Dice Roll! ~ Looking for Oddities in Phantom Roll

[JMunin]

In my down time, I've been playing with phantom roll and the outcomes with and without jobs in party. Though this, it could be argued, is an act of trying to find patterns where there are none--and in the process ascribing to a certain kind of gambler's fallacy--I found the experiment interesting nonetheless.

This began as an experiment to test AF for what might be hidden effects concerning rolling. I rolled Corsair's roll naked, with full AF and then with each individual piece for a total of 700 times. Though nothing really stood out from AF piece to AF piece, across the board something stayed consistent. The totals?

I: 16%
II: 17%
III: 14%
IV: 15%
V: 20%
VI: 18%

Corsair's lucky number is V, and with a corsair in party--me--it looked like V nudged out the other rolls just slightly. Since I had nothing better to do, I decided to test the other rolls and see how the probability changes when there isn't the specific correlating job in party to when the correlating job is in party. The following rolls were rolled 200x, 100x alone w/o the corresponding job, then 100x w/ the corresponding job.

Choral Roll -- Lucky #2

No Bard in Party

I: 15%
II: 12%
III: 15%
IV: 21%
V: 22%
VI: 15%

Bard in Party

I: 16%
II: 16%
III: 9%
IV: 21%
V: 17%
VI: 22%

It would appear, both with and without a bard in party, the roll is weighted toward IV~VI. Why I'm not sure, especially compared to how the probability for Healer's Roll turned out.

Healer's Roll -- Lucky #3

No White Mage in Party

I: 18%
II: 11%
III: 12%
IV: 18%
V: 19%
VI: 22%

White Mage in Party

I: 30%
II: 12%
III: 18%
IV: 15%
V: 9%
VI: 16%

I thought this was a real oddity: #1 turned up a whopping 30% of the time with a white mage in party. My parties with white mages in the past have supported this: it always seemed like it was easy to hit III--either right at the outside or doubling-up to it--or roll up high to X or XI with a white mage in party. Also, if you'll add up I~III, they appear 60% of the time. Why then, does IV-VI appear 60% of the time in Choral Roll?

Ranger Roll

No Ranger in Party

I: 19%
II: 14%
III: 13%
IV: 16%
V: 20%
VI: 18%

Ranger in Party

I: 15%
II: 20%
III: 15%
IV: 14%
V: 24%
VI: 12%

Why does V in this roll turn up so much? Otherwise nothing really stands out about the probability of this roll.

I decided to test Rogue's roll as well, as a contrast to the numbers put out in Corsair's.

Rogue's Roll

No Thief in Party

I: 15%
II: 24%
III: 23%
IV: 14%
V: 14%
VI: 10%

Thief in Party

I: 13%
II: 10%
III: 23%
IV: 18%
V: 19%
VI: 17%

III shows up a lot for no apparent reason in both tests. IV~VI rolls all went up with thief in party, and in the second they slightly edge out the I~III numbers in how often they turn up. However, nothing about the rolling leads me to think it really changed. However, I've never liked IIIs for this roll; III's always seem to lead to IXs.

From some of the pecularities I've seen here and in xp, it would seem that rolls might be weighted, that they tend to hit certain numbers. Rolls that support this are Corsair's (slightly heavier from IV~VI) and White Mage (heavier from I-III). Choral Roll seems an oddity: why would it be weighted for higher numbers? And Rogue's and Ranger I can't tell much from, except that Rogue's seems to roll heavy on III.

This was a small sampling, of course, and I might need to roll 1000x for each roll to see if anything is weighted or not.

And what are total %s, you ask? (700 Corsair, 200 each of Bard, White Mage, Ranger and Rogue's--for a total of 1500)

I: 17.4%
II: 15.1%
III: 15.8%
IV: 16.9%
V: 18.2%
VI: 16.7%

Looks like it evens out pretty well, except that V still edges out everyone. Though nothing is really conclusive, it is interesting.

I wanted to ask any other Corsairs whether or not they've noticed anything different about how they roll when the corresponding job is in party. If in fact the probability changes, it makes the arguement for using the appropriate job roll all the stronger.

All tests took place in Nashmau with the corresponding jobs at level 75. Roll once > subjob change > roll again. Nothing was ever doubled-up.

[Priran]

dangly parts

[Delekii]

I know it sounds kinda arsish to say, but none of the results are particularly standout considering the sample size.. a couple of outlying "that's odd" numbers, but in a sample of 1-200 rolls I would expect inconsistencies of those kinds of magnitude..

I know people say that they feel that the dice might be weighted, but I still stand by the fact that I think any weighting people are seeing is somewhat placebo.

