The COMG column refers to the COMG retail chain in Japan.  Each point represents 1 pre-order for that game at their store(s).   They publish these pre-order numbers every day on their web site.   We have a long running thread for the data.

http://gamrconnect.vgchartz.com/thread.php?id=159902&page=1#

The sales column is the actual units sales for the first week for each title.
The ratio is the comparison of actual sales to the pre-order figures.   A 1:1000 would mean 1 pre-order for every 1,000 unit sales.

COMG pre-orders are often near that 1:1000 ratio so predictions for games based on those COMG pre-order figures are usually not too far off.   These charts show which ones were close to the 1:1000 ratio and which were way off.

COMG is a fairly small (~10 stores) chain of game shops in Japan, and on their website they keep a daily chart showing how much the top 20 most pre-ordered games have been pre-ordered (we assume, the chart will say 100points rather than 100 pre-orders for example, but it's widely assumed that 1pt = 1 actual pre-order)

Now generally speaking, the average ratio between the number of points on the chart a game has on its last day and the number of copies a game will sell in its first week is 1pt = 1000 sales. However, in actual fact this varies massively, as you can see in Kresnik's chart, from a few hundred to 2000 or 3000. However, if you compare the performance of a given game with other similar games in the past, you can get a better idea of how that new game might sell.

Say for example that there was a new Dragon Ball game on Vita, which got 52pts on COMG. Now the first Dragon Ball game on Vita had 23pts and sold 22,346 copies in its first week, meaning that its points ratio was 1pt = 972 sales. Now assuming that the demographics on COMG and the sorts of games its customers like and dislike stayed the same, we would use the ratio and times 52pts by 972 to get 50,544, which would be a rough idea of how the new DB game might sell in its first week

Of course this a more labour intensive way to use the chart - lots of people just like following it to see what's being pre-ordered lots, or which games are ahead of others, but you can take from it what you will. It's not perfect either - when games have more than one version available, for example a regular copy and a limited edition, one may show up but not the other, so we don't know the exact number of pts, which is why some of the figures in these tables have greater than symbols, because we at least know a minimum from the one or two versions that did turn up

Hope that helps, join us in the COMG thread if you want to follow the charts, they get posted everyday (exact time will differ with your timezone, they come out a midnight here in the UK now we're using daylight saving (GMT+1)

EDIT: Beaten, wasn't even close :(

Charts updated with Tokyo New World Record & Corpse Party.

Corpse Party, like all 5pb games seemingly, has a really poor ratio. COMG consumers seem to pre-order their games in big quantities.

Whereas Tokyo New World Record has a very good ratio. It's actually published by 5pb, so seems to be the first game in a while that's bucked the trend.

I'm confused, why are there so many tables with the same things inside?

The order is different.  First one is by release date; second one is by ratio; third one is by COMG points and fourth one is by FW sales.

Just if anyone needed a point of reference to make claims such as "games with high FW sales have poor COMG ratios" you can go straight to the table ordered by FW sales to make comparisons.

I see, thanks for letting me know.

Hell! I'd never realized COMG was so small! I've looked at the stores and it turns out there are all in Niigata prefecture, which is basically the countryside or almost (Niigata is pretty far North of Tokyo). No wonder there are discrepancies... but it's actually completely amazing that we can predict national sales with that level of precision from a couple of stores lost in Northern Japan!

I take from this take there must be a good concentration of Neptunia fans in Niigata. Next time I fly to Japan I go straight there, lots of good people it seems XD

These are my thoughts as well. It's super interesting.

Neglected this thread for a while, so got it up to date with Bullet Girls/Hayarigami/Monster Hunter.

ioi hadn't bothered to include Monster Hunter in the database so I simply added the Media Create numbers to the Famitsu numbers and divided by 2. I'm thinking of doing the same for Final Fantasy X HD as well since he hasn't included that.

Thread updated. I'd been neglecting it for a while, so finally got around to it.

Interesting things to note since last update:

- Love Live! now has the biggest ratio of any game. That's mostly to do with the fact that it was 3 SKU's and none of them showed up on COMG.

- Toukiden Extreme's ratio was actually better than the first game.

- Legend of Heroes' ratio was much worse, as the first one didn't chart at all.

- DanganRonpa: Another Episode had exactly the same amount of pre-orders as 1-2 Reload.

Charts are updated as they're going to be for a while. Fate/Hollow Atraxia did extremely well in ratio; Dengeki Bunko did poorly.