Massive Refine Failure

Aubree - Dreamweaver · December 2014

freygin wrote: »
About the chances for refining from pwpedia using mirage only :
Refine  Mirage Only 	 
+1 	50% 	 
+2 	30%  
+3 	30% 	 
+4 	30% 	 
+5 	30%  
to get +5 from zero the chance would be 50% * 30% * 30% * 30% *30% = 0.00405 , the average needed mirages then would be 100000/405 = 246.9~247 mirages.

Is that right ?

I tried a couple times using Asterelle's refining simulator the range of mirages used varied from 20 to 990, a million tries probably would get an average close to 247 mirages.

Also a chance like 50% should always be 50% no matter how it's randomized.

Sounds about right for what I reported my refining was. +6 two times with 511 mirages. That is not counting how many times I got 4 or 5 before failing. I used no refining aides. When I see the window fail 10x in a row I know a lucky streak is coming soon. b:laugh

A friend of mine got +8 with no fails at all but on the remainder of his gear could barely get +7. This was at the same time I was refining.

Deadalus - Harshlands · December 2014

Poopinpanto - Heavens Tear wrote: »

I only use mirages to plus stank gear or to get a start on gear butusually just go ahead and spend coin to + with orbs.But even that's been cheap lately due to being able to get +6 and +7 orbs from RA primal.

Can we get on this again? what do you mean with RA primal? do i miss something? b:shutup

Aubree - Dreamweaver · December 2014

Deadalus - Harshlands wrote: »

Can we get on this again? what do you mean with RA primal? do i miss something? b:shutup

Pretty sure they mean the chest rewards for achieving primal sky levels. You get a 6 and 7 star orb.

opkossy · December 2014

Deadalus - Harshlands wrote: »

Can we get on this again? what do you mean with RA primal? do i miss something? b:shutup

He's talking about http://www.pwdatabase.com/pwi/quest/30572 and other such glowy chest rewards.

Deadalus - Harshlands · December 2014

Oh that... i thought i missed out on something b:surrender

c44102 · December 2014

Ive refined a few things to +5 since the update with out using refining aids. Like normal somethings very easy. But like normal somethings very hard. The hard stuff seems alot harder to get to +3 then normal. I have tryed different places but i still have got some stuff to +6 and +7 without orbs. Overall cost still cheaper then orbs. I did try with a G16 wep and 500 mirages ( i know thats only 100 trys ) Best i seen was a +3 that was very odd normaly in that many trys i would seen atleast a +4.

Azura - Lost City · December 2014

I don't think there has been any changes. I just +9 another necklace and I haven't noticed refining becoming harder to do.

riffraffz · December 2014

OPKossy wrote: »

For the last time (oh who am I kidding? People are always gonna whine about nerfs when they run into a string of bad luck) it has not been changed. Has never been changed.

We had someone pull the data straight out in comparison of the ancient values and what was current at the time and there was zero change. We've had MULTIPLE people do that MULTIPLE times when folks were crying nerf.

There. Is. No. Nerf.

**** luck happens. It's part of the system. Just a few hours ago I had two pieces of the exact same armor. Same sockets, stat bonuses, etc on both. One of them I got to friggin +9 with less than 300 mirages and about 20ish refine aids. The other one I had to settle for +5 after burning through ~800 mirages and accompanying refine aids. It's all down to luck.

My advice? Just save this post as i think you'll be needing it often in the future b:chuckle

SylenThunder - Twilight Temple · December 2014

OPKossy wrote: »

-snip-
Edit: Man that came off as super aggressive upon a reread. Oh well, not changing it just to find a way to make it sound less mean.

Funny, it sounds just like I would have posted it, with one exception....
I'm one of the people that debunks most of these claims. I've proved it for TT many times over. Though I haven't looked at the code on this most recent expansion, they haven't changed the refine rates in 7 years, I really doubt they did it this time. It would require a massive code change as each item has it's own % chance for refines/rewards.

Salari - Raging Tide wrote: »

No you really missed the point as any algorithm coded in the game has nothing to do with the client/server files or percentages. A simple check on any number of values in the game server wide could alter what the outcome would be. A simple inventory system can do this. An example would be say packs for example. If (number_of_vit_stones_from_packs_lasthour => somenumber) then ignore(roll); return_token. Another example is getting stats on armor that does not exist in the server/client files. Without access to the source code no one really knows.

Actually, there are some of us that do. I've done it before.
Gear for example, has a set percentage for each refine level to succeed. It rolls that against the engine's randomization, (which sucks, but isn't broken), and then you get the hit. It's pretty much just like 90% of randomization in games out there.

