Support NeoGAF

[Hcoregamer00]

Well, at least Nintendo made of their fans the best charts analysts ever. Christ ! When finally an epic game succeeded in drawing attention, the whole clan reunites to predict doom and gloom. Can't people enjoy games instead of killing each others over COMAPNIES hit-parade smashes ?

What the hell are you talking about, I am a Sony Fan and the arguments about the PS3 makes sense.

If the PS3 had a constant flow of quality games following MGS4, the console would in effect be "saved" in Japan. Right now, having one high quality game with very few releases coming up in the following weeks will eventually lead to the PS3 going to the status quo, which is having sub 10k weeks.

If Sony of Japan got off their collective asses and start making deals like MS and start securing big exclusive RPG's and other games that appeal to the Japanese market, then MGS4 would be a sea change because it changed Sony's mindset. After all, the reason why the console is doing so terribly in Japan is not only a perception problem, it is a corporate problem.

 

[Ceres]

If 100% of MGS4 sales were attributed to current PS3 owners (they weren't), then 25% of the PS3 base now owns that game in Japan.

That's awesome for MG4, but sad for the PS3. Why? First of all, no sales spike leading up to the title. At 10,586 units sold last week, the anticipatory buy-in was a resounding "meh." That is the first red flag.

That would be an interesting point except for the fact of the MGS4 PS3 bundle. Why go in the week before when you can get the PS3 with the game you're buying it for?

 

[XiaNaphryz]

What the hell are you talking about, I am a Sony Fan and the arguments about the PS3 makes sense.

It's a common argument made by those with a vested interest in the stupid console wars. You could look last gen and replace Nintendo and Sony and it would still apply. Did everyone forget all the bashing and poking at Nintendo back then?

Most of the regulars in the MC threads just like talking about the data. If it seems like they're biased towards one side or the other simply through their comments based on the data, then you're obviously coloring what they're saying through your own filter first.

 

[schuelma]

Back to other topics...any other ToS:KoR predictions?

 

[XiaNaphryz]

That would be an interesting point except for the fact of the MGS4 PS3 bundle. Why go in the week before when you can get the PS3 with the game you're buying it for?

There have been several PS3 bundles in the past for franchises considered bigger than MGS, so we do have past data to compare that with. Can anyone bring up past bundle first week numbers and compare them all?

 

[Hero]

I imagine most of the MGS fanbase in Japan have probably bought a PS3 (and the game) by now. It's a hyped title and has been anticipated for years.

I expect PS3 to be around the 20K range next week.

 

[Hcoregamer00]

Back to other topics...any other ToS:KoR predictions?

Hmm.....

First Day: 200k
First Week: 300k
LTD: around 450k

Wii Sales: 100k

Guessing totally out of my ass, but hey, guessing is fun.

 

[schuelma]

Hmm.....

First Day: 200k
First Week: 300k
LTD: around 450k

Wii Sales: 100k

Guessing totally out of my ass, but hey, guessing is fun.

I agree it could do very well, but like I said in an earlier case I'm wondering if Namco is going to undership this.

 

[Tron 2.0]

Does the Mario baseball game come out this week as well?

I really don't know what to expect for ToS: KoR.

 

[batbeg]

ToS:KoR Prediction:

First Day: 120k
First Week: 200k
LTD: 420k

Wii Sales: 75k

As random of a guess as my Metal Gear Solid guess, but hopefully a bit closer this time and slightly more researched (ie actually seeing how well ToS did).

 

[schuelma]

Does the Mario baseball game come out this week as well?

Yes. Should get 1st day numbers tomorrow. I don't expect too much, honestly.

 

[Hcoregamer00]

I agree it could do very well, but like I said in an earlier case I'm wondering if Namco is going to undership this.

It is not the question whether Namco is going to undership it, the question is how much Namco is going to undership it.

After all, it is Namco, they get everything wrong half the time.

 

[Stumpokapow]

Does the Mario baseball game come out this week as well?

It's out today.

Click here and read the bottom of my post for some of the basic info re: sales

 

[Ceres]

There have been several PS3 bundles in the past for franchises considered bigger than MGS, so we do have past data to compare that with. Can anyone bring up past bundle first week numbers and compare them all?

