I crafted 301 6-property chest armors, and wrote down the results of every craft. Not only that, I devised a formula to compare them directly to each other no matter what stats exist on them based on a series of weights that so far has seemed to work out quite well for me.
Not only THAT, but I wrote down every price I was able to sell them at, and hope to soon be able to come up with a way to automatically price them according to past sales, adjusted slightly due to the lowered costs of production as essences go down in price and the less demand over time that is to be expected.
As of two full recorded crafts now, every good piece I noticed while browsing through them manually has appeared near the top of a sorted "average property value" list. That's good enough for me even though the formula could use some work I'm sure, to account for good goldfind/radius combos and such.
Copy the spreadsheet and put your own values in and see where they match up to see if it works as well as I think it does!
If you want more info on how I came up with the formula and why it makes sense to me, here's a TL;DR reddit post (reddit.com) I made on the subject. Up-vote it if you found it useful!
10500 exquisite essences, average 1500 each: 15750000 gold
1500 iridescent tears, average 100 each: 150000 gold
3000 tomes of secrets, average 1000 each: 3000000 gold
300 crafting cost, 91494 gold each: 27448200 gold
Total crafting costs: 46,348,200 gold
I updated the spreadsheet with all current sales tallied, I sold 53,875,000 gold worth before blizz's 15% cut, after 15% cut it's 45,793,750 gold. I've just about broken even with plenty of good chests still left to sell. I'm wearing two of the best pieces I crafted, and just recently crafted 300 belts that I'm currently logging, and have 7 million total gold. I "wasted" about 20 million or so gold on crafting 150 helms of which less than 10 sold but otherwise took up auction spots and I gave up on after making back all 20 million on them. I still have like 140 helms cluttering up characters, stash and mailbox. On that note, I documented all the helms I crafted so If anyone wants to see what I have and would like to make an offer on taking all of them, I'd be interested in that.
edit: by the way there are a good 20 or so undocumented sales because someone is helping me sell the lower items for 75-100k each while I sell the better ones in my spots.
- 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.
I think I'm most interested in the affix combinations and limitations. I'm loving this data.
The most interesting things I've noted are the probabilities with stat rolls. For example: I never realized that the resists were a positively skewed bell curve (it's very very likely to roll high resists as opposed to rolling low resists) and I didn't realize that there were two different affix strengths that the primary stats fall into.
Maybe this info will help you in identifying the rarest of your chests:
I think I'm most interested in the affix combinations and limitations. I'm loving this data.
The most interesting things I've noted are the probabilities with stat rolls. For example: I never realized that the resists were a positively skewed bell curve (it's very very likely to roll high resists as opposed to rolling low resists) and I didn't realize that there were two different affix strengths that the primary stats fall into.
Maybe this info will help you in identifying the rarest of your chests:
Keep up the good work and give me more data to work with!
I would love access to the Helm spreadsheet you've been keeping. Depending on the rolls, I might be interested in an upgrade to my own..
If you can make use of it, awesome. I have a spreadsheet somewhere where I tried to assign values to everything and whatnot, but I went a little crazy adding different combinations and stuff and I found it kinda useless. There's just too much to consider like what the value of something that has magic find but no socket compared to magic find and a socket, etc. This is just the raw data from 150 helms. I sold maybe 10-20 of these but if you want any of them, I can see if I have it and sell it to you for cheap.
edit: I think the reason helms sell slower is because every class but DH and barb has a class-specific helm, which max out at 62 so you're not doing any amazing favors by crafting ilvl 62 generic helms. WDs and wizards in particular rely on mana regen and arcane power from crit, which these helms don't provide. Also, everyone thinks sockets are necessary for a MF or GF build (understandably), so those nice combos are useless unless a socket is there, and a socket is only on about 1/3 of the crafted helms it seems.
Just sent an access request for that spreadsheet. I'll start looking into trends and such. I'm not necessarily interested in what combinations of affixes happen, just the affix probabilities themselves.