Also, depending on where you are, there are a few "tweaks" you can utilize to slightly increase your odds, because of the way the NPC's function and how the randomization works.

Your code of "If (number_of_vit_stones_from_packs_lasthour => somenumber) then ignore(roll); return_token.", is completely silly. The dice is rolled on each individual roll. There are no averages that play into it, and this aging title in no way has the capability of keeping track of that kind of information. (Though the new engine may, they would not waste time completely re-writing the code to support it. They have a hard enough time releasing the new data in the time constraints forced on them by management. I covered that in another post as well.)

If you want to get your hands on the source code, and the tools to look into it, it's too easy to even mention here. Before the expansion went live, we had our private servers running at the same version as the live client. It's not terribly hard, and Wanmei has been leaking the data faster the past year or so.

Salari - Raging Tide · December 2014

You are confusing game files with source code. The game files are not the source code. Yes there are several flavors of the game that can be downloaded, but is not the source code to the game. The tools you mention are for editing the game files, like the shop, char files npc's instances ect not the source code itself.

Your code of "If (number_of_vit_stones_from_packs_lasthour => somenumber) then ignore(roll); return_token.", is completely silly. The dice is rolled on each individual roll. There are no averages that play into it, and this aging title in no way has the capability of keeping track of that kind of information. (Though the new engine may, they would not waste time completely re-writing the code to support it. They have a hard enough time releasing the new data in the time constraints forced on them by management. I covered that in another post as well.)

lol at that, just tells me you have no idea about what a game engine can and can not keep up with. Older engines that were more advanced than the one pw is based on kept up with that information and a lot more

Auldwulf - Dreamweaver · December 2014

My take on it is this; I sincerely believe that when a character is made, there is a random % of "luck" given to that character.

Because I know a few that seem to have amazing luck at refining, opening packs, ect., while most of us seem to just get hosed.

Face it, it would be in the company's best interest to do so. Give most crappy luck, then a majority of those will buy zen, to get the refining material they need. Give a few good luck, to give the rest of us hope, and incentive to keep the cycle going. (Half said jokingly). b:chuckle

filipenogueras · December 2014

Looks like with the update the places, the timing and sequence changed/reseted. I had trouble with +3 and +4 too.

Before expansion I got all my set +7, and +6 was very easy.

I had to find a new sequence, timing and place to refine.

Trying the old way on new Expansion wasted me 2000 mirages for 4x +5.

Once I found a new good combination of place, timing and sequence, it took me 600 mirages for 4x +6 and 10x +5 (not mention all the +3). Ps: I guess it can get even better

Bosk - Sanctuary · December 2014

I think there is perhaps some confusion between the CHANCE of successfully refining an item to a particular value, and the NUMBER of mirages expected to be used in doing so.

For example, the chance of getting +2 is about 1 in 6.7 (50% x 30% = 15%) however the number of Mirages that will be required to get a piece of gear to +2 will be 10, 20 or 50, depending on the type of gear. Even gear that only needs 1 mirage per refine attempt, the average is 10 not 6.7.

Objective:  To Refine to +N (N = 1 to 5 for this example)			
Table of AVERAGE number of Mirages required based on 1x, 2x or 5x per try			
			
N	     1x		     2x		     5x		Odds: 1 in
+1	    2.00	    4.00	   10.00	  2.00
+2	   10.00	   20.00	   50.00	  6.67
+3	   36.67	   73.33	  183.33	 22.22
+4	  125.56	  251.11	  627.78	 74.07
+5	  421.85	  843.70	2,109.26	246.91

Why?

Well, consider what happens while refining. A typical pattern to get to +2 would be:

+0: START
+0: FAIL (50% chance)
+1: SUCCESS (50% chance)
+0: FAIL (70% chance)
+0: FAIL (50% chance)
+1: SUCCESS (50% chance)
+0: FAIL (70% chance)
+0: FAIL (50% chance)
+1: SUCCESS (50% chance)
+2: SUCCESS (30% chance)

Number of times the item went from +0 to +0: 3
Number of times the item went from +0 to +1: 3
Therefore, there were 50% fails and 50% successes at +0 (which is the stated success rate).

Number of times the item went from +1 to +0: 2
Number of times the item went from +1 to +2: 1
Therefore, there were 67% fails and 33% successes at +1 (which is just slightly off the stated success rate).

The total number of refines required was 9.

Since the actual chance of a fail on +1 is 70%, on average there will be slightly more mirages required: the final expectation is 10.

