TBB's statistics were based off stores that were stocked, and then an average was calculated. There were many places that weren't operated and were just taking up space, which were ignored. A system monitoring RoB would have collected information from everywhere, even stores that hadn't retailed for weeks, where TBB only counted a number of stores that were functional.
Because of this, the prices were very accurate in reflecting the bulk of stores in the central area.
There's several flaws in your model: First, you were only measuring stocked stores. The actual typical price is how much was recently PURCHASED. The real price of a good is the highest price of stores that are getting SOLD OUT despite being regularly restocked. Consider the following car prices:
CAR PRICES AND MONTHLY SALES:
12 cars at $10,000 (sold)
37 cars at $12,000 (sold)
900 cars at $14,000 (sold)
11,200 cars at $16,000 (200 sold)
28,039 cars at $18,000 (3 sold)
This does not make the price of cars $17,000. The mean price is about $14,400.
The second flaw is that you can actually track which stores are doing business. Maybe they just sold out every day before you came on.
I could count any number of more flaws depending on how strict I want to be in judging your due diligence and your lack of transparency in your data gathering methods, your statistical model, and your confidence levels.
The only global system I tracked was sales for a good over the lifetime of the game, as well as the average price of a good over the lifetime of a game. I never cared to track shorter-term price fluctuations since it wasn't important to why I wanted the data, but I easily could have. One thing I can say with absolute confidence, however, is that your prices were usually way off, both in what I inferred from the global stats and my own personal experience.
