Drop rates for Featured. Are they weighted probabilities? And how does RNG algo work?
Dude17
Member Posts: 133 ★★
I know that we have a 100% chance to pull a 6* champ in a Featured crystal but are probabilities weighted by champ? eg. newly released champs vs other champs in the crystal.
1) I’m assuming not or it would need to be specified in the drop rates, yes?
2) Also curious how the RNG algorithm works, ie. is it affected by pop/spin method, bulk openings, current progression title, spending habits, alliance prestige, current roster, etc? Or is it literally a random number generator based on no additional data points? There’s so many possibilities, so am genuinely interested, thx.
1) I’m assuming not or it would need to be specified in the drop rates, yes?
2) Also curious how the RNG algorithm works, ie. is it affected by pop/spin method, bulk openings, current progression title, spending habits, alliance prestige, current roster, etc? Or is it literally a random number generator based on no additional data points? There’s so many possibilities, so am genuinely interested, thx.
5
Comments
Expecting an “even” distribution over a small sample is what gets people into trouble. That’s possibly one of the least likely outcomes.
Here are my distributions over about 40 pulls. Color coding corresponded to most/least wanted:
I don’t think that clumpiness is in any way “not random.” I’m happy the big numbers aren’t over Groot, Rocket, Thor and SL, but they could just as well have been. This is a pretty reasonable pRNG distribution over my ~40 pulls.
Dr. Zola
Edit: Not intended to be a flex. Just helps I think to see someone’s actual results instead of someone’s “impressions.”
Dr. Zola
QS 1 new
Titania 1 new
Rocket 1 new
Miles 1 new
SWitch 1 new
Rogue 3
X23 3
SL 3
Groot 3 new
Thor 3
Warmaxhine 1
Professor 2
Manthing 1
Only 4 Champs I'm looking for, all new.
Galan gorr negs maw
Nothing affects the drop rates. Spinning a crystal doesn't matter as I'm the reel is for entertainment or agony only. Your champ is decided the second you click on it to stop spinning or it stops on its own.
Each crystal is independent of others and the previous pull has no bearing on the next.
Rogue 4x
Professor X 2x
Winter Soldier 2x
Galan 1x NEW
Archangel 1x NEW
Man-Thing 1x NEW
Ebony Maw 1x
Scarlet Witch 1x NEW
War Machine 1x
X-23 1x NEW
Miles 1x NEW
Groot 1x NEW
My biggest wants now are Valkyrie, Negative, Quicksilver and dupes on AA and SWitch.
ScarletWitch 2
Starlord 2
Thor 2
Titania 1
X23 2
WarMachine 1
MilesMorales 2
Ebony maw 1
Galan 1
Bwdo 1
Mr negative 2
Hulkbuster 1
Rouge 1
AA 1
Prof x 2
Groot 1
Shangchi 1
Quicksilver 1
No. No. No. No, No, no, no. Yes.
Crystals like the featured 6* crystal simply roll a random number using a computer PRNG and drop a champ determined by that roll using some form of pigeonholing algorithm. Which is to say, if there are 24 champs in the crystal, the game could randomly choose a whole number from one to 24 and use that to decide which champ to drop, but there's also ways to choose a random number between zero and one and just slice that range up into 24 partitions and drop based on what pops out. In other words, pick a random number between zero and one, if it lies between 0/24 and 1/24, drop the first reward; if it is between 1/24 and 2/24 drop the second reward, and so on.
By default, all drops are weighted evenly. When this is not the case, this is specified by the crystal. For example, a standard Cavalier crystal has a 2.4% chance to drop a 6* champ, a 12% chance to drop a 5* champ, a 31% chance to drop a 4* champ, and a 50% chance to drop a 3* champ (plus a 3.2% chance to drop a 5* Nexus and a 0.6% chance to drop a 6* Nexus). For crystals like this, the game rolls a random number to decide which of those types of drops it will drop. So for example, it picks a random number between zero and one, and if it is between 0 and 0.024 it drops a 6* champ, if it is between 0.024 and 0.03 it drops a 6* Nexus, if it is between 0.3 and 0.158 it drops a 5* champ, if it is between 0.158 and 0.19 it drops a 5* Nexus, etc. Then if the drop is determined to be a 3 or 4 or 5 or 6* basic champ, it rolls again to determine which champ from among the champs currently listed in the basic pool.
