The Math of Fusings

"

Aelloon wrote:

"

sage2050 wrote:

because the math has been done, several times, independently, and the odds are already known.

Why do I keep hearing largely different numbers from different people then? What are the actual odds that are known?

because people keep trying to figure out the odds without knowing the mechanics. someone just bumped a thread with the actual values, go read that.

Well the odds for juwelers are fairly known, the odds for fusings aren't.

But the mechanic is probaply the same. If you want a 6s item the best way is to use juwelers on 1s items, since you can't get what you already have. So the best way is to figure out what linking-structure has the highest chance of appearance. The problem with that approach is that you won't have the choice of a lot of items, so getting a 6s from 1s items is faster than getting a 6l from any 6s item, but it doesn't matter.

If you have a nice rare and you want it to be a 5l you first have to roll juwelers on it. Whenever it might be a 5s you have a chance of getting 5l (propably the same chance as if you are using fusings). So using juwelers after getting the required number of sockets is propably not a good idea. After that you are down to fusings. And it doesn't really matter how the chances are, since you only have that one item, and even if it would be more effective to roll a 5l with an item that has is a 3l+2l you need to use what you have.

I think it was stated that a 5l is about 1/300, which is a pretty low chance :P, in the mentioned thread someone just took the odds for getting 6s and used it for links, I'm not sure if this is the same, since this would mean a 5l in about every 50 fusings, which seems a low number.

"

sage2050 wrote:

"

Aelloon wrote:

"

sage2050 wrote:

because the math has been done, several times, independently, and the odds are already known.

Why do I keep hearing largely different numbers from different people then? What are the actual odds that are known?

because people keep trying to figure out the odds without knowing the mechanics. someone just bumped a thread with the actual values, go read that.

The fusing mechanics are known, but the actual odds are not known. They're not nearly as good as the odds on jeweler's orbs, that's about all we know.

im going to propose the question to a guy who does gambling odds for a living.

"

sage2050 wrote:

im going to propose the question to a guy who does gambling odds for a living.

To know what the odds are you need either the devs to tell you what the odds are or a heckuva lot of data. I'm guessing your gambling-odds-for-a-living friend won't have either and won't be much help.

Heck, I used to have log information on 7k to 10k random 6S configurations (these were generated using a vendor glitch that was patched long ago, which would give fusing odds in it's purest form) and of those 7000+ configurations, I only saw one 6L, although I saw dozens of 5Ls. Seeing a 6L that seldom was bad luck, but even rolling close to 10k fusings the data is so limited you can't draw any conclusions (when it comes to 6Ls odds).

It seems to me that the chance to link sockets is actually lower on items with more sockets. I used a few fusings on 5-6 link items, and got 0-3 links each time, when with equal chance 6 socket item should have 3x more chance for 4 links.
It's entirely possible that the algorythm is fancy; like, each socket tries to link to a random socket, and links only if that socket is near.

You used a few fusings and are drawing conclusions based on that?!

The mechanics are already known. I quoted myself explaining them earlier in the thread. This thread does a good job of explaining how it works. Mind you that the odds stated in it are based on the jeweler's odds and aren't even close to right (he mentions this at the top in a disclaimer).

The odds are most definitely better with more sockets, as you can get extra chances to roll links that an item with fewer sockets wouldn't give you.

For example, a 4 socket item only gets one chance to roll links between all 4 sockets. But a 5 socket item that rolls a one socket group (no links) on the first socket still has 4 sockets left to roll in which it could roll links between the last 4 sockets and give you the desired 4L.

"

TehHammer wrote:

"

sage2050 wrote:

im going to propose the question to a guy who does gambling odds for a living.

To know what the odds are you need either the devs to tell you what the odds are or a heckuva lot of data. I'm guessing your gambling-odds-for-a-living friend won't have either and won't be much help.

Heck, I used to have log information on 7k to 10k random 6S configurations (these were generated using a vendor glitch that was patched long ago, which would give fusing odds in it's purest form) and of those 7000+ configurations, I only saw one 6L, although I saw dozens of 5Ls. Seeing a 6L that seldom was bad luck, but even rolling close to 10k fusings the data is so limited you can't draw any conclusions (when it comes to 6Ls odds).

to do any gambling odds you generally view at least 1 million iterations. if he can write a program to emulate rolling a item i'm sure he would do at least that many.

"

sage2050 wrote:

to do any gambling odds you generally view at least 1 million iterations. if he can write a program to emulate rolling a item i'm sure he would do at least that many.

How is his program going to choose when to roll a 5L and when to roll a 6L? You have to feed the program numbers, you have to know the odds to feed it accurate numbers.

what? no you don't. if you know how the roll works (we do) you can write an algorithm for it. it's just a series of random selection.