from what I understand "distasteful" means "offensive to one's personal taste, unpleasant, objectionable because of poor taste", I hope you didn't really find Tamako that unpleasant to you?

GarLogan78 said:


I think they just need to stay away from original anime. Tamako Market and Munto both sold terribly.

jmal said:
By the way if you're looking for an evaluation of Stalker and Oricon, we've done a bit of that at Mania. Here's one graph that came out of it, thanks to ultimatemegax. I also provided some prediction vs oricon data (xlsx) for all of 2012 up to that point. He hasn't had a chance to visualize it yet but might do so if he gets a chance.

Basically, though 2012 up through part of the Fall season (a couple shows weren't out at the time), for all first volume titles that ranked (n=142), Stalker was on average -9.8% off, meaning underestimated 9.8% and the median is -2.3%.

So the average and median are fantastic but the range of values is very large: between -279% and 61%.[•] This shows that Stalker is very good in the aggregate but for any individual title it can be wildly inaccurate.

[•] -279% is Arcana BD v1, estimate 326, actual 1234.
+61% is Tari Tari BD v1, estimate 14298, actual 7758.

30 releases (21%) were overestimated by >20% while 41 (29%) were underestimated by >20%. This means exactly 50% of releases were estimated within ±20% and the same number were outside of a 20% estimate. Take any individual estimate with a grain of salt, but you can trust the overall model to at least ballpark non-storefront titles.

symbv said:
^ If it is not a very busy week like this week I think I can put up the Temporal Ranking earlier (but of course only when there are data to update when we see more sales for vol.1 of some titles)

Don't forget to look at the production company as well if you want to talk about pushing sales as it is the production company which is responsible for the marketing ranging from magazine ads, huge outdoor posters, to fan events and collaboration campaign with retailers or convenience stores etc. (so it is less confusing to say 4 anime made by Sunrise, instead of "produced")

And here Bandai really shines: Gundam UC, GalPan, Love Live, Horizon all are produced by Bandai and these titles are ruling the chart.

jmal said:
But precisely because of its shortcomings, it needs to be used carefully, and people need to take the time to understand those shortcomings - I won't rehash my spiel there though, as I think we're both in agreement on that.

symbv said:

CD Album
*3, 19,578 19,578 Kalafina "Consolation"

Source: Oricon Youtaiju

Going back to your mention of Nate Silver (I also followed his blog extensively during the election period, also Princeton Election Consortium), I think I can draw a decent comparison

Nachtwandler_21 said:
Could someone post temporal sales ranking of this season shows? I mean ranking of total maximal(like dtshyk did) or average by volume numbers with place.