10500 exquisite essences, average 1500 each: 15750000 gold
1500 iridescent tears, average 100 each: 150000 gold
3000 tomes of secrets, average 1000 each: 3000000 gold
300 crafting cost, 91494 gold each: 27448200 gold
Total crafting costs: 46,348,200 gold
I updated the spreadsheet with all current sales tallied, I sold 53,875,000 gold worth before blizz's 15% cut, after 15% cut it's 45,793,750 gold. I've just about broken even with plenty of good chests still left to sell. I'm wearing two of the best pieces I crafted, and just recently crafted 300 belts that I'm currently logging, and have 7 million total gold. I "wasted" about 20 million or so gold on crafting 150 helms of which less than 10 sold but otherwise took up auction spots and I gave up on after making back all 20 million on them. I still have like 140 helms cluttering up characters, stash and mailbox. On that note, I documented all the helms I crafted so If anyone wants to see what I have and would like to make an offer on taking all of them, I'd be interested in that.
edit: by the way there are a good 20 or so undocumented sales because someone is helping me sell the lower items for 75-100k each while I sell the better ones in my spots.
Another thing to consider is the value of the plan which is around 6 million gold. Although, if you can turn a profit consistently then the plan's value is just a sunk cost that's effect diminishes every day. The main thing is turning a profit.
My thought is that if a crafting player can break even over the 700-1200 (I've not seen a large enough sample to justify this range however) crafts it takes to roll a chest with 230+ vit, 65+ all res, 100+ main stat, 3 sockets, and some nice extras like high armor,life%,MF, etc, then that's where the big profit will come in.
10500 exquisite essences, average 1500 each: 15750000 gold
1500 iridescent tears, average 100 each: 150000 gold
3000 tomes of secrets, average 1000 each: 3000000 gold
300 crafting cost, 91494 gold each: 27448200 gold
Total crafting costs: 46,348,200 gold
I updated the spreadsheet with all current sales tallied, I sold 53,875,000 gold worth before blizz's 15% cut, after 15% cut it's 45,793,750 gold. I've just about broken even with plenty of good chests still left to sell. I'm wearing two of the best pieces I crafted, and just recently crafted 300 belts that I'm currently logging, and have 7 million total gold. I "wasted" about 20 million or so gold on crafting 150 helms of which less than 10 sold but otherwise took up auction spots and I gave up on after making back all 20 million on them. I still have like 140 helms cluttering up characters, stash and mailbox. On that note, I documented all the helms I crafted so If anyone wants to see what I have and would like to make an offer on taking all of them, I'd be interested in that.
edit: by the way there are a good 20 or so undocumented sales because someone is helping me sell the lower items for 75-100k each while I sell the better ones in my spots.
Another thing to consider is the value of the plan which is around 6 million gold. Although, if you can turn a profit consistently then the plan's value is just a sunk cost that's effect diminishes every day. The main thing is turning a profit.
My thought is that if a crafting player can break even over the 700-1200 (I've not seen a large enough sample to justify this range however) crafts it takes to roll a chest with 230+ vit, 65+ all res, 100+ main stat, 3 sockets, and some nice extras like high armor,life%,MF, etc, then that's where the big profit will come in.
Absolutely. I didn't include the plan costs just because they've been fluctuating so much, but that was something to consider too. FWIW, despite what my recorded costs and profit were like I mentioned before, I started all this only off of a couple decent drops, a dagger that I sold for about 20 million and a chest I was able to randomly snag for 3 bucks somehow that was worth 15 million.
I do wish it was a little more reliable to make back your money doing it, I can kinda imagine crafting 150 and if some those top chests on my spreadsheet hadn't been crafted, it would be much harder to break even.
I think I'm most interested in the affix combinations and limitations. I'm loving this data.
The most interesting things I've noted are the probabilities with stat rolls. For example: I never realized that the resists were a positively skewed bell curve (it's very very likely to roll high resists as opposed to rolling low resists) and I didn't realize that there were two different affix strengths that the primary stats fall into.
Maybe this info will help you in identifying the rarest of your chests:
Keep up the good work and give me more data to work with!
I would love access to the Helm spreadsheet you've been keeping. Depending on the rolls, I might be interested in an upgrade to my own..