With 34,170 Mirages, one would expect to see:				
				
Outcome		Results		Used		Successes	1 in 
 0=>0		10,000 (50%)			
 0=>1		10,000 (50%)	20,000		10,000		  2.00
 1=>0		 7,000 (70%)			
 1=>2		 3,000 (30%)	30,000		 3,000		 10.00
 2=>0		 2,100 (70%)			
 2=>3		   900 (30%)	33,000		   900		 36.67
 3=>0		   630 (70%)			
 3=>4		   270 (30%)	33,900		   270		125.56
 4=>0		   189 (70%)			
 4=>5		    81 (30%)	34,170		    81		421.85

And all this means ... ?

It means that the expected number of refine attempts is much higher than the odds of getting the desired refine level. Consider +4. The "odds" of getting to +4 are 1 in 74 or so, but that odds is based on starting at 0 and clicking on refine with no aides 4 times and only after doing so considering the result. If you were to do this for real, the expected number of mirages used would be 4x the odds (and then 1x, 2x or 5x that depending on the gear).

Instead of quoting the chance of success, the more appropriate value to consider is the expected number of mirages required. For +5 this is 422, 844 or 2,109 depending on your gear (see the top table). If you are getting +5 using fewer mirages, then you are lucky!

Good Luck!

P.S. This does not consider that there is a very wide distribution on the refine function. It is obviously possible to only require 5 refine attempts to get to +5, but it is almost as likely to require over a thousand.

Bearleeable - Lost City · January 2015

Bosk - Sanctuary wrote: »

Instead of quoting the chance of success, the more appropriate value to consider is the expected number of mirages required. For +5 this is 422, 844 or 2,109 depending on your gear (see the top table). If you are getting +5 using fewer mirages, then you are lucky!

Good Luck!

P.S. This does not consider that there is a very wide distribution on the refine function. It is obviously possible to only require 5 refine attempts to get to +5, but it is almost as likely to require over a thousand.

at that point isn't it cheaper just to use orbs? with orbs in the 130-160k area and mirages @ 10k you quickly come to a point where you should have just orbed.. or am I missing something?

My point was, I've never had as much trouble as I do now.

WannaBM - Archosaur · January 2015

I spent about 2000 event gold as well as 20.000 mirages getting 6 pieces to +9

All thanks go out to https://sheet.zoho.com/public/warren.forrest/pwi-refine-costs

shashahah · January 2015

i had issues refining as well, when b4 the update i could at least get to +5 quite ok
but now i cant get to +3 w/o tienkang and at least 200 mirages

what i also noticed it that this not also applies for refining but also for dying fashion (ye ik ik QQ more about something useless, but just to have it mentioned ffs)

usually im quite lucky with getting pure black within lets say 200 randoms >at most<
yesterday, however, i wasted almomst 3k randoms on a single piece and i didnt even get anywhere close to black..not even grey
all i got was mostly green/violet and some shetty brown
ive also talked to some others and they all said they had the same issue

Cotillion - Dreamweaver · January 2015

Its all luck. I think its funny people still have superstitions about timing, location, amount of people around and so on, like any of that matters.

I got an r9rr piece just after the expansion, less than 15 mirages to get it +7. So, I guess based on this we can assume PWE massively increased the chances of refining with the expansion?

No?

Or maybe its cause I was standing on my head at the time of a planetary alignment?

Cotillion - Dreamweaver · January 2015

Just got another piece of r9rr......refined it to +7 using roughly 300 mirages and 20 tienkengs.

KrittyCat - Dreamweaver · January 2015

Cotillion - Dreamweaver wrote: »

Or maybe its cause I was standing on my head at the time of a planetary alignment?

Wait, you can stand on your head?

Are you Tweedle-Dee or Tweedle-Dum? b:chuckle

In all seriousness, though, the superstitions have been around since people started refining. There's nothing wrong with them, and telling people they're wrong for believing something like that is a waste of time. They're going to do what they believe gives them the best chance, even if it means standing on their head at the time of a planetary alignment on the night of a full moon. b:mischievous

Asterelle - Sanctuary · January 2015

I went through about 10000 mirages recently getting 3 +9 refines. Here is data I collected from a pretty lucky run, Tienkang at +2 and Tisha at +5

3012301000100012300000010012001200000010101010010100120101230101001201012001001012000001001012300010123454565676545400100100010100012300123010100000100010010123001010123401230101230012012300101010000010000100100101201000101230010100000010012340101001001010123001234540001200100100123401012300100010000101001001010012300012000100100010010101010000100123400010010101000010010120010100100010010101010123000100001001001201012012010123012010101200001200120012000001000000120120123401010101012340010010000001001010101200123000101010120101010000120123010001001000100000000010010010123400010000012012000101010120101201012012001010101200100010101234001230001010101201001201010000001001001200010100010123000101000000120000120010100123456545654545456767878767876540010100001201000101230012000001230010001001000101230101001001010010001234565454001200123001010101000100100012001001000101001230012000100123454010120010120123401000010101201010001010123010001000123012345400100010000012000123456789