[Lithiani]

A COR I know claims that the game rolls 1-7, not 1-6, and a roll of 7 either displays as a "good" number (if the job is in the PT) or a "poor" number if not.

Which is why Evoker's Roll has an absolute genius for rolling double 6. 6 on Phantom Roll is bad because it's past the lucky number, and then bust or 9 are the only "worse" results it can go to.

It would not surprise me if the dice were weighted.

[Priran]

A COR I know claims that the game rolls 1-7, not 1-6, and a roll of 7 either displays as a "good" number (if the job is in the PT) or a "poor" number if not.

and you claimed in the other thread that debuffs can land and not take effect.

badum pssh

[Stilgar]

I thought priran had been banned?

[spooky]

Write a parser to record all the rolls for a month of playing COR, and then have a look at the results.

[Kainsin]

While I do recall rolling quite a number of I's when using Healer's Roll, I also believe that it's just luck with the numbers and no real weighting is going on.

That being said, you sure changed your subjob around quite a lot...how do you have the patience for that? >.>

[safer]

someone insert the dilbert cartoon and end this thread <_<

75 levels of playing cor.. and just as the job says... its random!

[Not Kuno]

someone insert the dilbert cartoon and end this thread <_<

75 levels of playing cor.. and just as the job says... its random!

http://i2.photobucket.com/albums/y38...no/dilbert.gif

[Kylen]

I know it sounds kinda arsish to say, but none of the results are particularly standout considering the sample size.. a couple of outlying "that's odd" numbers, but in a sample of 1-200 rolls I would expect inconsistencies of those kinds of magnitude..

I know people say that they feel that the dice might be weighted, but I still stand by the fact that I think any weighting people are seeing is somewhat placebo.

What Delekii said. Hate to break it to the OP, but 100 or even 200 samples isn't nearly enough to start drawing conclusions like that, especially considering the relatively low variance of your results.

Just for comparison, here's what I got from rolling a 6-sided die 100 times (sue me, I'm bored >.>)

1: 17%
2: 22%
3: 20%
4: 16%
5: 13%
6: 12%

This is about the same variance that you got in your results testing with different jobs, but I'm not about to go claiming that the die is weighted towards 2 and 3 just because they showed up a few more times than the other numbers. Your first sample with 700 COR rolls is more statistically significant than your other tests, and as I'd expect from a truly random system your results there show less variance. My guess is that the more tests you do, the closer your results will converge to 16.6% for each roll.

I appreciate that you did those tests, though, as it reinforces my belief that Corsair rolls truly are random (well, pseudo-random, since you can never get truly random numbers in programming) and not based on factors such as gear, jobs in party, etc. If you had rolled V 35% of the time in your 700 COR roll tests that would have been enough evidence to indicate that the rolls are indeed weighted, but unfortunately a +/- 3% variance from the mean is statistically insignificant considering the number of tests done.

[Priran]

I thought priran had been banned?

YOU SILLY NEGRO

[divisortheory]

If you do this test 700 times EACH the results will hold a little more weight. Unfortunately that would be quite annoying. Windower macroes doing an infinite loop, combined with a parser that can handle Corsair rolls (you'd need to write your own), could help here.

[JMunin]

Like I said in the OP, the sampling is small. I posted here for some criticism to determine if I should continue or not with some larger sampling. No one's broken my heart or written anything I didn't expect.

I originally wanted to see if AF or gear effected any rolls. I haven't tried trump gun--though I might now--and if phantom flueret would ever appear on pandemonium ah I'd try that as well.

All said, I would prefer if the die weren't weighted. Its just a fear I had. Do I think there should be items available to Corsair to improve chances of rolling better--or perhaps more consistently? Yeah, and hopefully we'll see something in the December update. But I think I'll take your post to heart, Kylen, and use the 700 Cor rolls as evidence that the rolls are, as far as any of us can assume, random.

And Kainsin: takes about 20 minutes to roll 100x. It wasn't too bad, really. 1500x total = 5 hours for the raw data. And once you got into the rhythm, it was relatively easy.

[LinktheDeme]

Dont mean to be toooo much of a necro but figured it was better than making a new topic on the same thing. Im leveling corsair and i notice a few things. I'm wondering if SE makes percentages for double up based on the following. . .

1) Based on the last number you rolled
2) based on the last number you double upped
3) Based on the total number of what you currently have in the roll.

Noticed that a lot of the times when i roll a 6 and i double up (I even wait like 10 secs too sometimes to see if its just if i double up too fast) that i'll double up another 6. I know people are gonna say the random thing but i do it more often then other numbers when i double up on 6. Most of the times its 2-3 times and in different parties as well. It would have to be pretty random to do consistant (1/6^2)^3.