I hate to be pedantic, but if a probability distribution has its bulk toward higher values, then it has negative skewness, *not* positive skewness. It is also said to be negatively skewed or left skewed.
I think I'm most interested in the affix combinations and limitations. I'm loving this data.
The most interesting things I've noted are the probabilities with stat rolls. For example: I never realized that the resists were a positively skewed bell curve (it's very very likely to roll high resists as opposed to rolling low resists) and I didn't realize that there were two different affix strengths that the primary stats fall into.
Maybe this info will help you in identifying the rarest of your chests:
Keep up the good work and give me more data to work with!
I would love access to the Helm spreadsheet you've been keeping. Depending on the rolls, I might be interested in an upgrade to my own..
I hate to be pedantic, but if a probability distribution has its bulk toward higher values, then it has negative skewness, *not* positive skewness. It is also said to be negatively skewed or left skewed.
Doh... I suppose it's been a while since I studied actual statistics. So yes. Negatively skewed.
I think I'm most interested in the affix combinations and limitations. I'm loving this data.
The most interesting things I've noted are the probabilities with stat rolls. For example: I never realized that the resists were a positively skewed bell curve (it's very very likely to roll high resists as opposed to rolling low resists) and I didn't realize that there were two different affix strengths that the primary stats fall into.
Maybe this info will help you in identifying the rarest of your chests:
Keep up the good work and give me more data to work with!
I would love access to the Helm spreadsheet you've been keeping. Depending on the rolls, I might be interested in an upgrade to my own..
I hate to be pedantic, but if a probability distribution has its bulk toward higher values, then it has negative skewness, *not* positive skewness. It is also said to be negatively skewed or left skewed.
Doh... I suppose it's been a while since I studied actual statistics. So yes. Negatively skewed.
I make the mistake all the time myself. Just busting your chops.
I started a similar exercise (isn't having something really boring to avoid at work great!) trying to reverse engineer the "item roll" process and noted similar trends.
I was trying to work out how far down each affix type an item would roll - the floor on most seem to be level 51/52 affixes (with the exception of pickup radius).
Then I tried to work out if it was 1 roll or 2 roll system - all possible affixes thrown in a big pool and rolled or roll "group", then roll within the group. Its clearly a lot more complex than that as you note there is clearly a level of weighting towards the highest affix in any given group.
At this point I was pretty much beyond my analytical capability so I'd be interested in suggestions on how to move forward (with the goal being to be able to work out the proportion of items with a given set of stats or better - a "rarity" score as it were).
I'm not really a statistician, but I think the graphs do support the premise that there are a few different stackable affixes a chest (or any piece of gear) can roll.
It lists several different combinations between Vitality and Dexterity rolls.
If you roll the best combination of affixes you can between Vitality and Dexterity, you come up with a maximum possible 300 Vit and 200 Dex: http://d3db.com/item...5994,3561578828
I believe those statistics were datamined from the game, but I could be wrong. Assuming we believe them, there is probably a relatively even distribution of possible affix combinations between each of them.
I'm willing to bet that the game randomly decides what affixes the item has based on a simple "pick X affix combos and roll the stats for each applicable combo" I'm also willing to bet the game won't allow you to roll more than 2 different combinations of any given stat (no more than 2 vit rolls, 2 dex rolls, etc.) and probably has a set probability of rolling each combination.
Best possible roll for a Vit chest would be a combination of 2 different Vitality rolls including the "pure" vitality roll (170-200) as well as a vitality/primary stat roll.
This is extremely important as it implies that a magic item can't have more than 2 different affix rolls, but can result in 3 different stats. If you can find a magic chest with more than 200 vitality AND more than 100 of any given primary stat, it would invalidate that claim as doing so would actually require more than 2 different affix rolls.
***EDIT***
And just looking around on the AH, I was unable to find any Magic items that had those stats. I suppose the correct word for these "rolls" is a property which can be a combination of different affixes.
We can confirm this hypothesis with the affix quantity distribution curve with relation to primary stats on the chests that Tybrone crafted as it simply says that it's much more likely to only roll a single set of vitality affixes than it is to roll double vitality affixes. Given the minimum and maximum possible ranges for those affixes, we can identify a "deadzone" so-to-speak of a range of vitality (or any stat really) that simply cannot occur on a given piece with given ilvl.
