General Discussion - Socket Chances Revealed! - Forum

" demivion a écrit : it's true, you can't roll the same socket combination that's already on an item Citation still needed, I am quite sure I saw that happening ( 5s => 5s )	Posté par aldorus#6039 le 23 mai 2014 à 09:12:49 Citer ce post
And please no simulations / scripts results can be taken seriously, they may or may not actually behave the same as the game itself does.	Posté par aldorus#6039 le 23 mai 2014 à 09:14:14 Citer ce post
" aldorus a écrit : And please no simulations / scripts results can be taken seriously, they may or may not actually behave the same as the game itself does. well if Chris's numbers are applied when item drops and the code behaves as "if #sockets == #sockets; then continue", then why not?	Posté par CCR5#7770 le 23 mai 2014 à 09:16:42 Citer ce post
" CCR5 a écrit : " aldorus a écrit : And please no simulations / scripts results can be taken seriously, they may or may not actually behave the same as the game itself does. well if Chris's numbers are applied when item drops and the code behaves as "if #sockets == #sockets; then continue", then why not? But we are back to my first question. I need a reliable source confirming your hypothesis, that Jewellers orb cannot roll 5S into 5S because I am not convinced about that.	Posté par aldorus#6039 le 23 mai 2014 à 09:20:21 Citer ce post
" aldorus a écrit : " demivion a écrit : it's true, you can't roll the same socket combination that's already on an item Citation still needed, I am quite sure I saw that happening ( 5s => 5s ) Same, I'm pretty sure I've seen a 3S2L become a 3S0L and the like during socket attempts before. No. Calm down. Learn to enjoy losing.	Posté par b15h09#7812 le 23 mai 2014 à 09:22:09 Citer ce post
another question is real probabilities instead of rounded numbers (not that it mattter much) using 3268/9999 obviously works, but... and "tickets" numbers give different probabilities, given naive assumptions about what "tickets" are and what 20% would do Dernière édition par CCR5#7770, le 23 mai 2014 à 09:25:22	Posté par CCR5#7770 le 23 mai 2014 à 09:22:27 Citer ce post
I think I have a very large amount of Jewelers sitting back in my Standard stash, I guess its time to let them roll, for the science! :)	Posté par aldorus#6039 le 23 mai 2014 à 09:25:55 Citer ce post
" aldorus a écrit : And please no simulations / scripts results can be taken seriously, they may or may not actually behave the same as the game itself does. " chris_wilson on Reddit wrote The third number is the number of "tickets" it has, but I wouldn't read too deeply into these numbers. In general, the actual authoritative stuff is server-only (the "ots" file that you'll notice is thankfully missing). Edit: here are the correct numbers for a weapon that's able to get six sockets. 1 socket: 32.68% 2 sockets: 29.41% 3 sockets: 26.14% 4 sockets: 9.80% 5 sockets: 1.63% 6 sockets: 0.33% I'm not going to provide numbers for linking sockets, sorry. Chris used the word "tickets" and percentages he posted exactly match 100/(100+90+80+30+5+1)=0,326797386 and values for other numbers of sockets. I'd like to find a citation, that same number of sockets can't be rolled, too. But once I'm sure that this is true, I will be sure simulation should be taken seriously no matter how exactly GGG excludes "same roll". Dernière édition par SunnyRay#1519, le 23 mai 2014 à 09:28:07	Posté par SunnyRay#1519 le 23 mai 2014 à 09:27:43 Citer ce post
" CCR5 a écrit : another question is real probabilities instead of rounded numbers (not that it mattter much) using 3268/9999 obviously works, but... and "tickets" numbers give different probabilities, given naive assumptions about what "tickets" are and what 20% would do Real probabilities are (for 0%): 1s = 100/306 2s = 90/306 3s = 80/306 4s = 30/306 5s = 5/306 6s = 1/306 Source of ticket numbers: http://gyazo.com/c73d32096b0fdae0601e2b5429b46e3d These values match rounded numbers Chris posted. These are probabilities for a drop, when no number of sockets is "banned". Dernière édition par SunnyRay#1519, le 23 mai 2014 à 09:34:45	Posté par SunnyRay#1519 le 23 mai 2014 à 09:32:22 Citer ce post
" SunnyRay a écrit : ... I ran a simulation to get real numbers. Number of "jewellers used": 100 million. Here are the results: Percentage: 1 sockets: 30,192223% 2 sockets: 28,497462% 3 sockets: 26,504156% 4 sockets: 12,149895% 5 sockets: 2,207363% 6 sockets: 0,448902% Fraction: 1 sockets: 1/3,312111 2 sockets: 1/3,509084 3 sockets: 1/3,772993 4 sockets: 1/8,230524 5 sockets: 1/45,302925 6 sockets: 1/222,765773 An average amount of jewellers needed to get 6s is 223. ... Edit: fixed a minor bug in the script. Values are slightly more accurate now. These numbers are quite accurate (as you'd expect in a simulation this deep). I threw it in a spreadsheet (here) for an easier way to see what's happening. One thing I did was eliminate rolls once an item has 6 sockets, since that's now impossible (it used to be possible and I've actually had this spreadsheet for ages). Also, Chris posted these numbers in Feb of 2012. Nothing has changed to the base (non-quality) odds. How Fusings Work: http://www.pathofexile.com/forum/view-thread/38585/page/3#p1451934 IGN: TheHammer Dernière édition par TehHammer#0539, le 23 mai 2014 à 09:44:45	Posté par TehHammer#0539 le 23 mai 2014 à 09:33:27 Banni(e) Citer ce post