Jump to content
CeladonMD

[GAME] General - There may be an issue with the shuffling system not being random

Recommended Posts

overcast_kid

Well this got thread got moved? why?

Share this post


Link to post
Share on other sites
Bobix86

It's very interesting that you brought this up right now CeladonMD, as i have been having a mostly ok experience with shuffles up until this month when the game started to almost constantly "rubberband" me (got to learn a new word here, yay), anyway i fail to get the cards i need (plus getting a bad starting hand) now in 80% (or at least very close to) of matches played.

This is also an excellent topic to keep on pressing for a true rng implementation. Now i did try to point out that a true rng would be much better for this type of game, but i was disputed as it was a topic regarding only coin flips and i didn't care to elaborate further. I don't know what rng they are running on this game, but i'm fairly certain it's a pseudo rng, probably not a very good one from my experience. The problem is when you use pseudo rng's, they reset and go trough same algorithms each time you request random numbers/strings ( hope i got it right, i'm not an expert but i have been following the outcomes of many games rng's just as any inquiring player), while a true rng reads from atmospheric noise or similar truly random source and then it takes the results trough those algorithms thus getting better (more random) results. Now there is a service that provides truly random numbers, random org, and if implemented right, it would make for much better gaming experience with very few if any complaints. But the implementing should be done so that it contacts the site for every event separately, meaning when you get the starting hand, it contacts the site, when you draw card/cards (excluding specific cards) from the deck, it contacts the site,when shuffling the deck, same, for each coin flip separately, it contacts the site for random numbers. Reason for this kind of implementation is that atmospheric noise changes with time, and so do the readings, so even those few milliseconds it takes to get the numbers might make the difference and better randomness. Hope i explained it right, and that players, even mods/developers) understood me. I'm aware that the load might be potentially high, and that it would be a payed service, but we would certainly imo, get a better rng, and it would be worth the money.

To get back on the topic, i would say it wasn't biased if you asked me 10 days ago, now i'm not sure what to say, something has changed at least for me. I still feel we wouldn't have a reason to complain if the game ran a true rng.

 

To name a couple examples from what i remember:

 

1. This happened just right now so it's fresh in my memory, i just play tested a fairy deck, though i won 6 times in a row, 5 out of 6 times only pokemon i had in first hand draw was spritzee, luckily i had gotten supporters so i made it trough. Have 12 basic pokemon in that deck in case anyone wants to do the maths.

 

2. Yesterday i was playing against a deck (don't remember which think it was some dragon type) with my dark discarder deck, opponent was on his last pokemon but had only 1 prize left to take, i had houndoom with dark roar, he had two cards in hand, an energy and another card which he didn't have use of i know cause i used ghetsis, he only needed that one energy to win otherwise i'd have knocked out his last pokemon on the next turn, guess which card was discarded.

 

Now there was much more "bad luck" that didn't involve me winning. But curiously it kind of started this month, call it selective memory, but i really don't remember noticing these excessive bad draws/luck until this month, and no, i didn't stumble upon this topic until today.

Edited by Bobix86

Share this post


Link to post
Share on other sites
Stratadrake

1. This happened just right now so it's fresh in my memory, i just play tested a fairy deck, though i won 6 times in a row, 5 out of 6 times only pokemon i had in first hand draw was spritzee, luckily i had gotten supporters so i made it trough. Have 12 basic pokemon in that deck in case anyone wants to do the maths.

How many Spritzee were you running?  If I assume four, then:

 

- Chances of having no Spritzee in your opening hand = chances of drawing seven non-Spritzee cards = 56/60 * 55/59 * 54/58 ... 50/54 = ~60%

- Chances of having at least one Spritzee = 40%

Edited by Stratadrake

Share this post


Link to post
Share on other sites
Bobix86

How many Spritzee were you running?  If I assume four, then:

 

- Chances of having no Spritzee in your opening hand = chances of drawing seven non-Spritzee cards = 56/60 * 55/59 * 54/58 ... 50/54 = ~60%

- Chances of having at least one Spritzee = 40%

It would be ok if it was four but i'm running two, reason is i only have two aromatisse.

 

Anyway i remember doing about 20 more matches, same deal, i got spritzee only, 75-80% of the time, now tell me that's normal. Ain't, right?