Also in other FF games SE tends to rig what your next outcome will be like with setzers slot machine. Just thought the maker of this post would be interested in this. That and he/she seems like a more dedicated person to test this than I.

P.S. Im also wondering if phantom rolls randomosity is just a time based thing like every X.X seconds will change numbe, or if its a random number generator.

[aurik]

Dont mean to be toooo much of a necro but figured it was better than making a new topic on the same thing. Im leveling corsair and i notice a few things. I'm wondering if SE makes percentages for double up based on the following. . .

1) Based on the last number you rolled
2) based on the last number you double upped
3) Based on the total number of what you currently have in the roll.

Noticed that a lot of the times when i roll a 6 and i double up (I even wait like 10 secs too sometimes to see if its just if i double up too fast) that i'll double up another 6. I know people are gonna say the random thing but i do it more often then other numbers when i double up on 6. Most of the times its 2-3 times and in different parties as well. It would have to be pretty random to do consistant (1/6^3)^3.

Also in other FF games SE tends to rig what your next outcome will be like with setzers slot machine. Just thought the maker of this post would be interested in this. That and he/she seems like a more dedicated person to test this than I.

P.S. Im also wondering if phantom rolls randomosity is just a time based thing like every X.X seconds will change numbe, or if its a random number generator.

The chance to bust doubling on a 6 is 1/6, not 1/6^3.

[jibini]

Pfft, nonsense, experience has taught me that the chance to bust doubling on a 6 is about 95%.

Seriously, I only ever roll a 6 when I'm already at 6, and if I'm at 6 I can almost guarantee that doubling up will bust me.

[LinktheDeme]

Dont mean to be toooo much of a necro but figured it was better than making a new topic on the same thing. Im leveling corsair and i notice a few things. I'm wondering if SE makes percentages for double up based on the following. . .

1) Based on the last number you rolled
2) based on the last number you double upped
3) Based on the total number of what you currently have in the roll.

Noticed that a lot of the times when i roll a 6 and i double up (I even wait like 10 secs too sometimes to see if its just if i double up too fast) that i'll double up another 6. I know people are gonna say the random thing but i do it more often then other numbers when i double up on 6. Most of the times its 2-3 times and in different parties as well. It would have to be pretty random to do consistant (1/6^3)^3.

Also in other FF games SE tends to rig what your next outcome will be like with setzers slot machine. Just thought the maker of this post would be interested in this. That and he/she seems like a more dedicated person to test this than I.

P.S. Im also wondering if phantom rolls randomosity is just a time based thing like every X.X seconds will change numbe, or if its a random number generator.

The chance to bust doubling on a 6 is 1/6, not 1/6^3.

sorry was typo mean for it to be (1/6^2)^3 ^2 for having 6 twice in a row and ^3 for that situation happening 3 times in a row ill fix it in my post

[JMunin]

Dont mean to be toooo much of a necro but figured it was better than making a new topic on the same thing. Im leveling corsair and i notice a few things. I'm wondering if SE makes percentages for double up based on the following. . .

1) Based on the last number you rolled
2) based on the last number you double upped
3) Based on the total number of what you currently have in the roll.

Noticed that a lot of the times when i roll a 6 and i double up (I even wait like 10 secs too sometimes to see if its just if i double up too fast) that i'll double up another 6. I know people are gonna say the random thing but i do it more often then other numbers when i double up on 6. Most of the times its 2-3 times and in different parties as well. It would have to be pretty random to do consistant (1/6^2)^3.

Also in other FF games SE tends to rig what your next outcome will be like with setzers slot machine. Just thought the maker of this post would be interested in this. That and he/she seems like a more dedicated person to test this than I.

P.S. Im also wondering if phantom rolls randomosity is just a time based thing like every X.X seconds will change numbe, or if its a random number generator.

Though I still think their may be some pecularities hidden in how phantom roll works, we need to acknowledge that our brains are looking for patterns where there very likely are none. I only suspect because I don't trust SE to make it completely random and they might sneak a variable in there.

That said, none of the rolls I tested used any double ups. These are pure "first rolls" and double-ups weren't touched my sampling. Though a double-up testing might be warranted, testing it thorougly strikes me as troublesome, particularly if it's:

1) Based on the last number you rolled
2) based on the last number you double upped
3) Based on the total number of what you currently have in the roll.

However, what I would think would make it simple is if someone simply kept track of how many times they bust when they double up on a VI. In fact, I'll start keeping tract from now on in merits and events on my corsair.

I'll do it this way: every time I roll a six, I'll double-up. I'll write down the number rolled or if I bust, and I'll do it until I have a fairly large sample size. It might be awhile though--I haven't merited much lately and I'm going on vacation in a week. So if anyone else wants to try, I'd be interested in seeing it.