I'm not sure what has been bundled in Japan already. I can't think of any major bundles in the US. Europe had Resistence and something else I believe but that is hardly a big franchise. I think most people buy a console well before a particular game they want comes out. However, any console bump due to game hype is certainly not going to happen in the weeks before a bundle with said game is released.

 

[Tron 2.0]

It's out today.

Click here and read the bottom of my post for some of the basic info re: sales

Cool. Thanks very much for the info.

 

[Flakster99]

Back to other topics...any other ToS:KoR predictions?

Ohh, new prediction time!

First Day: 90k
First Week: 145k
LTD: 285k

Wii Sales: 70k

 

[Rolf NB]

So MGS4 just outsold the rest of the top 10 combined by 2:1, and most likely outsold the entire top 30 minus itself. Konami must be pretty happy about that.

Sony may just have sold eight weeks worth of hardware in one, but if I'm extrapolating the non-existant long-term effect of GT5P, which should have been a similarly huge game (appeal to a wide audience, strong presentation, beloved franchise), we'll be back to <20k in three weeks, and who knows where in five weeks.

 

[liuelson]

Math question for salesagers - do week to week software sales follow a geometric series? I'm using the idea that week(n+1) / week = some constant r, and LTD = [Week 1 x (1 - r^n+1) / (1 - r)]. Is that reasonable?

 

[Vilix]

I'm not sure what has been bundled in Japan already. I can't think of any major bundles in the US. Europe had Resistence and something else I believe but that is hardly a big franchise. I think most people buy a console well before a particular game they want comes out. However, any console bump due to game hype is certainly not going to happen in the weeks before a bundle with said game is released.

There was a DMC4/PS3 bundle.

 

[Stumpokapow]

Math question for salesagers - do week to week software sales follow a geometric series? I'm using the idea that week(n+1) / week = some constant r, and LTD = [Week 1 x (1 - r^n+1) / (1 - r)]. Is that reasonable?

Well, I don't think anyone has really tried to do any kind of regression to derive a more complete week x to week x+1 formula, but the tried and true napkin math is:

Day one * 2-4 = Week one.
Week one * 2-4 = Lifetime.

With the exception that Nintendo games in specific tend to have longer legs and higher weekend boosts, puzzle games and such have longer legs in general, very very hardcore games have very frontloaded sales (Musou games, SRW games, etc).

Let me see if I can use Josh's numbers to try to compute some basic info for you.

 

[jarrod]

There was a DMC4/PS3 bundle.

There was also a Mingol 5 bundle iirc.

 

[Rolf NB]

That would be an interesting point except for the fact of the MGS4 PS3 bundle. Why go in the week before when you can get the PS3 with the game you're buying it for?

A more desirable populace to attract would have been people who already had an interest in the PS3 in general, but for whom it took one big game, MGS4, to push them over the edge.
To strengthen the PS3 software market, you want people who are willing to continually buy games. If they really only want that one game, even if they wanted it so much that they bought the system for it, that doesn't improve the prospects for software sales much.
That's just outlining a potential problem. I'm not sure the pre-MGS4-week data alone allows a conclusion to be drawn.

 

[Jokeropia]

And games like Beautiful Letter Training sell close to 300,000? that is fucked up. It is even educational.

It teaches Kanji (and supposedly does it quite well), so yes.

PS3 is nearly 2 years old, with a price no cheaper than a launch console, and still being the highest among all current competitors. I think the majority of people are expecting it to have a price point lower than that of PS2's launch at the very least.

PS2's first pricedrop also didn't come until almost two years after release.

 

[noonche]

There was also a Mingol 5 bundle iirc.

Hasn't there been a bundle for every major PS3 release?

 

[Frillen]

ToS2:

First Day: 110k
First Week: 165k
LTD: 350k

 

[KajunW]

Math question for salesagers - do week to week software sales follow a geometric series? I'm using the idea that week(n+1) / week = some constant r, and LTD = [Week 1 x (1 - r^n+1) / (1 - r)]. Is that reasonable?

That seems unlikely. Game sales are usually highly skewed in the sense that the sales in the first week are huge and the dropoff after the first week is immense. After that, the sales decline seem to slow down.

There's probably too much noise (seasonal variation, sales from competition, pricing changes in software or hardware, and so on) to describe the sales over time accurately by simple univariate ways, but a better option that fits abovementioned criteria would be a time relationship based on exponential decay.

