Monday, February 18, 2013
Chazhang's Affix Rolls by Item Level
The following comes courtesy of Chazhang, a fellow theorycrafter kind enough to forward me the results of some of his affix roll testing a week or two ago. This is only a snippet of the work he did, but I find it particularly relevant to 1.0.7 crafting. In short, it suggests that top-tier rolls should be much more common on the new crafted items due to them being ilvl 63.
Subscribe to:
Post Comments (Atom)
I see a broken link. It may not be shared.
ReplyDeleteUh oh, that's not good. It seems to be working on both computers I've used today, but I'll look into it.
DeleteYep - no link for me either. I'm really curious to see this as I want to understand exactly how affixes are rolled, and the ranges and probabilities for each roll.
ReplyDeleteI was thinking of testing some of your earlier predictions sometime this weekend. I noticed you grabbed data from diabloprogress - did you ever try crafting to get some sample data as well? That would remove any bias that may be inherent in diabloprogress data.
Ok, should actually be working now! I was using an unsupported filetype to post the image, apparently, but Blogspot wasn't notifying me of that.
ReplyDelete@J Mueller, I have not collected any data from crafting, but Chazhang did, which is why his input was so valuable. He confirmed some of my original findings, as well as adding in some twists that my method wouldn't have picked up on. This is one of those twists.
So it looks like Chazhang sorted all the affixes into different ilvls, but didn't look at the specific affix rolls and compare them to predicted values. Basically, in the 200 ammy rolls, you might expect to get some number of hits between 21-30 all res (or dex or vit or any of the other 30ish stats) based on your model. Do you know if anyone has done a comparison like that?
ReplyDeleteMy aim would be to determine the probability of rolling a specific affix in a specific range. If we can understand that, we can figure out the expected number of rolls we'd have to do on the archon crafts to be able to hit specific affix combos in specific ranges. That'd allow us to very accurately price specific affix combos.
Over the weekend I'm planning on rolling 100 resplendant warlord boots and charting the specific affix values on each roll. I picked that craft because 1) it's cheap to craft 2) only needs hell mats (so it's really cheap) and 3) has 5 affixes per roll. Let me know if you'd want to work together on figuring out the probability of hitting rolls in specific ranges for this craft.
In this clip, what Chazhang did was look at the occurrence of affix level and how it relates to item level. My work focused on figuring out the relative probability of different affixes rolling, but not the probabilities associated with the affix rolling within a particular range. Theoretically, though, if you combine our data you can get closer to a full picture of what the probabilities are of rolling good endgame items.
DeleteI considered keeping track of my Archon crafts, but it's just so much damn work. It takes long enough to sit there and click the "craft" button 100 times that actually entering the affixes on a spreadsheet was too painful for me to endure. Maybe one of these days.
Got it. Well, I'm curious and this problem has piqued my interest, so I think I'll plow through a hundred of those boots and record the affixes. If you're interested, I'll let you know what I find.
DeleteBy the way, I just read your "about me" and noticed that you're a physics grad student. Me too! I actually was wondering if you were in physics or math based on how you were analyzing your findings. Glad to see your analytical skills are being put to good use! Feel free to add me on d3 if you want. I'm WoodsSaxon#1770.
Nice! Glad to see some fellow physics grad students in D3. Most of my friends in the department quit playing over the last few months. I'll add you in-game next time I'm on!
DeleteSo I rolled those boots, and found a few interesting things:
ReplyDelete1) Some of the boots rolled with physical, cold, and fire resistances below 20. These affixes would be hardened (physical), nomadic (cold), and pure (poison). But these boots are iLvl 60, so based on your theory they should only roll with affixes of hermetic (physical), sheltering (cold), and untarnished (poison). It seems like the name of the affix does not constrain the available rolls.
2) One of the boots rolled with a fire resistance of 43. Based on d3own.com , this is an iLvl 61 roll, not an iLvl 60 roll. Either the iLvls of the affixes have changed since 1.04 (used on d3own.com ), or crafted items can roll above their iLvl.
Are you planning on updating your model of how the affixes roll? If so, I'd be happy to compare it to my data.
Keeping track of the rolls actually wasn't that bad. I used a dictation program and just read out the rolls aloud as they came, then parsed and analyzed the text file.
Ha, a dictation program is a pretty smart way to go, hadn't even considered that. Maybe I'll look into it.
DeleteAs far as the ilvl 60 boots go, it's hard to work that info into my theory. I don't know how low-level crafted items work because I think that system is a little archaic (i.e. not updated to coincide with farmed item rolls for the newer patch mechanics). I'm not really sure, but there's any numbers of things that could be going on with low-level crafted items, and they may or may not apply to the new ilvl 63 crafts or farmed items. I actually started keeping track of my rolls on the new ilvl 63 items for 11 different categories where there isn't any overlap in rolls--that way I can know exactly what affix level was rolled. I only just started, and I'm limited by how quickly I can find Demonic Essences, but in the next week or two I should have enough data to say something interesting, hopefully.
Nice - looking forward to it!
Delete