Magic Craftsmanship and Orb Change

[Kestiel]

For a while now I’ve seen several crafters touting that Magic Craftsmanship is responsible for orb change, and others who subsist that the ratio of orb change is weighted heavily towards any specific orb. While I’m aware that crafting is due for a change and to have much greater clarity (if Yoshida follows through with what he has said) I wanted to find out through actual testing rather than “eyeballing it”. I went into this unsure of what I would find, mostly because as any human would tell you, I see patterns where there are none. I see prismatic less often because I use it for Magnum Opus, or I see red all the time when I want white.

What I’ve found is that there is no correlation between Magic Craftsmanship and orb change; furthermore I did tests on each and every stat, Control, Craftsmanship and even low stats across the board. I also did a wide variety of different synths on different classes all with a mix of stats for my final test. None of them altered the base percentage out of what could be expected given the margin of error for each test was ~3.1%. Below I’ll list the individual tests I did for anybody who cares to see the numbers.

Method: I used primarily carpenter because they have a cheap and easy 50 synth (Dated Walnut Mask) to provide dLVL0 so that I could rule out dLVL as an influence. Additionally to rule out any sort of issues due to action failures I equipped 100% success and progress abilities for the first five trials, the sixth randomized trial only had 100% progress abilities. Using these I felt I had isolated the synthing mechanics enough that I could get an accurate representation of what I was after. I needed ~1,000 samples to hit an agreeable margin of error (~3%) and thus make sure my results were statistically significant.

Abilities: Unerring Hand, Grand Design, Tricks of the Trade, Greater Calling* (*Only the first five trials).

Trial #1: Low Magic Craftsmanship
Craftsmanship: 112
Control: 114
Magic Craftsmanship: 76
Samples: 1,004

White: 264 (26.294%)
Yellow: 255 (25.398%)
Red: 246 (24.501%)
Prismatic: 239 (23.804%)

Note: As you can see they’re all pretty even, and with a margin of error of 3%, they could easily swing (assuming 25% is the true base) from 22%-28% as we’ll see in later tests.

Trial #2: High Magic Craftsmanship
Craftsmanship: 99
Control: 111
Magic Craftsmanship: 143
Samples: 1,000

White: 273 (27.3%)
Yellow: 244 (24.4%)
Red: 245 (24.5%)
Prismatic: 238 (23.8%)

Trial #3: Low Control
Craftsmanship: 116
Control: 67
Magic Craftsmanship: 105
Samples: 1,115

White: 302 (27.085%)
Yellow: 262 (23.497%)
Red: 254 (22.78%)
Prismatic: 297 (26.636%)

Trial #4: High Control
Craftsmanship: 95
Control: 166
Magic Craftsmanship: 81
Samples: 1,221

White: 287 (23.505%)
Yellow: 321 (26.289%)
Red: 299 (24.488%)
Prismatic: 314 (25.716%)

Trial #5: Low Stats
Craftsmanship: 47
Control: 37
Magic Craftsmanship: 17
Samples: 1,149

White: 282 (24.543%)
Yellow: 279 (24.281%)
Red: 294 (25.587%)
Prismatic: 294 (25.587%)

Trial #5: Randomized Stats & Synths & Disciplines
Samples: 2,560

White: 661 (25.82%)
Yellow: 625 (24.414%)
Red: 636 (24.843%)
Prismatic: 638 (24.921%)

Conclusion: As you can clearly see, the orbs are on a set %, more than likely a completely equivalent 25% for each orb. Particularly the last trial is most telling because I switched to WVR, LTW, BSM, ARM and ALC for a wide variety of synths from 36 to 54. With a larger sample size the breakdown of the orbs was less varied than in most of the other trials, with the largest variance being only 0.82%.
So, I hope all those people who keep advocating to “lower your Magic Craftsmanship because it will increase your orb change rate” will understand that they have been perpetuating a rumor that has no factual basis. It simply isn’t true, so go and stack Magic Craftsmanship as much as you like without the worry that your prismatic orbs are going to vanish.

TL;DR Magic Craftsmanship doesn't affect the color variance of the orb.

[REDace0]

Finally! Actual hard data on crafting.

Thanks for being all scientific and debunking the rumors. Can't give you enough likes for all your work.

[tymora]

Interesting. 25% at DLvl 0. Thanks.
Orb Ratio however varies with DLvl.