Edited by Bobix86

Share this post


Link to post
Share on other sites
overcast_kid

Id also like to being up something I've noticed, when you have a type advantage the Game tends to let your opponent draw the necessary cards to bench more pokemon/ draw more useful cards. Like the game won't let you get a one hit ko first attacking turn. (Not a one turn win but not a first turn KO)

Share this post


Link to post
Share on other sites
tarif15

to add up to this discussion , i played my newly made yveltal deck and i forgot to add 1 jirachi ex :) before i added the jirachi ex , i played about like 6-7 games and i realized i didn't add it (cause i wasn't lacking supporters obviously :P ) but when i realized it and added the jirachi .... guess what happened . played 3 games after the addition of jirachi ex and they give me ONLY jirachi ex in hand as a pokemon . i dont think thats NORMAL :/ before the jirachi ex addition , i had gotten yveltla , yveltal ex and toad ex in my first hand .

 

so the question is .... does ptcgo have a bad shuffling system ?

 

OR does the game simply like cards like Shaymin ex (RS) , Jirachi ex , spritzee (as stated above) , celebi ex (u get the point :P ) ?

 

btw , i had about 11 basic pokemons BESIDES the 1 jirachi ex .

Edited by tarif15

Share this post


Link to post
Share on other sites
overcast_kid

Also not to be all conspiracy-guy but why did this topic get moved? Should be in game feedback. Not pokemon video games.

 

"Game Feedback

Share and discuss your thoughts on the Pokémon TCG Online."

Edited by overcast_kid
  • Upvote 1

Share this post


Link to post
Share on other sites
Stratadrake

It would be ok if it was four but i'm running two, reason is i only have two aromatisse.

 

Anyway i remember doing about 20 more matches, same deal, i got spritzee only, 75-80% of the time, now tell me that's normal. Ain't, right?

With only two Spritzee, your chances of getting at least one in your opening hand should be only about ~25%.  Yeah, I'd say that sounds broken.

Share this post


Link to post
Share on other sites
CeladonMD

Fellow Trainers,

 

  Thank you for your thoughtful contributions. I am taking an extra amount of time to perform a proper statistical analysis of my observations as several of you have supported doing so. I studied enough stats as an engineer and med student, hopefully it will be useful in some way. 

 

Please continue to relate any new experiences on biased shuffling and card draws. 

Edited by CeladonMD
  • Upvote 1

Share this post


Link to post
Share on other sites
arklumpus

Meanwhile, I did some tests too, and wasn't able to find any statistically significant deviation from randomness...

 

 

 

I used the following (test) deck: 1 Pikachu (XY 42), 10 Grass Energy, 49 Fire Energy.

 

 

 

Over the course of 19 games, 132 hands were drawn (19 of which had a Pokémon in it).

 

 

 

Expected values were obtained through ******* Carlo [i don't get why random words are banned, in this case it was the Italian/Spanish translation of "mountain"] simulation (didn't have time to do proper computations .-.).

 

 

 

The results were the following:

 

 

 

Hands without Pikachu:

 

- With 0 Grass energies: 34 (EV: 29.3)

 

- With 1 Grass energy: 43 (EV: 47.8)

 

- With 2 Grass energies: 30 (EV: 29.3)

 

- With 3 Grass energies: 4 (EV: 8.7)

 

 

 

Hands with Pikachu and/or with more than 3 Grass energies: 21 (EV: 16.9)

 

 

 

χ2 ≈ 4.8, with 4 degrees of freedom; this means that the probability that the observed distribution be coherent with "true randomness" is about 31% (which is actually not enough neither to completely exclude bias, nor to definitely say there is one). Moreover, note that most of the observed difference comes from the datum with the lowest expected value (3 grass energies), and this further substracts to the credibility of the bias [incorporating this with the "other" data group gives a χ2 ≈ 1.3 with 3 degrees of freedom, which translates to a probability of "true randomness" of about 73%].

 

 

 

More data would probably allow us to arrive to a statistically significant conclusion, but without clear signs for a bias I don't think it is worth to spend more time testing with this deck...