This run wasn't so lucky:

656540120000123012000100120012340123012010100010012001001230000101010000123010120101001230010000101201200000012300101200123001234010000000001200010001200001000010123000001001200101200000100010001010012010120101000123401010100012340010101010010100101201010100012010012000123400120001001001010100100101200001012010101000010000101001000000000100100120101000010001012000123000001010101000120010100100101000010100012010012300101001012010012010100100010101001230012010101000012001010120120100000100120100010100101201000010010012010101010100010101010001000120010001230000010000123001001010000010010012001000100000000010120000101001200000012300001012300012000100012300101201234001010010010010001010101230123400100101200001234565456765400120001010012001012340001000012300100101001001200000001000101012345454540100101230100101000010100000101201000001001000120010001010120010001010000012010101230000101234545401010001201010010010001010010123001234010101000120100001001000010000101230120100010010000120120000123454565656565401010010010101230100010101001010010123000101222001010010101200000101001012000000123010120101010001000000010012340010101201000000100001201201010001230120001001230100001012000000000012345454000010001000001012001012010100012010001230012010001010100120101000101010101001001230100101234010101200123400100010012300101200120010010120100012000101230001200101001010101010001012345401001200010001000010100000010000000100000010101000010001200000123001230100000120101001012345654010101001000101230101012301230120010010101010100000101010010010012345400010120101012001200123454545654001001201201200101200010100100010010100120100000012010000101010010010101201230120101000100123400120101001000012345654010010000010012301234012345456765655401012001010010010120010101200001010010101010123012000010001234010010100101000012340000001001010001010000012012340010120101234001010012001010100012012340010101001230010012012012301230001230000000101010000001001201012010001001010101201234567654010012301010100000101012010100010001001000120123001000100120120101001000001000101230123000101010010000000001001200000012001201001010101001010101001230101234010001201000010010101012301230010101201200010101000101200012000100101012000100000000101000000000010100101001010000101200010000100010101012001200100120101000100012010000000100101001000120012010100100000001200010000100012012301201234012012010000101001010101000010101201000120012001200010123010100100001001010120000012301200001000010001001001010001200001010101001001234545454010012010120001001001010123401010100010120012001010010100120101234540123001010001201001010120100010101000012001010100120000010100120000010000001012345400100001234001010123010101200012300101010012012345454010101234545454540001010012300012301010100001001012345654012345400101200000001234010120123456540000123000100012345456545676787654010120001000001001001010010100010100012001001200100100100101234000100010001010012300101230100100101012000012012301010101010010000101230010100000120001012001012301001201

Some basic stats from this, expected values come from http://pwi-wiki.perfectworld.com/index.php/Refining
For every single one the expected value falls within the 95% confidence interval of the measured value

01 = 1085 (50.5% +/- 2.1%) Mirage (Expected 50%)
00 = 1063
12 = 354 (32.6% +/- 2.8%) Mirage (Expected 30%)
10 = 730
23 = 153 (43.2% +/- 5.2%) Tienkang (Expected 45%)
20 = 201
34 = 62 (40.3% +/- 7.8%) Tienkang (Expected 45%)
30 = 92
45 = 52 (45.6% +/- 9.1%) Tienkang (Expected 45%)
40 = 62
56 = 24 (31.5% +/- 10.4%) Tisha (Expected 33.5%)
54 = 52
67 = 10 (29.4% +/- 15.3%) Tisha (Expected 33.5%)
65 = 24
78 = 6 (40% +/- 24.8%) Tisha (Expected 33.5%)
76 = 9
89 = 1 (20% +/- 35.1%) Tisha (Expected 28.5%)
87 = 4

Fail_BM - Raging Tide · January 2015

56540000123000100012345456545676787654010120001000

b:chuckle

Did you input manually every number?

bloodedone87 · January 2015

Asterelle - Sanctuary wrote: »

I went through about 10000 mirages recently getting 3 +9 refines. Here is data I collected from a pretty lucky run, Tienkang at +2 and Tisha at +5

