Duelyst Forums

Replace Mechanic Broken - PLEASE LOOK INTO

#22

MATH POLICE, REPORTING FOR DUTY

Combinatorics are not my strong suit (I’m more of a physics/calculus person), so please point out any mistakes I make. Lets assume that the draw was random each time and calculate the probability of none of the next draws being exactly the replaced card. I will be generous and assume that no additional card draw happens and the average game lasts 8 or so turns. In this hypothetical situation, the player replaced once each turn and draws no additional cards. Also, I will be considering the case where you draw EXACTLY the card you replaced, not another copy of the same card.

Since you draw 5 at the start of the game, the starting deck size is 34 cards. When you replace a card, your deck size does not decrease. Now, the deck contains 33 regular cards and 1 freshly replaced card. As such, the probability of not that freshly replaced card is 33/34.

On the next turn, a similar situation occurs. Now, the deck contrains 32 regular cards and 1 freshly replaced card after you replace. No, the probability of drawing a regular card is 32/33

So, the probability of NOT drawing a freshly replaced card in an 8 turn game is

(33-0)/(34-0) * (33-1)/(34-1)…(33-7)/(34-7)= 0.7647

Which is equal to about 0.7647, which is very close to 75%. In other wards, around every 1/4 games, it is expected that you will draw the exact same card you replaced on that turn at least once in one game. This is far from improbably and does not even factor multiple copies of the same card in the deck. This around 25% chance of this happening at least once each game combined with conformation bias (people remembering the times where it did happen over the times where it did not) probably spawned these threads

Now, onto the next phase. I’m still working on this, but I’m playing a 1-of-each deck in practice mode and recording my replaces to see if the result matches these numbers. So far, this theory looks like it’s to be debunked.

6 Likes
#23

Good on you for bringing up the concern to CP, I’m sure they’ll patch it if there’s any truth to it.

#24

this right here is why i love this community so much. hell yeah duelysts :smiley:

#25

The guys a genius

#26

For shure - but nobody was talking about “once in a game”. :stuck_out_tongue_winking_eye:

#27

It’s called fact checking and it’s a method that allows to discuss about something real, not an impression or a feeling but using genuine data.

#28

No, it’s rhetorical because the only thing we all know is the fact we do not know the facts.

But I’m out now. As for me: Replace mechanic is ok the way it is.

#29

wait no, asking for an exact number isn’t the equivalent of “shut up” it only feels like that because people are inherently lazy to test things out that takes time and doesn’t benefit them

#30

Yeah well, same thing really, who the fuck has the time to get > 1000 samples of that, which is bpring and doesn’t get you anywhere, when CPG can just look at/tell us the code for it.

As for the rest, pretty sure, that this is just confirmation bias, or working as intended, i don’t think they would severely fuck up sth. that is as closely related to the draw mechanic.

#31

:warning: This is a sample of how the experience could be conducted, NOT actual numbers :warning:

It’s the opposite. If someone join the discussion with :

  • made 20 matches in practice, took me 2 hours
  • test deck was only combination of 13 sets of 3xsame card
  • replaced a total of 100 cards
  • over these 100 replacements, got the same card 65 times !
  • according to these results, something is really broken

That would be real data, interesting, and concerning.

That is another way to say “Shut up”. :wink:

1 Like
#32

No. That’s the opposite. It says: “I shut up now!” :stuck_out_tongue_winking_eye: :kissing_heart:

#33

that’s interesting, something might actually be broken.

Illuminati confirmed.

#34

Those are some might fine numbers. But you still missed the heart of the question: how to apply this to Astral Crusader?

#35

I thought the in-game text said somewhere that replacing will never return the same card as was replaced.

#36

Correct, when replacing during your turn you cannot get back the same card you replaced. it is possible to receive the same cards from your opening mulligan or when you draw for end of turn however, and it is the same possibility as receiving any other card.

#37

Oh you are going to do this? Ok, make sure to replace LAST, as the last thing you do. It sort of seems that replacing then ending turn has a higher chance of you drawing that card back.
Just like in multiple rng picks in games if you spam-click (and mostly in older games) you are more likely to get the same result twice.

Also, don’t know which RNG they use, but some RNG’s are known to be ‘too streaky’.
There are papers etc. online about it.

1 Like
#38

Knowing me, I would probably procrastinate to no end. Thanks for the advice though! I never knew about the spam-click thing but it makes some sense. If that is the case, it should probably be patched ASAP. I’ll do some research on RNG in the mean time.

Again, thanks for the advice!

1 Like
#39

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.