So a few month ago I was thinking it would be fun to have some data on torch farming. The easiest way to get that was to do a 100 Hellfire Torch runs. It took me two month and it was very expensive as i didn't farm keys but trade it all and sold all the torch i got. I'm still trying to calculate how many Ist it cost me in total but i can tell you (as expected) it's not lucrative at all. I got a few good ones and got to upgrade some of my existing torch along the way.
Here are the result I got :
Amazon : 19
Assassin : 18
Barbarian : 15
Druid : 16
Necromancer : 13
Paladin : 12
Sorceress : 7
PS : All my runs happend on EU server with a Paladin Smiter lvl 85
So a few month ago I was thinking it would be fun to have some data on torch farming. The easiest way to get that was to do a 100 Hellfire Torch runs. It took me two month and it was very expensive as i didn't farm keys but trade it all and sold all the torch i got. I'm still trying to calculate how many Ist it cost me in total but i can tell you (as expected) it's not lucrative at all. I got a few good ones and got to upgrade some of my existing torch along the way.
Here are the result I got :
Amazon : 19
Assassin : 18
Barbarian : 15
Druid : 16
Necromancer : 13
Paladin : 12
Sorceress : 7
PS : All my runs happend on EU server with a Paladin Smiter lvl 85
On one hand, I kinda think "Why would they even rig this, what's the point? Surely just bad luck!"
Rigging for the devs is probably as easy as editing a text file with a proprietary extension. I suspect it's being done far more often than we might expect.
What I would like to know is where are all these unid Annihilus' coming from? Anyone can get one for a single HR.
These results don't Shout "1/7" chance for each character class. This seems similar to the latent charms, and Gheed's Fortune. I've yet to see one, but I've picked up many (not 100, but probably 20ish) latent charms.
In just 100 runs, getting exactly 14 torches per character is equally unlikely than this result. On 1000 runs, you will get closer to a 143 average and after 10,000 runs, RNG will even out almost completely.
In just 100 runs, getting exactly 14 torches per character is equally unlikely than this result. On 1000 runs, you will get closer to a 143 average and after 10,000 runs, RNG will even out almost completely.
Sounds like a challenge, but not for me. I understand how "odds" work, but even in the trade forum, very few sorc or warlock torches (which lots of people like to play), and LOTS of druid and necromancer (which nowhere near as many play), specifically on ladder. Something doesn't add up...but I'm a skeptic!
...but even in the trade forum, very few sorc or warlock torches (which lots of people like to play), and LOTS of druid and necromancer (which nowhere near as many play), specifically on ladder. Something doesn't add up...but I'm a skeptic!
But you have given the explanation yourself:
1. There are less Sorc torches for sale because the demand is much higher, since the Sorc is much more popular than Dudu and Necro but the amount of torches dropping is the same for all classes.
2. For Lock the mismatch is extreme. Everyone plays him (high demand) but there are no stocks of Lock torches and if you drop one, you sell it within hours or even minutes (low offer). That does not mean that they don't drop or drop less often.
...but even in the trade forum, very few sorc or warlock torches (which lots of people like to play), and LOTS of druid and necromancer (which nowhere near as many play), specifically on ladder. Something doesn't add up...but I'm a skeptic!
But you have given the explanation yourself:
1. There are less Sorc torches for sale because the demand is much higher, since the Sorc is much more popular than Dudu and Necro but the amount of torches dropping is the same for all classes.
2. For Lock the mismatch is extreme. Everyone plays him (high demand) but there are no stocks of Lock torches and if you drop one, you sell it within hours or even minutes (low offer). That does not mean that they don't drop or drop less often.
I wasn't only referring to torches for sale, but torches in general. I did a quick price check search for each class to find out how many sold, then searched the trades for active for sale, then totalled them to find how many have been listed so far (S14):
Now I realize this isn't anwhere near perfect statistics, I'm sure not everyone sells every torch they come across, some sales aren't reported, Warlock is the desired class at the moment (high demand), etc, but with a supposed 1/8 (12.5%) chance to get any class torch, these numbers don't reflect anywhere near that. Can anyone dig into the code to determine if the odds are in fact 1/8?
I wasn't only referring to torches for sale, but torches in general. I did a quick price check search for each class to find out how many sold, then searched the trades for active for sale, then totalled them to find how many have been listed so far (S14):
How are you referencing/researching torches "in general"? If you searched here, then torches "in general" must equal torches that are or were for sale. People don't make trade posts to show off their finds. We have the forums for that, but that isn't tracked.
Botters bot with sorc's. Botters also farm torches. Botters keep their sorc torches for their bots.
Sorcs are popular. Players play with more than one sorc. I also farm torches occasionally. I have 5 sorcs, 3 warlocks, 1 necro, and 1 druid. Meaning when I do put torches up for sale, necro and druid ones disproportionately go up for sale compared to sorc and warlock.
Can anyone dig into the code to determine if the odds are in fact 1/8?
All unique items are defined in uniqueitems.txt (you can see what this file looks like for RoTW 3.2 here, and for LoD 3.2 here). Each line in the file defines one unique item (Hellfire Torch is defined on line 403). It's a tab-separated text file so it's best to view it in an editor that can properly display it as columns.
Each unique item can have up to 12 properties, defined by columns propX, parX, minX and maxX. For Hellfire Torch, the +3 to a class' skill levels is the first property (all attributes is 2nd, all resistances is 3rd, light radius is 4th, Firestorm ctc is 5th, Hydra charges is 6th). In the uniqueitems.txt file we can see that prop1 = "randclassskill", par1 = "" (empty), min1 = 0 and max1 = 7. So we can see that the class roll is a random value in the range [0, 7] (inclusive). That means there are 8 values (0, 1, 2, 3, 4, 5, 6, and 7) that it can roll. That's it. Looking at the LoD version of the file, everything looks the same except max1 = 6, because there are only 7 classes.
There's functionally no difference between the class roll and any other type of roll - there's a min value, and a max value.
I'm in CET (Central European Time)
My profile says Ladder, but I play bothLadder and Non-Ladder.
My profile says Softcore, but I play bothSoftcore and Hardcore.
My Holy Grail Tracker is for Offline Hardcore.
Holy Grail Death Count: 1
List trades and earn trust
Level up as you post
Give and Receive Likes
Discreet Notifications
Avatar & Profile customisation
Private Messaging
Bookmark Pages & Posts
Use Drafts & Revisions
Attach Images
Squelches this trader, making their offers on your trades invisible to anyone except themselves. Their trade listings will also be hidden from your marketplace.
Your squelch list can be managed and edited from Top right dropdown > Account Prefs > Squelches.
Results of 100 Hellfire Torch runs 
Hide ads
brownSys, 
departure, 
jrlafrance, 
klmklm27, 
ShadowHeart, 
Yahoo [Bot] and 355 guests.
La_Brick_48
7