3012301000100012300000010012001200000010101010010100120101230101001201012001001012000001001012300010123454565676545400100100010100012300123010100000100010010123001010123401230101230012012300101010000010000100100101201000101230010100000010012340101001001010123001234540001200100100123401012300100010000101001001010012300012000100100010010101010000100123400010010101000010010120010100100010010101010123000100001001001201012012010123012010101200001200120012000001000000120120123401010101012340010010000001001010101200123000101010120101010000120123010001001000100000000010010010123400010000012012000101010120101201012012001010101200100010101234001230001010101201001201010000001001001200010100010123000101000000120000120010100123456545654545456767878767876540010100001201000101230012000001230010001001000101230101001001010010001234565454001200123001010101000100100012001001000101001230012000100123454010120010120123401000010101201010001010123010001000123012345400100010000012000123456789

This run wasn't so lucky:

656540120000123012000100120012340123012010100010012001001230000101010000123010120101001230010000101201200000012300101200123001234010000000001200010001200001000010123000001001200101200000100010001010012010120101000123401010100012340010101010010100101201010100012010012000123400120001001001010100100101200001012010101000010000101001000000000100100120101000010001012000123000001010101000120010100100101000010100012010012300101001012010012010100100010101001230012010101000012001010120120100000100120100010100101201000010010012010101010100010101010001000120010001230000010000123001001010000010010012001000100000000010120000101001200000012300001012300012000100012300101201234001010010010010001010101230123400100101200001234565456765400120001010012001012340001000012300100101001001200000001000101012345454540100101230100101000010100000101201000001001000120010001010120010001010000012010101230000101234545401010001201010010010001010010123001234010101000120100001001000010000101230120100010010000120120000123454565656565401010010010101230100010101001010010123000101222001010010101200000101001012000000123010120101010001000000010012340010101201000000100001201201010001230120001001230100001012000000000012345454000010001000001012001012010100012010001230012010001010100120101000101010101001001230100101234010101200123400100010012300101200120010010120100012000101230001200101001010101010001012345401001200010001000010100000010000000100000010101000010001200000123001230100000120101001012345654010101001000101230101012301230120010010101010100000101010010010012345400010120101012001200123454545654001001201201200101200010100100010010100120100000012010000101010010010101201230120101000100123400120101001000012345654010010000010012301234012345456765655401012001010010010120010101200001010010101010123012000010001234010010100101000012340000001001010001010000012012340010120101234001010012001010100012012340010101001230010012012012301230001230000000101010000001001201012010001001010101201234567654010012301010100000101012010100010001001000120123001000100120120101001000001000101230123000101010010000000001001200000012001201001010101001010101001230101234010001201000010010101012301230010101201200010101000101200012000100101012000100000000101000000000010100101001010000101200010000100010101012001200100120101000100012010000000100101001000120012010100100000001200010000100012012301201234012012010000101001010101000010101201000120012001200010123010100100001001010120000012301200001000010001001001010001200001010101001001234545454010012010120001001001010123401010100010120012001010010100120101234540123001010001201001010120100010101000012001010100120000010100120000010000001012345400100001234001010123010101200012300101010012012345454010101234545454540001010012300012301010100001001012345654012345400101200000001234010120123456540000123000100012345456545676787654010120001000001001001010010100010100012001001200100100100101234000100010001010012300101230100100101012000012012301010101010010000101230010100000120001012001012301001201

Some basic stats from this:

01 = 1085 (50.5%)
00 = 1063
12 = 354 (32.6%)
10 = 730
23 = 153 (43.2%)
20 = 201
34 = 62 (40.3%)
30 = 92
45 = 52 (45.6%)
40 = 62
56 = 24 (31.5%)
54 = 52
67 = 10 (29.4%)
65 = 24
78 = 6 (40%)
76 = 9
89 = 1 (20%)
87 = 4

What is the meaning of this? Cause I understand nothing from this numbers? seems like alien launguege for meb:chuckle

Aasaf - Sanctuary · January 2015

bloodedone87 wrote: »

What is the meaning of this? Cause I understand nothing from this numbers? seems like alien launguege for meb:chuckle

+6>+5>+6>+5>+4> busted 0> +1> +2> busted 0> 0> 0 so on and so forth.

WannaBM - Archosaur · January 2015

Fail_BM - Raging Tide wrote: »

b:chuckle

Did you input manually every number?

Ashty is our hero.

Probably ran it on his/her own server software or something like that. And even if its bots doing the refining on the pwi server, we dont care because asty is our hero b:pleased

All you need to learn from the post is that the chances are still the same and noone needs to worry

Asterelle - Sanctuary · January 2015

Nothing fancy, one button i pressed for fail and another for success, used a keylogger to log it and later reconstructed the sequence. But ya, over many thousands of mirages everything is statistically still very much consistent with expected.

Massive Refine Failure

Comments

Categories