Sharky..I think there are a number of ways to get information.
1) Console and publisher shipment information
2) GFK, NPD, MC, and the rest all rely on retailers assembling data for the week so that the agencies can extrapolate it. ioi uses the agencies as reference points, but primarily gets the data from whatever assembled retailer information he can get his hands on.
3) Regression formulas. Yeah, there probably are some complex algorithms involved in assembling the data. I would think once a game stops selling at a certain level, it is just entered into the algorithm to determine how fast sales will regress to zero new units being sold at retail. If you are only covering 2% retail for instance, and the game sold only 5,000 copies this week, it is entirely possible none of your covered stores have the information, which would mean you need the formula to estimate the nuances of small scale national sales.
4) Historical trends & current trends. Big games are actually easy to predict, and small games are the ones likely to be off but since no one (except for small communities) really cares about them, it is somewhat irrelevant to publishers and viewers if the sales of those games are off by 5% or 100% since sales are so small (talking games well under 100,000 worldwide). Medium/Typical hits (100,000-300,000) worldwide are the most difficult to predict I think, since they are so common and so variable...
Does that help?
People are difficult to govern because they have too much knowledge.
When there are more laws, there are more criminals.
- Lao Tzu
