Originally Posted by Sharlikran
If you went that route that would be different. I don't have anything against TTC personally, but it's rather common that I see in chat that everybody thinks it's useless because certain items are ridiculous as far as the price range. Say 3,000 to 120,000 ????
There are just too many people out there that try to skew prices and make people think that certain things are a good deal or worth more then thay are. Things like a green recipe that isn't worth more than a 100 gold but they're trying to sell it for 5000.
There are just too many people out there that just waste too much time trying to sell stuff that isn't really worth selling. Guilds that make billions and billions of gold per week, those people don't have spread sheets, exe files of exported MM data, and they don't care about anything more then 30 days old. Gold mats, people usually use the MM modifier and look at 10 to 15 days.
As as far as trying to keep unwanted data out of a data set that's going to be nearly impossible because people will just do stupid things anyway. Like buying and selling on multiple accounts so that the sale goes through. So I don't care if you use predictive min-max kind of modeling using one or two levels of the standard deviation nor do I care if you want to try to check against the quartile range you're not going to prevent price skewing at all.
So having a TTC like project is great, be creative nothing against that IMO, but just having 20 to 1000 data points on certain items is enough to see a trend over 90 days. Trying to retain one year of data and using the MM max setting of up to 10,000 sales before the data is trimmed just takes longer to process the information to draw the dots for the scatter plot graph. It doesn't give you a more meanigfull average. If I outpot the mode you would see there are pleanty of sales clustered all in the same value. With that, an average will be the same anyway. You are just dividing by a greater number of X.
100 + 100 + 100 + 100 + 100 + 100 + 100 + 100 + 100 is still an average of 100. I don't care how many times you add 100 together to get the average. That's why I think it's silly that people want so many data points. The outliers are the only values changing the average and they usually aren't even valid sales. I can add people to the (Renamed from previous versions) the "Account and Guild Name Filter" (to be more politically correct) and once a seller that is clearly skewing the price the Deal calculator shows the item at a Green or Blue level instead of Purple or Yellow. Because the outliers were removed.
Too much data still doesn't help you get the right price and it still doesn't account for the great amount of competition in selling items anyway.
Bonanza should change that slightly but I just don't know what effect it will have as of yet.
|
Absolutely agree with all of that. In reality I'm seeing this more as a way to get insight into the economy as opposed to (another) price checking tool. Being able to see inflation, how much of the market share an individual item takes up, how trends in what people are buying and selling change over time. A lot of it can probably be guessed by any savvy player, but if nothing else it would just be a cool tool to have.
At the end of the day though this all comes down to me needing something productive to do over the summer, and this was the best idea I came up with
.