Absolutely nothing you do when you spin the crystals affects the drop, or can affect the drop. That's because the crystal drops are calculated on the game servers, and the game servers have no idea what you're doing in the game client. The servers don't know if you're using the pep method or swirling the crystals around or waiting for the crystal to zoom in from the bottom right corner. The servers don't know, so you have just as much effect on the crystal drop as you do when you yell at your favorite sports team while watching them on TV. They can't hear you, so this is literally impossible.
There are people who will tell you otherwise, and will tell you they have convincing evidence. You should treat such reports as you would someone who claims that when they wear their lucky T-shirt their favorite sports team plays better.
The other things that people claim influence crystal openings - how much you spend, what your current roster is, which items you currently have in inventory - these are all at least theoretically possible, but Kabam has explicitly disavowed. And while it is impossible to prove absolutely that they don't do these things, every controlled statistical test has failed to uncover any such crystal rigging. *If* it is happening, it is happening in an undetectable way. And crystal odds tampering that we can't detect through statistical tests of tens of thousands of crystals is tampering you will never notice yourself opening a couple dozen.
So for example if Kabam says there's a 3% chance for a Cav crystal to drop a 6*, there's no way you or I could ever tell if they rigged it to be 2.99999 instead. I couldn't possibly disprove this, but I also couldn't care less. But if they rigged it to be 2% or 1%, that would be detectable, and that's easier to disprove.
There's also an idea floating around out there that the crystals are programmed with an algorithm to have a higher probability of giving you what you don't need. So for example, if you are overflowing with tech T4CC but have no skill T4CC, the T4CC crystals will have a higher probability of giving you tech T4CC. This is ridiculous, easily disproved, and also a lunatic thing to believe in. Because how would the game *know* what you want and don't want? Sure, you could argue that it will give you more of what you already have a lot of, but what if that's what you want? Does the game read my mind and know that I'm not going to use those catalysts? Does it calculate that it is unlikely I will want even more of them to do a rank up? There's no good reason for the game to even want to do this, much less trying to figure out what sort of rules the game would follow to try to figure this out in the first place. Needless to say, anyone who believes this is, in my opinion, completely nuts.
Groot x1
HB x2
Negative x1
Racoon x2
Manthing x2
Titania x1
Galan x3
Starlord x1
Wiccan x1
X23 x1
1 - Man Thing (sigs)
2 - Gorr (new)
3 - Gorr (awaken)
4 - Shang Chi (sigs)
5 - Miles Morales (sigs)
6 - Hulkbustere (awaken)
7 - Black Widow DO (new)
8 - Titania (new)
9 - ProfX (awaken)
10 - Quicksilver (new)
11 - ProfX (sigs)
12 - Titantia (awaken)
13 - Winter Soldier (sigs)
14 - Titania (sigs)
15 - Titania (sigs)
16 - Archangel (sigs)
17 - Hulkbuster (sigs)
18 - Winter Soldier (sigs)
19 - Hulkbuster (sigs)
20 - X-23 (sigs)
21 - Quicksilver (awaken)
22 - Quicksilver (sigs)
23 - Thor (awaken)
24 - Archangel (sigs)
25 - Gorr (sigs)
26 - Archangel (sigs)
27 - Rogue (sigs)
28 - Hulkbuster (sigs)
29 - X-23 (sigs)
30 - Groot (awaken)
31 - Galan (new)
32 - Shang Chi (sigs)
33 - Galan (awaken)
34 - Wiccan (new)
35 - Galan (sigs)
36 - Black Widow DO (awaken)
37 - Star Lord (new)
38 - Titania (sigs)
39 - Winter Soldier (sigs)
40 - Quicksilver (sigs)
41 - X-23 (sigs)
42 - Miles Morales (sigs)
43 - X-23 (sigs)
44 - War Machine (sigs)
Distribution:
Titania: 5
Quicksilver: 4
Hulkbuster: 4
X-23: 4
Galan: 3
Gorr: 3
Archangel: 3
Winter Soldier: 3
BWDO: 2
Miles Morales: 2
ProfX: 2
Shang Chi: 2
Wiccan: 1
Groot: 1
Man Thing: 1
Rogue: 1
Star Lord: 1
Thor: 1
War Machine: 1
Valkyrie: 0
Scarlet Witch Classic: 0
Ebony Maw: 0
Mister Negative: 0
Rocket Raccoon: 0
16 of 44 are featured champs, which is somewhat higher than average (statistically speaking you'd expect 11 on average). This doesn't prove the crystal is biased towards featured champs (given the margin for error with this sample size this would be a very low confidence result) but it does more strongly suggest the crystal is not biased away from featured champs (if we expect 11 and we actually get 16, that's moderately higher than expected, but if someone believes the crystal is biased away from featured champs and actually expects something like 8 or 5 to show up, then 16 would be far less likely to happen if that were true).