Much lower samples than OP's but should still show a nice relationship.

Orb Ratio   DLVL   White        Yellow       Red               Rainbow       
           -40   0.262712   0.152542   0.186440678   0.398305       
            -20   0.25876   0.177898   0.169811321   0.393531       
             -9   0.352941   0.154412   0.139705882   0.352941     
             -7   0.277108   0.210843   0.162650602   0.349398       
              0   0.296167   0.202091   0.216027875   0.285714

[Green_and_a_Half]

i have changed to the conclusion that Magic Craft and Craft should be double the synth lvl and have abandoned the lower magic craft = rainbow theory.

I would love to see that test if you feel like it as i am all 50 and dont really craft that much now.
also i have been telling people stat = 0.1% quality
meaning +10 stat = 1% quality boost
this one is all personal experience and also needs to be tested

so for me, its 120 crafting 120 magic crafting 147 control all day.

great job doing the tests, at 50 synths an hour, you did 140 hours of testing?

[Kestiel]

Interesting. 25% at DLvl 0. Thanks.
Orb Ratio however varies with DLvl.

Much lower samples than OP's but should still show a nice relationship.

Orb Ratio   DLVL   White        Yellow       Red               Rainbow       
           -40   0.262712   0.152542   0.186440678   0.398305       
            -20   0.25876   0.177898   0.169811321   0.393531       
             -9   0.352941   0.154412   0.139705882   0.352941     
             -7   0.277108   0.210843   0.162650602   0.349398       
              0   0.296167   0.202091   0.216027875   0.285714

What's your sample size? You need ~1,067 samples (1,000 is good enough for a ~3.1% margin of error) otherwise you run the risk of getting skewed results that don't show the actual rates. I did synths varying from -15dLVL to +5dLVL and those were mainly in the last 2,000 or so synths, and they show the same rates of orbs.

[Kestiel]

i have changed to the conclusion that Magic Craft and Craft should be double the synth lvl and have abandoned the lower magic craft = rainbow theory.

I would love to see that test if you feel like it as i am all 50 and dont really craft that much now.
also i have been telling people stat = 0.1% quality
meaning +10 stat = 1% quality boost
this one is all personal experience and also needs to be tested

so for me, its 120 crafting 120 magic crafting 147 control all day.

great job doing the tests, at 50 synths an hour, you did 140 hours of testing?

;_; Please don't remind me how long I spent...I try so hard not to think about it. Plus the hours of working out a way to get Notepad++ to cancel out the additional orb info from Greater Calling on the first five trials.

Edit: 50 synths an hour is a bit generous. I clocked around 65-75 seconds per synth if I was paying extra-special attention, which is not possible all the time. So some synths took 2-3 minutes or longer if I got up and did something else.

[tymora]

What's your sample size? You need ~1,067 samples (1,000 is good enough for a ~3.1% margin of error) otherwise you run the risk of getting skewed results that don't show the actual rates. I did synths varying from -15dLVL to +5dLVL and those were mainly in the last 2,000 or so synths, and they show the same rates of orbs.

Only a couple hundred.
Yes, I know the values are inaccurate. Unfortunately no time to do more tests.

Clean slate is a good skill to do testing with. Effectively doubles the sample size from 1 synth.

[Kestiel]

I've not used it since the update, if it doesn't refresh abilities too it incurs the risk of synth action failure having an effect on orb color (which can erroneously skew results) and disregarding that, would make the synth take longer on average than using the 100% progress abilities. Of course if failed synth actions don't attribute to color change then that'd definitely be a preferred way to go. After I saw the new updates (recently returned ~1mo ago) to clean slate I pretty much disregarded it and haven't looked back on it.

[KumaAkuma]

It would have been kind of ridiculous if magic craftmanship gave more white orbs. You want rainbows when trying to HQ.

[pompey_dan]

this is what i find when crafting at least it appears to be true for all the new recipes i synth, it may sound stupid but please try it out before ruling it out

magic craftmanship - higher stat gives a higher quality gain with a successful synth.
craftmanship - higher stat gives more progress gain with a successful synth.
control - higher stat gives better chance of a successful synth.

I still dont believe the whole one stat for one synth sort of deal, i know people who stack purely craftmanship for hq nuggets whilst others stack mag craftmanship n both believe they have the best set up lol.