So,

y = C*exp(-k*t)

where C is some non-negative constant (in this case),
k the decay parameter,
and t the time.

C and k obviously would differ per game, per system (perhaps a non-linear mixed effects model?).

No clue what sales analysts use, but it's probably some fancy GARCH model.

 

[Knocks]

I'm more interested in the hardware sales for that week. Are the numbers out yet?

 

[distantmantra]

I'm more interested in the hardware sales for that week. Are the numbers out yet?

They'll be out tonight.

 

[d+pad]

It's out today.

Click here and read the bottom of my post for some of the basic info re: sales

Has the game really been lambasted as much as you say? I know some early hands on reports suggested there were some control issues, but that was early on. Nintendo didn't fix them?

If it sucks, it's kind of a shame. I'm no fan of baseball games, but recent vids of this game have grabbed my attention for some reason.

Guess I'll wait and see what other GAFers think (if any of them pick it up).

 

[Vinci]

Has the game really been lambasted as much as you say? I know some early hands on reports suggested there were some control issues, but that was early on. Nintendo didn't fix them?

Nintendo didn't make it.

If it sucks, it's kind of a shame. I'm no fan of baseball games, but recent vids of this game have grabbed my attention for some reason.

Grab MLB Power Pros. It's a good one.

 

[d+pad]

Nintendo didn't make it.

OK, Namco didn't fix them?

Oh, well, I'll spend my money on something else. Both Boom Blox and Chocobo's Dungeon are calling to me anyway...

 

[XiaNaphryz]

Grab MLB Power Pros. It's a good one.

Has 2K released an official date for Power Pros 2008 yet? It's almost July...

 

[d+pad]

^^ I'm pretty sure I saw the cover on the Internets recently, so it's gotta be out soon...

 

[Vinci]

Both Boom Blox and Chocobo's Dungeon are calling to me anyway...

Dear god, yes it is! It's fucking singing hymns to get you to buy it. I'm not kidding. Buy that damn thing.

@ Xia: The 29th according to Wiki.

 

[jarrod]

Grab MLB Power Pros. It's a good one.

Imagine if Nintendo had partnered with Konami instead. Super Mario Power Pros would be a phenomenon imo...

Honestly, Nintendo's deals with Namco have all been pretty miserable. Mario Baseball sux, Star Fox Assault sux, Donkey Konga was okay but way weaker than it's Taiko source material, Link rocked SC2 but Nintendo lost out on SC3/4, they bought Monolith but not Baten Kaitos... work with SEGA more instead plz!

 

[liuelson]

a better option that fits abovementioned criteria would be a time relationship based on exponential decay.

So, y = C*exp(-k*t)

where C is some non-negative constant (in this case),
k the decay parameter,
and t the time.

I started thinking along those same lines as well. The key is the "k" decay parameter, which gives a sense of "legs." Lower k means slower decay and better legs; higher k means faster decay and more front-loaded sales.

I only have MGS series data given all the recent threads and discussion, and using a geometric series model overestimates week 2 sales quite a bit. Using an exponential decay model, you can adjust the parameters to get it to fit week 1 and week 2 data. Here's what I come up with:

MGS2 Week 1 / Week 2 / LTD = 457k / 88k / 789k
k = 1.65, C = 237.5k, predicts 456k / 88k / ~570k

MGS3 Week 1 / Week 2 / LTD = 487k / 122k / 820k
k = 1.4, C = 197.5k, predicts 487k / 120k / ~650k

Obviously, a purely exponential decay model underestimates LTD quite a bit.

What about MGS4?
MGS 4 Week 1 / Week 2 / LTD = 476k / ??? / ???
k = 1.65, C = 2480k predicts 476k / 91k / ~600k
k = 1.4, C = 1930k predicts 476k / 117k / ~635k
k = 1.15, C = 1500k predicts 475k / 150k / ~700k

Edit: Google Doc (see sheet 2)

 

[d+pad]

Dear god, yes it is! It's fucking singing hymns to get you to buy it. I'm not kidding. Buy that damn thing.

Ha ha! I take it you like the game?

Actually, I just remembered that Zack & Wiki are calling to me, too. Shit, looks like I have to buy a few games in the next 2 weeks :lol

 

[Vinci]

Imagine if Nintendo had partnered with Konami instead. Super Mario Power Pros would be a phenomenon imo...