 

 

 

Finally, note that this test does not exclude an "intelligent" bias (which has been previously called in this post "rubber banding"), as to test for that would require a more adequate deck (if you have one to suggest, I will be happy to test that as well, if I can get a hold of the required cards).

 

 

 

These are the hands I obtained (each hand is represented by a single digit signifying the number of Grass energies in the hand):

 

Hands without Pikachu: 12100021102221111102010112122002211110011101212012211301300202010100120120213022144022211121222020101001013101001

 

Hands with Pikachu: 1020012202012010311

Edited by arklumpus
  • Upvote 1

Share this post


Link to post
Share on other sites
overcast_kid

I want to be very clear my stance on this shuffling. Im not saying that the shuffling is "broken" as in the random cards drawn is favoring one or another. Im saying I think there is a deliberate formula put in by pokemon tcgo from the creators to have rubberbanding. When we look at our own hand it will of course seem random but we should take into account what our opponent has, his options with the cards he can play vs what we have and can play. 

 

Play a game and pay attention to what you have in your hand that can be played onto the board first turn for instance.

 

Lets say both players at the beginning put one pokemon into the active postition and no benched pokemon.

Now say on the first hand you put down thunderus ex with raiden knuckle and muscle band and one energy-turn 1

your opponent puts down a 50 hp pokemon and would have nothing to stop this attack

your opponent will it seems be gauranteed a way to bench another pokemon next top deck or draw something to disrupt like laser that will i assure you be heads to make your pokemon sleep.

 

Also another example is malamar with flips attacking. I think the numbers of heads flipped to do 60 damage is proportional to how much damage the opponent has from zero. A fully healed 180 hp pokemon vs a for example 20/180 hp pokemon has better odds of you flipping not enough heads to knock it out.

 

I have tested chillmax vanilluxe with over 15 energies attached multple games and only been able to knock out a fully healed ex pokemon in one attack like never. 

 

I think this should be looked into, if its true then this whole game would be an insult to our intelligence and more importantly our wallets.

Edited by overcast_kid

Share this post


Link to post
Share on other sites
arklumpus

As I said, in my opinion finding such a reliable "intelligent" method of balancing the cards dealt would be non-trivial at the very least... Moreover, since this would imply an algorithm to find the best possible thing to do with a certain hand, and if this had been found, the fights against the AI would not be as ridiculous...

 

Anyways, if you'd like to test the coin flip thing, I've got a Malamar EX: I could make a deck with it and dark energies, and we could do a private match (you should use a deck with only high-hp Pokémon...). Would you be available - say - in the next couple of hours?

Share this post


Link to post
Share on other sites
CeladonMD

Of note, for those keeping count, the thread was renamed from "Biased Shuffling System? Please contribute" to "There may be an issue with the shuffling system not being random".

 

Not a major issue, but I know some users are sensitive to such changes. 

 

Arklumpus, a noble contribution, thank you. You do observe though, that this issue of "rubberbanding" was one of the requisites in my original post. Perhaps you could run the same deck in versus, and see if your Pikachu draw rate changes.

 

Keep it up everyone!

There may be an issue with the shuffling system not being random

Share this post


Link to post
Share on other sites
arklumpus

@Celadon: my tests were, indeed, done in the normal "versus" battle (i.e. against other real players)...

Share this post


Link to post
Share on other sites
CeladonMD

@Celadon: my tests were, indeed, done in the normal "versus" battle (i.e. against other real players)...

Yes, wasn't doubting it, but did you identify a change in the Pikachu draw rate after consecutive wins/losses?

Share this post


Link to post
Share on other sites
arklumpus

Ah ok... The problem is then that it's difficult to win a match with a single Pikachu :P

Share this post


Link to post
Share on other sites
CeladonMD

Ah, but it's not hard to lose, no? You could test the potential positive effects of losing on draw rate.

Edited by CeladonMD

Share this post


Link to post
Share on other sites
overcast_kid

bump

Share this post


Link to post
Share on other sites
Mod_Poplar

Hello overcast_kid,

 

Bumping a thread is not considered constructive or on-topic, so it is not permitted. This issue is now being investigated by the Dev team, so I'll be closing this thread.

 

Thanks for your reports! :)

Share this post


Link to post
Share on other sites
Guest
This topic is now closed to further replies.

×
×
  • Create New...