What would truly be interesting is identifying what affixes were rolled on a piece of gear. I think we could do so if we ran through each piece of gear and basically just confirmed or denied that a given combination of affixes is possible for that piece of gear. It is only then that we could determine what the probability of rolling a particular property is and then coming up with a distribution curve for a particular affix.
From there, it's just a matter of figuring out the probability of finding that gear and, assuming it's a good combination of properties, valuing it accordingly.
****EDIT #2****
Figuring out what properties a piece of armor had would largely be trial and error but mostly process of elimination. It doesn't look like that website I linked is entirely accurate... but regardless, it's a lot to work with.
Rollback Post to RevisionRollBack
To post a comment, please login or register a new account.
Not only THAT, but I wrote down every price I was able to sell them at, and hope to soon be able to come up with a way to automatically price them according to past sales, adjusted slightly due to the lowered costs of production as essences go down in price and the less demand over time that is to be expected.
As of two full recorded crafts now, every good piece I noticed while browsing through them manually has appeared near the top of a sorted "average property value" list. That's good enough for me even though the formula could use some work I'm sure, to account for good goldfind/radius combos and such.
Copy the spreadsheet and put your own values in and see where they match up to see if it works as well as I think it does!
Here's the spreadsheet (google docs)
If you want more info on how I came up with the formula and why it makes sense to me, here's a TL;DR reddit post (reddit.com) I made on the subject. Up-vote it if you found it useful!
My I suggest freezing the top 3 rows so that one could easily see which column represents which stat when reading lower on the spreadsheet.
^this^
but otherwise gj
I'll update the spreadsheet with a new sheet mentioning the profit now, and get back to you
Forgot to re-add that earlier, gonna add it now
10500 exquisite essences, average 1500 each: 15750000 gold
1500 iridescent tears, average 100 each: 150000 gold
3000 tomes of secrets, average 1000 each: 3000000 gold
300 crafting cost, 91494 gold each: 27448200 gold
Total crafting costs: 46,348,200 gold
I updated the spreadsheet with all current sales tallied, I sold 53,875,000 gold worth before blizz's 15% cut, after 15% cut it's 45,793,750 gold. I've just about broken even with plenty of good chests still left to sell. I'm wearing two of the best pieces I crafted, and just recently crafted 300 belts that I'm currently logging, and have 7 million total gold. I "wasted" about 20 million or so gold on crafting 150 helms of which less than 10 sold but otherwise took up auction spots and I gave up on after making back all 20 million on them. I still have like 140 helms cluttering up characters, stash and mailbox. On that note, I documented all the helms I crafted so If anyone wants to see what I have and would like to make an offer on taking all of them, I'd be interested in that.
edit: by the way there are a good 20 or so undocumented sales because someone is helping me sell the lower items for 75-100k each while I sell the better ones in my spots.
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.
The most interesting things I've noted are the probabilities with stat rolls. For example: I never realized that the resists were a positively skewed bell curve (it's very very likely to roll high resists as opposed to rolling low resists) and I didn't realize that there were two different affix strengths that the primary stats fall into.
Maybe this info will help you in identifying the rarest of your chests:
https://docs.google....bEU0ZWgyX2NJSmc
Keep up the good work and give me more data to work with!
I would love access to the Helm spreadsheet you've been keeping. Depending on the rolls, I might be interested in an upgrade to my own..
If you can make use of it, awesome. I have a spreadsheet somewhere where I tried to assign values to everything and whatnot, but I went a little crazy adding different combinations and stuff and I found it kinda useless. There's just too much to consider like what the value of something that has magic find but no socket compared to magic find and a socket, etc. This is just the raw data from 150 helms. I sold maybe 10-20 of these but if you want any of them, I can see if I have it and sell it to you for cheap.
https://docs.google....TnF4bWNkd2RTR0E
edit: I think the reason helms sell slower is because every class but DH and barb has a class-specific helm, which max out at 62 so you're not doing any amazing favors by crafting ilvl 62 generic helms. WDs and wizards in particular rely on mana regen and arcane power from crit, which these helms don't provide. Also, everyone thinks sockets are necessary for a MF or GF build (understandably), so those nice combos are useless unless a socket is there, and a socket is only on about 1/3 of the crafted helms it seems.