Why didn't they? Seems like it would've made a lot more sense considering Konami's general quality in its titles.

@ d+pad: It's an awesome game. Truly. I'm hoping it has awesome legs 'cause it deserves every sale it gets. Also, Z&W is fantastic too. Buy them both.

 

[Kasumi1970]

when do the hardware numbers come in?

 

[PantherLotus]

That would be an interesting point except for the fact of the MGS4 PS3 bundle. Why go in the week before when you can get the PS3 with the game you're buying it for?

Excellent point. No rebuttal needed.

I would say that the bundle would factor in as lead-in sales that usually accompany a similar title that doesn't get a bundle.

==========

To conclude the earlier discussion on MGS4 vs. PS3, this is what I came away with.

Two fundamental positions:


PantherLotus
MGS4 first week sales compared with increased PS3 hardware sales for the week show that 12.9% of those sales were to new PS3 owners. (assuming a total of 70k ps3 sales per famitsu).

Conclusion: This is good for any individual game, but it shows that the adoption of new consoles based on major franchies has already happened. If the PS3 ever climbs out of the hole, it will not be on the backs of major titles such as these, because that audience predominantly already owns one.

Rebuttals:
0. If you don't know about the MGS series sales data, you cannot extrapolate inferences or conclude with any accuracy. (INVALID) -- MGS4 and other games of its hardcore-nature all follow the same basic model outlined by Stumpa a few posts above.

1. You can't assume that all the people that want a PS3 for certain games already has one based on this one game alone. (VALID) -- Decent point, but I am using conjecture and educated reasoning based on what I know of gamers and their buying habits. I am making a conclusion based on the idea that if there were certain games unreleased that some imaginary amount of people are waiting for, that this game (and others of its caliber)

2. You can't assume that this game won't continue to bring in new PS3 owners, like a snowball effect of word-of-mouth advertising.

3. You can't assume that any one game would impact hardware sales that much over a given period of time.

4. The data shows a significant conclusion that I don't want to hear, so you must be a Nintendo fanboy.

5.


boss coming.

 

[XiaNaphryz]

boss coming.

This is why you create your post in notepad or Word, then just paste it in after. Duuuuuh.

 

[Vinci]

boss coming.

Is he a Nintendo fanboy too?

 

[Arde5643]

Is he a Nintendo fanboy too?

I think the boss is more of a slave to the wage fanboy than anything really.

 

[Stumpokapow]

I started thinking along those same lines as well. The key is the "k" decay parameter, which gives a sense of "legs." Lower k means slower decay and better legs; higher k means faster decay and more front-loaded sales.

I only have MGS series data given all the recent threads and discussion, and using a geometric series model overestimates week 2 sales quite a bit. Using an exponential decay model, you can adjust the parameters to get it to fit week 1 and week 2 data. Here's what I come up with:

What you have to keep in mind is that it makes more sense to get your decay constant using a regression on ALL known data for each of those two games than just looking at the decay from week one to week two because the game doesn't actually follow an exponential pattern; there is variance within week to week while it slowly trends downwards.

 

[Rolf NB]

It's not strictly exponential decay, particularly not so for anything that has what sales-agers call legs. If you want to model stuff like NSMB, WiiFit or lately Monster Hunter, you have to find a way to express a residual level of sales at which decay effectively stops (or at least slows down so greatly that you're not making a huge error). For most games, this leads to nothing, but it's often precisely this tail-end behaviour that marks the difference between solid performers and huge successes.
Not sure which maths to use. Easiest would be to just add a constant after the power function, but I can already tell that's quite flawed.

 

[KajunW]

I started thinking along those same lines as well. The key is the "k" decay parameter, which gives a sense of "legs." Lower k means slower decay and better legs; higher k means faster decay and more front-loaded sales.

I only have MGS series data given all the recent threads and discussion, and using a geometric series model overestimates week 2 sales quite a bit. Using an exponential decay model, you can adjust the parameters to get it to fit week 1 and week 2 data. Here's what I come up with:

MGS2 Week 1 / Week 2 / LTD = 457k / 88k / 789k
k = 1.65, C = 237.5k, predicts 456k / 88k / ~570k

MGS3 Week 1 / Week 2 / LTD = 487k / 122k / 820k
k = 1.4, C = 197.5k, predicts 487k / 120k / ~650k

Obviously, a purely exponential decay model underestimates LTD quite a bit.

