Possibly wrong listed chances for Piscea

Platform, device version and operating system:

Windows 11

Screenshot or image:

What you were expecting to happen, and what actually happened:
Skill text of Piscea’s Fish Sign says “2 independent 30% chances to devour”.

I cast it three times in row, on these two stunned Takshakas and devoured none. (Got one on 4th cast).

If I count right, chance to not devour anyone is 0.7x0.7=49%.

So chance to repeat it trice is about 12%. One time per eight cases, at most.

And today I seen it four times already (I am battling in PvPs Geheron with Blue restriction). In fact, in about half of my battles it didn’t work from 2nd or even third cast.

Am I calculating something wrong, Piscea has not 30%, but significantly less, or I am simply so unlucky today?

How often does this happen? When did it begin happening?
Noticed it today.

Steps to make it happen again
Use Piscea in the (any?) battle.

1 Like

That’s the problem with RNG, you can never be quite sure. The only practical answer is to actually document your attempts (spellcasts vs. devours) and watch how the data piles up over time.

2 Likes

Instances of chance is not the same as X in Y times.

So if you have cast it 8 times, there is no guarantee you will devour in those 8 times. Each cast is one single instance of chance. The more you cast it, it does not affect any future instances of chance.

3 Likes

Good time to learn how statistics work.

Nah, I was playing against teams with it on, and every time it cast, it devoured two of my troops. So, therefore, it has a 100% chance to devour on every cast.
(Source: Trust me, bro)

No, but seriously, flip a coin four times in a row and you could get four tails or four heads. That doesn’t mean that is the actual odds of getting that result. To accurately work it out, you need to record large quantities of examples and use that to work out the chance. Using just a handful of cases can be wildly misleading.

Exactly.

3 Likes

Roll a dice. Just because there’s a 50% chance to get heads or tails, there will be times when you just keep getting a Bulette head over and over again, instead of getting a Piscea tail.

This is what drives gamblers crazy. They think they have better odds of winning when they keep losing…

3 Likes

There’s some meme about a surgeon telling a patient that they need a surgery that has a 10% chance of success, but not to worry because the last ten times he’s done it, he’s succeeded, and how different people react. Some people will think their surgery is bound to fail, because he’s already succeeded a bunch, whereas others will think their surgery is bound to succeed, because he’s on a roll. The reality is you still have a 10% chance, in theory, because the success of the other patients doesn’t affect you.

2 Likes

Documenting attempts on a human coded project is a fool’s errand. No matter how many attempts someone does, someone else will always say it’s not enough sample size. Then you’re asking people to do it against computer programming which is prone to human errors as well. I don’t agree with it

In reality, probability works as Jeto states.

However, there’s also never full certainty that things are always properly programmed from the outside. Because even if there was an error, most development teams will probably never acknowledge it. All we have on the outside is that the placeholder text says it should do something.

All it takes is a misplaced typo to alter an outcome. See: Aliens: Colonial Marines enemy AI getting wrecked by a single typo. I’m sure some people were saying the AI was working as intended as originally released :face_exhaling:

Do I think Piscea has an error? Very likely no.
But the other direction where people immediately handwave everything as working as intended can be very concerning..

1 Like

What’s the alternative, an AI / LLM (“vibe”) coded project? :rofl:

Yes, I am aware about big numbers and RNG swings. So I don’t proclaim “it’s wrong”, but only suggest it. And yes, of course it was much more than 8 casts - I believe, I played about two dozens of battles, give or take, most goes with much less chances than I expected it.

But then I noticed that in a half of such cases I played against heroes (can anyone tell me how to find if hero is immune to devour? I never was able to discover it) and immortals (obviously immune), and not always enemy has the full cast of troops, so, without knowing how “two attempts” work, it’s quite useless to count chances. If, say, it randomly choose 2 slots from 4, it’s one scheme, if it’s 2 separate tries against full cast it’s another, and if it’s two attempts against existing troops (which I subconsciously expect, but as GoW shown, it’s totally not always right) - it’s a totally different case.

So no, I am not proclaiming anything. :slight_smile:

Several Hero classes gain access to the “Fortitude” Talent at high enough Class Levels; among the “grab bag” of immunities it offers, Devour is included. You’d have to check the individual Hero’s talents to be certain, but it’s tolerably safe to suggest that any eligible Hero is going to have it equipped. Elementalist and Titan are two of those Classes, and those are two Classes you’re going to see a lot on opposition Heroes.

1 Like

Which is better than what you mentioned in the original post, but probabilities are wonky at times, so even that is probably too small a sample size to be sure, and it sounds like you weren’t tracking it so much as just getting a feel for how often it worked, which is open to biases. I’m just saying it’s impossible to tell how accurate the odds are that way.
I agree it often feels like the RNG numbers don’t match up with the descriptions, but without massive numbers to be sure, we can’t know how close they are. But it’s possible the game is rigged. We know they deliberately rigged dungeons against players, going out of their way to code them in a way that screwed players over, and it was only because players gathered huge numbers they were able to prove it and force the devs to change it. So, it does happen. I don’t know if this is one of those cases or not.

It would be in their traits. To find them, select the enemy hero (I’m on Playstation, so that would be R1 and then the relevant button), scroll over to the third page, and you’ll see the three main traits. On the left hand side, there should also be a dark panel with their class. It’ll say their class level and champion level for that class, then if you scroll down, below those will be all the unlocked traits for the class (on Playstation, use the left stick).
As was said, if it has Fortitude, it’s immune. Otherwise, you just have to look what traits it has, and they state what they do, so you can see what it’s immune to.

As I understand it, it picks two at random immediately, but you’re right that it’s hard to tell, and it might be picking the same troop twice. It shouldn’t, but in this game you never know. Usually if it’s doing two things at once, it selects two different targets, and the spell seems to be doing that, but it’s hard to be sure. If it was selecting the same troop at all, that should show in the percentage chances, so if would affect the numbers in a large enough pool of data, but it would be possible to find out, if we had that data.
But also, yes, if any troop is immune, it can select that and fail at devouring it, even if it succeeds at the roll. But in that case, you will see on the troop that is selected, the name of their trait (the one protecting them from devour) will pop up on their troop image on the right of the screen. You may have to record your screen and watch it back to be sure, but you can use that to tell whether or not those troops have been selected and the troop did attempt the devour and get blocked by immunity. It could be clearer, but at least it’s something.

1 Like

I think Bless also blocks Devour? Be watchful for that, too.

2 Likes

What you do have is a 100% chance to have a 70% chance to fail every time you cast it

N this is why hki is so much better you can make his a 100% chance to be a 100% chance to devour