Written by Wait, what union? Sounded good in my head…
I’ve been on an AH vacation for the past few weeks, and a RL vacation this past week, but I’ve still been quite busy, so I just wanted to give everyone an update of what’s been going on in my head.
Goblin Index Rating
I don’t know why I made a header for this, since it’s pretty much the only thing I wanted to talk about. With the nearing release of Blizzard’s AH API, I’ve really been trying to lock down an effective way of making this work. The question that I would really like this process to answer is: How can I take the data that is available and make it more accessible. The Undermine Journal does a fantastic job of presenting all of the data and offering tools to make getting at the data easier, such as the notifications and the category pages. The interpretation of the data is still left up to each individual person, though, and for some that can seem like a daunting task. I would like the GIR to fill that gap; to add that context of interpretation in order to get down to what really matters, making as much gold as possible. I’m going to throw out a pot-pourri of ideas I’ve been staring at over the past few weeks to see if anyone out there has some feedback.
- The rankings would be based on factors that relate to trends: IE. identifying when a market is saturated either with extra volume, or additional competition, etc. (this is mostly to compensate for a lack of sold data. Knowing actual sales patterns would be ideal, but I would argue that weekly trends and how they relate to what is happening ATM would be much more significant than what is happening ATM by itself)
- There would be some correlation between finished products and their reagents. IE. If savage leather took a dive, the ratings for leg armor and pvp gear would change based on how that should influence prices, both now AND down the road.
- The ranking would take server size, type and progression into consideration. For example, PvP based items or enhancements would carry a higher value on a PvP server than a PvE server, and vice versa.
- Be able to expand backwards and rank things like entire professions, or even servers. I think it would be really neat to try and identify which servers are most conducive to gold-making.
- Some way of customizing it for each person’s individual routine/setup – so that you can more effectively plan and analyze your own goblin activity (this is a pretty lofty dream, and would most likely end up as a “down the road” type of feature)
So what do you guys think? I will also admit that while I am fairly oblivious as to the space requirements for housing a significant database, this is starting to feel pretty big. I might have to move the actual GIR away from this little blog and get it a house of its own. I’d prefer to plan with expansion in mind than to “make do” and have things take off faster than I can keep up with.
If anyone out there would like to bounce some ideas around, feel free to send me an email: kathroman@breakthebakwow.com or a PM from the Consortium site. I’m more than happy to turn this into a collaborative project (especially since it seems to keep getting bigger and bigger in my head). A few of you have already helped get this to where it is now, and I will recognize your efforts accordingly. If you think it’s a flat out waste of time, don’t be afraid to speak up and voice that opinion, either. You’d better bring a pretty solid case, though, if you intend on persuading me to scrap it because I’m a pretty stubborn guy and unlikely to let a few obstacles derail my master plan…
I look forward to hearing from everyone.
-Kathroman
Without sold data, which I think Blizzard has said they don’t plan to release it would still be hard to identify “hot” markets.
Only thing is if blizzards AH API info is produced more often then every hour – we may very well see even better “recent sales data” via TUJ.
Yeah. It feels like the AH API will be pretty much the same as what TUJ pulls, ATM.
One idea I had was to compare current data to 7 day max and min for areas like market value, quantity available, competition, etc. This could give an idea of where prices “could” or “should” go.
The thought process is that pricing can be artificially manipulated, but is less likely over a longer period of time. 7 day values would still be fairly relevant, since they would factor in basic pricing trends, such as maintenance, weekends, weekdays.
If we study the right variables in the right way, we should be able to predict pricing trends. I feel pretty good about this, since even historical sales data will have it’s own limitations – such as markets like glyphs, where you are only selling to any given customer once.
We’ll see, but I’m starting to get pretty optimistic about things again. It’s just a question of sorting things out in my head.
Yea I’ve mentioned it before too, calculations of these sorts are useless without accurate sold data in my opinion. Now i’m not saying tracking broader trends isn’t useful, but the more detailed you want the more inaccurate/useless it becomes without confidence in your sold data.
All that being said, I wouldn’t be opposed to helping out on a project like this.