Another thing to consider is the value of the plan which is around 6 million gold. Although, if you can turn a profit consistently then the plan's value is just a sunk cost that's effect diminishes every day. The main thing is turning a profit.
My thought is that if a crafting player can break even over the 700-1200 (I've not seen a large enough sample to justify this range however) crafts it takes to roll a chest with 230+ vit, 65+ all res, 100+ main stat, 3 sockets, and some nice extras like high armor,life%,MF, etc, then that's where the big profit will come in.
Absolutely. I didn't include the plan costs just because they've been fluctuating so much, but that was something to consider too. FWIW, despite what my recorded costs and profit were like I mentioned before, I started all this only off of a couple decent drops, a dagger that I sold for about 20 million and a chest I was able to randomly snag for 3 bucks somehow that was worth 15 million.
I do wish it was a little more reliable to make back your money doing it, I can kinda imagine crafting 150 and if some those top chests on my spreadsheet hadn't been crafted, it would be much harder to break even.
I hate to be pedantic, but if a probability distribution has its bulk toward higher values, then it has negative skewness, *not* positive skewness. It is also said to be negatively skewed or left skewed.
Doh... I suppose it's been a while since I studied actual statistics. So yes. Negatively skewed.
I make the mistake all the time myself. Just busting your chops.
I'm not really a statistician, but I think the graphs do support the premise that there are a few different stackable affixes a chest (or any piece of gear) can roll.
Take this page for example: http://d3db.com/item/i/archon-armor
It lists several different combinations between Vitality and Dexterity rolls.
If you roll the best combination of affixes you can between Vitality and Dexterity, you come up with a maximum possible 300 Vit and 200 Dex: http://d3db.com/item...5994,3561578828
I believe those statistics were datamined from the game, but I could be wrong. Assuming we believe them, there is probably a relatively even distribution of possible affix combinations between each of them.
I'm willing to bet that the game randomly decides what affixes the item has based on a simple "pick X affix combos and roll the stats for each applicable combo" I'm also willing to bet the game won't allow you to roll more than 2 different combinations of any given stat (no more than 2 vit rolls, 2 dex rolls, etc.) and probably has a set probability of rolling each combination.
Best possible roll for a Vit chest would be a combination of 2 different Vitality rolls including the "pure" vitality roll (170-200) as well as a vitality/primary stat roll.
This is extremely important as it implies that a magic item can't have more than 2 different affix rolls, but can result in 3 different stats. If you can find a magic chest with more than 200 vitality AND more than 100 of any given primary stat, it would invalidate that claim as doing so would actually require more than 2 different affix rolls.
***EDIT***
And just looking around on the AH, I was unable to find any Magic items that had those stats. I suppose the correct word for these "rolls" is a property which can be a combination of different affixes.
We can confirm this hypothesis with the affix quantity distribution curve with relation to primary stats on the chests that Tybrone crafted as it simply says that it's much more likely to only roll a single set of vitality affixes than it is to roll double vitality affixes. Given the minimum and maximum possible ranges for those affixes, we can identify a "deadzone" so-to-speak of a range of vitality (or any stat really) that simply cannot occur on a given piece with given ilvl.
What would truly be interesting is identifying what affixes were rolled on a piece of gear. I think we could do so if we ran through each piece of gear and basically just confirmed or denied that a given combination of affixes is possible for that piece of gear. It is only then that we could determine what the probability of rolling a particular property is and then coming up with a distribution curve for a particular affix.
From there, it's just a matter of figuring out the probability of finding that gear and, assuming it's a good combination of properties, valuing it accordingly.
****EDIT #2****
Figuring out what properties a piece of armor had would largely be trial and error but mostly process of elimination. It doesn't look like that website I linked is entirely accurate... but regardless, it's a lot to work with.