Summoners, due to a technical issue, this week's series of Alliance Quests will be cancelled at 4pm PT. There will not be any Rewards for this series. The next series of Alliance Quests will start on April 18 at 12pm PT. We will be assessing the impact on Alliances and compensating accordingly, as well as returning Alliance Tickets.

# Have the odds of pulling a 5* from GM crystals been changed recently?

Dr_ARCHer
Posts:

**125**★
For those of you who buy a lot of GM crystals, have you noticed that the odds of pulling a 5* been reduced recently? The odds of pulling a 4* seems to be about the same, but the odds of pulling a 5* seems to have dropped. This seems to happen just before 16.0 patch, but it could have happened with the 16.0 patch.

Just want to compare notes and data points.

Just want to compare notes and data points.

2

## Comments

4,514★★★★★Dr. Zola

125★For the GM crystals, my odds for 5* was about 1 out of 80 to 100, and for 4* about 1 out of 8 to 10. Recently though, my odds are still zero out of 200, but 4* odds remain at around 1 out of 8 to 10.

As for GM featured crystals, the odds used to be the same, with the additional odds of pulling the featured 5* around 1 out of 200 to 250. I don’t have enough data points on GM featured crystals to give more accurate estimates of the odds because they were available only twice.

301★★This is just normal randomness. Randomness is always streaky. If everyone were consistently getting the same results they got in the past, then it wouldn't be random.

125★The dgp (data generation process) is unknown, so I’m merely trying to increase the sample size to try to get a better estimate of the unknown population.

301★★For one thing, even if you got a hundred responses, that still wouldn't be a good sample size.

For another, the people who are most likely to respond to a forum post are the ones who have extreme experiences. Most of the people who haven't noticed a difference aren't even going to click on the thread. The responders will be as far as you can get from a random sample.

There are threads like this all the time. People are constantly having a string of good/bad luck with crystals or champs and asking if there has been a change. It's just randomness being clumpy.

125★For example, in my case, I’ve added another 30 GM crystals and still have not pulled a 5*. That makes it about 230 crystals with no 5*. If I can add on more data points, I can at least infer what sort of odds. Of course there are certain biasness in the information, but there is no other source available.

125★12,059GuardianIf the odds of pulling a 5* champion are about one in a hundred, you'll have a reasonable sample size to confirm those odds have increased to more than one in two hundred after opening about six thousand crystals. But if they've only been increased to about one in a hundred fifty, you'll need more than ten thousand crystal openings. Let me know when you get there.

No, I'm not kidding nor am I flippantly guessing. If the odds are about one in a hundred, then you have about a 50/50 chance of pulling at least one in 66 openings. But to make the claim that you are seeing an unreasonable level of bad luck such that it is more likely the odds of the crystal have been changed you need to see a level of bad luck you aren't likely to ever see. For what you see to be a one in a hundred chance requires seeing something like no pulls out of 450 openings. For what you see to be a one in a thousand chance requires seeing something like no pulls out of 680 openings. Even that could just be bad luck. You'd need to observe something like ten openings out of 1500 pulls to believe that the odds are somewhere around one in 150, but you'd have a certainty level of maybe plus or minus 10%. To measure more accurately than that you'd need several times that number, at least four times that many realistically.

The problem is that if you decide to measure the odds of a coin flip being 50/50, if you flip enough times you will eventually get three heads in a row, and then four heads. Eventually you'll get nine heads in a row. That doesn't mean the coin suddenly broke. It just means random probability generated that result, which it is eventually going to do.

I used a source for random bits from the internet and generated this sequence of random ones and zeros: 11100101 00100111 11000011 11111100 00000101 00011110 00111101 10101011 10101000 00101100. That sequence is generated randomly. Notice that at one point it spits out eight ones in a row followed by seven zeros in a row. Someone that doesn't understand random statistics might claim that the odds against that happening are 32768 to one. But that's not true, because randomness doesn't work that way.

I've done these tests myself in other contexts, and the problem is that to do a test right, you can't measure the gaps between success results. You have to measure enough successes and measure the number of trials to get there. Seeing nothing out of 230 pulls doesn't say anything. Seeing ten out of 1500 says something. But to be confident in the number you'd really have to pull at least fifty, more likely a hundred featureds and measure how many pulls it took to get there. And as I said, that's likely to involve ten thousand pulls or more.

If people are going to be suspicious about the odds on the crystals changing all the time, you will simply never have enough data to prove to anyone that the odds didn't change yesterday or won't change tomorrow. You're never going to see enough openings in a short enough period of time to deal with that level of suspicion.

125★Thanks very much for the detailed explanation. My point here is that based on my previous openings of over a 1000 GM crystals, my odds seem to indicate a drop rate of 1%. With 16.0 or just before 16.0, my odds surprisingly have been lowered. The odds of 4* seems to habpve I proved slightly, and I do then to get one or 2 4* for 10 crystals while previously, such occurrences were fewer.

While I am sure I will reach another thousand crystals or so, I’m just checking if others have data points to contribute.

125★125★331-340 - 2 4*

228★228★742★★★3,460★★★★2,933★★★★826★★125★351-360 - 0 4*

542★★14 - 3*

0 - 4*

0 - 5*

So.. I never spend in this **** anymore

I got Kamala 3* FOUR TIMES

125★361-370 - 0 4*

371-380 - 1 5* 1 4*

1,102★★★125★381-390 - 0 4*