What about MGS4?
MGS 4 Week 1 / Week 2 / LTD = 476k / ??? / ???
k = 1.65, C = 2480k predicts 476k / 91k / ~600k
k = 1.4, C = 1930k predicts 476k / 117k / ~635k
k = 1.15, C = 1500k predicts 475k / 150k / ~700k

Edit: Google Doc (see sheet 2)

It's not strictly exponential decay, particularly not so for anything that has what sales-agers call legs. If you want to model stuff like NSMB, WiiFit or lately Monster Hunter, you have to find a way to express a residual level of sales at which decay effectively stops (or at least slows down so greatly that you're not making a huge error). For most games, this leads to nothing, but it's often precisely this tail-end behaviour that marks the difference between solid performers and huge successes.
Not sure which maths to use. Easiest would be to just add a constant after the power function, but I can already tell that's quite flawed.

Oh, I didn't mean to imply that an exponential decay was accurate by any means; just that sales resemble more of an exponential decay than the constant liuelson initially proposed.

I see that the exponential decay is too strong, and that LTD sales are not approximated well. I am not a sales analyst, but I am a mathematician/statistician. If a statistician doesn't have a function on theoretical grounds, he will use something like splines to model the non-linear relationship (if time series like GARCH etc. are no option). But that has little prognostic value. So one might try different types of functions but that just is a lot of (educated) guessing and takes a lot of finetuning.

Anyways, you can try something until you find something satisfactory. You can try composite exponential decay functions. Another option would be inverse functions:

y = C/t^a

where
y is the sales at time t (t>0) and C is a constant (assumes value of first week sales, as y=C when t=1). a is the power that varies amongst each game, and dictates the decay speed.

e.g. if we were to focus on approaching LTD sales after 5 years, then we get --
MGS1: C=316,833; a=1.645
MGS2: C=456,747; a=1.913
MGS3: C=487,156; a=1.956
MGS4: C=476,334; a=?

It offers a much more even decay, but I can't really say if the decay is realistic enough or not. Don't have data to back it up.

 

[Phife Dawg]

PS2's first pricedrop also didn't come until almost two years after release.

I think that's not true, PS2 was launched March 2000 and the first price cut happened in the middle of 2001 (40k to 35k). Don't have any links though.

 

[viciouskillersquirrel]

I started thinking along those same lines as well. The key is the "k" decay parameter, which gives a sense of "legs." Lower k means slower decay and better legs; higher k means faster decay and more front-loaded sales.

I only have MGS series data given all the recent threads and discussion, and using a geometric series model overestimates week 2 sales quite a bit. Using an exponential decay model, you can adjust the parameters to get it to fit week 1 and week 2 data. Here's what I come up with:

MGS2 Week 1 / Week 2 / LTD = 457k / 88k / 789k
k = 1.65, C = 237.5k, predicts 456k / 88k / ~570k

MGS3 Week 1 / Week 2 / LTD = 487k / 122k / 820k
k = 1.4, C = 197.5k, predicts 487k / 120k / ~650k

Obviously, a purely exponential decay model underestimates LTD quite a bit.

What about MGS4?
MGS 4 Week 1 / Week 2 / LTD = 476k / ??? / ???
k = 1.65, C = 2480k predicts 476k / 91k / ~600k
k = 1.4, C = 1930k predicts 476k / 117k / ~635k
k = 1.15, C = 1500k predicts 475k / 150k / ~700k

Edit: Google Doc (see sheet 2)

I guess you could put in some terms that correct for the underestimation. I put in t for the x axis because I feel I understand it easier to view the function in terms of time:

y = (C*exp(-k*t) - m*x)*&#928;(t/S-S/2)

Where m is a downward-sloping gradient some kind, &#928;(t/S-S/2) is a pulse function (basically, it equals one between 0 and S) and S is the amount of time the games spends on retail shelves.

 

[jts]

It really is a fascinating battle. I expect Wii Fit to win in the end, but I never thought MH would have the legs its displayed, so who knows

MH legs are good but not comparable to Wii Fit's, believe me.

 

[viciouskillersquirrel]

Or could game sales follow a Poisson distribution with a unit lambda? The only thing that I could see messing that up is that the sale events are not purely random.