LikePuma, on 23 July 2012 - 07:10 AM, said:
Just a few ideas I've got off the top of my head:
- While both primary stat and vitality are important, they aren't worth the same, i.e. 30dex+170vita would have a different value from 100dex+100vita (or 200dex).
- Sockets shouldn't always be worth +58stat, since you're not gonna put the highest gem into a 100k chest armor. Socket value should be dependent on the overall value of the item. (e.g. 38stat for items under 100k, 46stat for items around 1-2mil, etc.)
- Values exceeding certain round numbers should be worth some extra, e.g. +101dex is worth more than +99dex by more than just +2stat points, since people tend to search AH for stats higher than certain values, such as 100+, 150+, etc.
- Specific resistances are worth almost as much as all resist on pieces with dexterity (for monks), but they wouldn't be worth as much for other classes.
Having said all that, please don't treat my points as "you got this wrong!", it's more meant like "wouldn't this be worth considering?".
Cheers mate, keep up the good work.
I like constructive criticism. However, I do have specific responses to all the points you mentioned
First, you're absolutely right about the primary stat and vitality. The reason they are grouped together actually completely has to do with sockets. I think most people would agree that despite what many have said about sockets not being better than another, perfectly rolled property, pretty much everyone wants sockets on their chests and pants (not to mention weapons, which they are probably the best stat).
With that in mind, I considered the fact that everyone would have their own way of socketing their items. For example, even if I put 3 emeralds in a chest piece I feel like works best for a monk, maybe there's 149 strength and 150 dexterity, and I just went with dexterity. Awesome amount of dexterity can go in there right? But a barb with 300 million gold wants the chest, and will put his own gems in, and found it just searching for 140+ strength. Maybe that barb wants to put in 3 vitality gems instead, so he can get other items that have less vitality and still have 40k health.
Really, grouping all the stats together was my way of comparing a chest piece with 3 sockets and some amount of strength and vitality to another piece that has maybe higher amounts of strength, dexterity and vitality but no sockets, considering that despite whatever search priorities people use, they still will have their own ideas about sockets. That brings me to my next point.
You might notice on the spreadsheet that I also considered giving sockets just 34 stats per socket, there's a column just for that. I ended up deciding to go with the 58 per socket, despite how ridiculous that sounds, because not only will a top tier demon hunter with 3 radiant star emeralds laying around favor 3 socketed pieces for his alt monk, people just prefer 3 sockets on their chests anyway. It's fair IMHO to bias towards the sockets providing more value as a property than a primary stat just for the customization. I've certainly noticed my top selling items (you can see my evidence on the sold items sheet) have 3 sockets. Really, if you think about it, since I already decided to just lump all the stats together as one big "primary stat" value, that gets multiplied by amount of properties, all it ends up doing is adding a slight amount of value to sockets compared to a high/perfect primary roll. It also kinda works because it just ends up averaging even when there's no primary stats to speak of, it's hurt when it's only 3 sockets and no other primary stats.
Theoretically, and I've worked this out, an item with 3 sockets, 200 primary stat, 300 vitality and then something like perfect all resist and armor for its final two stats ends up having the highest average values as I'd imagine a chest piece like that should. With the current formula, that theoretical piece would have a 0.88 average, and guess what, there are pieces on my list that even have a higher defense value than that, like 1.325 vs 1.38 for something that has high armor, all resist and fire resist.
About the rounded values, that's a very good point and I could consider making the formula more complicated based on what people would search for. I'd be partial to something like a slight added value based on hitting a round number, just because I know an easy way to add that, rather than a whole formula based around it. For its current purposes though, I have to admit having a value based around actually being better rather than perception is more interesting to me, even though I'm mainly using the formula to find AH values anyway.
Finally to address the monk thing, and also to add in a note about gold find+ pickup radius, I actually tried that when I logged 150 helms and hated how skewed the list would be based on those values. I figured I could change the sale value myself if it had an extraneous value based on combos and certain builds.
I know more than anything this is by no means perfect, I agree, but in reality this whole project's goal was to put these items closer to the top than other pieces. Then I can just look at the list and choose which ones to sell.
Edited by tybrone, 23 July 2012 - 08:44 PM.