Anyone else noticing Devour doesn't work right in pet quest?


#1

So, I’ve been using Kerberos to get through the pet quests; it’s basically the only way to do them if you don’t have a perfect loop team yet. I’ve been noticing though that something seems a bit off in this game mode compared to other modes. I am very aware of how often the devour spell works, because it is a marathon of 8 matches on a timer. so I’m paying very close attention.

What I seem to be noticing though, is that Devour seems to work roughly once every 4-5 casts, which would put it at a consistent 20-25% success rate, where it’s listed percentage is 40%. Is this happening with anyone else?


#2

I think Great Maw 100% devour rate one time cast is better for this than Kerberos at 40%

Great Maw
Mercy
Anything
Anything


#3

Rate does seem a bit down for my doomclaw and kerb. But then he did two devours on stage 8 for a huge win


#4

There is no difference in the devour chance between pet rescues or any other game mode. :slight_smile:


#5

RNG either loves or hates you


#6

In that case, it must not be working right in any mode. I don’t track devour success in the other modes as closely, but in pet quests I am watching quite closely because I’m on a timer and have a set number of fights. After doing these for a while, the lower success rate than listed on the card are pretty consistent.


#7

What exactly is your data set?


#8

I have to say I have never in my life seen a Kraken devour…
Even though I know it can.
Devour chances too small to matter anyways.
During pet quest if I get goblins 4x gobchomper retreat if none devour, lol.


#9

As i said, I am observing a successful devour roughly every 4th to 5th cast on Kerberos. This can vary slightly, as random chance is like to do, but the 4-5 casts per devour has been roughly consistent. Given that Kerberos has a 40% devour chance, it should activate roughly every 2-3 casts on average over time.


#10

It’s the same as any other “I didn’t see it trigger enough” post.

Take a coin out of your pocket or wherever you have a coin. Flip it once. Now flip it again. Did you get the same thing? Flip it a total of 10 times. Did you see 5 heads and 5 tails? If so, make your way up to 20. Is it still 10 by 10? Keep going, 10 tosses at a time, until it’s not.

There will be a number where you don’t have 50/50 in your results. That’s because there isn’t a Probability Fairy who makes sure the coin balances. Gambling wouldn’t work if there were a Probability Fairy. You could watch a roulette wheel long enough to figure out some number or other is “expected” and break the house. Same with slot machines, you could carefully count pulls without a jackpot and only choose to play when the Probability Fairy is about to force a win.

Now, one can argue Poker or Blackjack are subject to card counting. That’s because dealing cards out of a deck is NOT an independent event. Every card that’s dealt changes the deck, so the next card has more well-understood probabilities. If you know there are 4 decks, and you’ve seen three of a particular card dealt out, you understand the odds of getting that card are extraordinarily low.

But Devour is not a poker game. Every devour is independent. That means even if you have a 90% chance, it’s possible to fail 2 out of 3 attempts with a not-insigificant probability. If it were NOT possible, the RNG would be just as broken as if it were common.

This is very hard to understand. Our brains do not intuitively understand probability. This is why gambling works and why casinos and F2P games are very rich.

Even 10,000 trials showing a large aberration is not enough to prove a problem. It takes several sets of players showing 10,000 trials going wrong to suggest a problem. The larger ‘several’ gets, the lower the probability it could happen and the more the problem is suggested.

So when we’re talking dozens of pet rescues, or even hundreds, I’d argue it’s very likely we could find 10 people who think devour never works on the forums. Sometimes you have to be really lucky to encounter a form of bad luck.


#11

Well sure, over time. I mean, how many trials have you tracked it for? Hundreds, or “a bunch”? Are you writing down every piece of data, or eyeballing it and relying on your memory?

I ask because, historically speaking, every single time this has ever come up before, anecdotes have always been “the chance is broken!”, followed by someone gathering hard data, followed by the data proving that the chance is exactly dead-on accurate over time.


#12

I’ve been looking at the over-time, not the per-match average. Obviously you can have two successful devours in a row in a single match, or fail every time in a match. This is why I’ve waited for pet quests to be out for a while before making this thread. I’ve been watching the average over a length of time. It’s coming out to around 4-5 casts per devour on average. the point of this thread though is to ask if other people are having the same experience or if I’m coming up with a lengthy set of unusual results.


#13

The only devour troop I use in Pet rescues is Gobchomper and I am not seeing anything that causes me any concern with respect to the 50% devour rate.

Outside of that, my feelings follow @Amadan’s.


#14

Oh? Gobchomper is proof of how screwy the rng is on xbox. With 4 Gobchomper, 1 in 16 chance they all miss. Complete miss way more often than that. Over time the miss / devour ratio might be equal, but it’s extremely streaky.


#15

You have accurately defined a RNG. ;p


#16

Nope, it has to also follow a normal distribution.


#17

The only thing a percentage-chance random system has to do, in order to be working correctly, is produce its expected percentage of results over time. That’s it. Real randomness is streaky and clumpy - and then nice and even - and then streaky - and then nice and even - and then a big burst of streaky! - because it is, by definition, random.

A lack of streaks and clumps is, in fact, a sign that a system is being deliberately tampered with.


#18

When we’re talking about probability, nothing is proof of anything. The OP asked if other people are having the same issue as he is. I am not, and said so. No one else has jumped in with any impressionistic data and the OP hasn’t provided his data so I’m satisfied that, so far as percentage chance devours go, things continue to be fine/normal.


#19

Normal distribution is also necessary. I did some lengthy testing, which took a while because pet quests don’t exactly happen constantly, and in the long run it “technically” balances to 40% but it does so with streaks of 5+ misses, then a string of alternating hits and misses until it catches back up to balance, then misses several times in a row again.

That doesn’t really act much like random chance. It seems like a heavy bias towards missing that is course corrected periodically to get it back into balance. True random would look more like “Hit, miss, hit, miss miss, hit, miss, miss, miss, hit hit, miss” this looks more like “miss, miss, miss, miss, hit, miss, hit, miss, miss, miss, miss, miss, hit, miss, hit, miss, hit, miss, miss, miss…”


#20

This is the thing about random: if it is truly random, you can’t say what it should look like. If you’re getting your 40% on average, that’s all you need to worry about.