I was initially missing ten of the twenty four champs in the crystal. Statistically speaking I would expect to get between seven and eight of those ten after 44 crystals. I got seven. That's basically what I would expect, and there's no evidence the crystal was deliberately trying to avoid giving me what I was missing.
Is the crystal weighted towards the bottom of the barrel champs? Well, I think most would assert those were Rocket, Groot, Winder Soldier, and Star Lord. That's four out of 24 or one in 6. I would expect to pull about seven of those out of 44 openings. I actually pulled six. That's basically what I would expect, with no evidence of them coming out a lot more frequently than statistically likely.
Now, which champ did I want the most out of this crystal? Actually, with all the great champs in here, the one I was hunting for the most was actually Scarlet Witch Classic. As I have yet to pull her I guess that's some slight statistical evidence that Kabam has managed to implement mind reading in the crystal technology. But as this only has 6.5 to one odds against, its not super strong evidence for computational telepathy yet.
Also, I pulled Titania, then awakened her, then added sigs to her, then added sigs again to her, then immediately ranked her up and used her against EoP Terrax because I heard she was good for that fight. This has to be the worst failure of crystal rigging Kabam has ever implemented.
As an example, some of my most wanted pulls from this featured have been SW and Starlord (judge me all you want) and I have pulled and duped both of them so far, as well as OG Thor. Just because you aren’t getting the champs that you want doesn’t mean Kabam is somehow unfairly making that happen. I’ve also gotten my Man-thing from sig 100 to max sig in my first 16 crystals. Just cause I pulled him a lot doesn’t prove bias in crystals.
I still think Kabam needs to contract out DNA to do a “Video Game Statistics” series that I can just paste every time someone asks these questions.
For illustration I will share my 5 pulls:
Winter Soldier (2x)
Groot
Star-Lord
Mister-Negative
Seems rigged in favor of bad old champs right? Now check the pulls of other people here and you'll see that slowly the amounts of pulled champs are becoming closer and more or less even, depending on how many crystals they have opened.
Lastly, the RNG is same no matter whether you pop or spin, open 1 or 10 at a time.
People are bad at statistics because it requires being good at four things people are bad at. It requires setting aside biases when observing the world. It requires keeping track of small details. It requires abstract thinking. And it requires rigorous analysis. All four things would get our caveman ancestors killed. But all four things are necessary to be able to say anything about the way statistics works in the world.
Most people honestly hook their feet right into the starting gate and decapitate themselves at the start of the race as it swings open. Which is: what is random(ness)? When asking "are the crystals random" most people cant' even define what that question means in a practical way, which is why you'll see people say things like "computers can't really choose random numbers so the crystals aren't really random." This sentence is true only in an impractical academic setting that most people are unqualified to participate in.
The crystals are random enough because they satisfy the colloquially reasonable definition of randomness: the crystal drops cannot be predicted ahead of time by the players, they exhibit no observable statistically significant pattern, and they are not correlated with each other in any way. Basically, you can watch crystals open one after the other until the game shuts down, and you will not be able to either predict the next crystal nor state that any one particular outcome is more likely than any other.
That's all we need for random crystals. Can't predict. No one outcome is more likely than the other. And watching no reasonable sequence of any number of openings helps you find any pattern in the drops. The same randomness we see in slot machines in Vegas that underlie billions of dollars of transactions a year.
AA x2
WS x2
Thor x2
WM x1
RR x2
X23 x2
Groot x2
Mr Neg x5
Gorr x3
QS x1
Galan x1
Valkyrie x1
SW (classic) x1