Jump to content

RNG Experiment: Attack Coinflips


  • Please log in to reply

17 July 2017 - 12:48 PM

#101

harshu

    Elite Trainer

  • harshu

I don't know about what you have learnt on the randomness being produced by the computer. But it is definitely possible to generate something random for 1 time and then the pattern is formed. You wouldn't need an algorithm to form a pattern, because then it wouldn't remain random. This can be done by using the mechanical disturbances within the server sometimes called as inefficient disruptions which are used to create randomness. Because these are created by disruption, they follow the the code properly and only evade from the desired result by the use of those mechanical number. This is how randomness is generated. But due to the inability to fully use this disruption, computers and codes or anything digital can never be 100% random and hence pretty much predictable.

 

 

On a side note, always keep count of the heads and tails in a game and even more importantly what the last flip yielded ( I start my day with a heads ) --> Rest all matches, till you are flipping the coin, you are likely to win the coin tosses at the start of the match xD ! ;)


Probably the MOST HATED WRITER on PTCGO !
If you need any Deck help just visit this thread - The Best Decks in TCG - http://forums.pokemontcg.com/topic/28694-the-best-decks-in-tcg/
If you are new, and looking for a Beginner's Guide, click the following topic, to get reach to a Helpful Guide. - New Player Guide - http://forums.pokemontcg.com/topic/31931-new-player-guide/
Click the names, and you can reach those Topics :)

  • 0

04 August 2017 - 05:01 PM

#102

Synn88

    Rookie Trainer

  • Synn88

I have only been playing for a few months but I have to say that as far as my account goes there seems to be some variance from your collected data, especially withcrushinghammers. I have collected some data, although not as thorough as yours and have noticed that my experience is dramatically different. I used a Dragon deck with as many coin flips as possible including Kingdra, Crushing Hammers, and other flip attacks to get as many coin flips as possible. My collection is limited, so I could not build the deck described in the initial post.

 

What I have discovered is as such:

 

Through well over 100 matches I received the following flip results:

Noibat- H-16 T-53

Zweilos- H-24 T-32

Skrelp- H-11 T-25

Horsea- H-3 T-19

Seadra- H-25 T-19

Kingdra- H-36 T-72

Crushing Hammer- H-21 T-102

 

This does not take into account flips initiated by opponents cards. 

 

Although some of the cards seem to be within the acceptable range there is a very clear discrepancy when it pertains to Noibat, Skrelp, Horsea, Kingdra, and especially Crushing Hammers. The overall flip ration is horribly off as well. 

 

My wife and 2 sons also play and don't seem to have the issues I am. I also have a startling number of seemingly impossible draws. Not so much with mulligans, but frequently starting with cards I only have 1 or 2 of in opening hand, including but not limited to Kindra and both of the electric energies at once (only have 2 in deck because Kingdra is only card that uses any and requires only one).

 

In light of my experiment I have 2 questions:

 

1) Is the RNG run Client or Server side?

 

2) Is it possible that there is a glitch affecting my account (as this happens to me on my Desktop, Laptop, and Tablet but does not seem to affect others that use the same devices)? 

 

I am completely prepared to accept that I simply have ridiculously bad luck. I haven't gotten a GX or EX in my last 35 booster packs. lol

 

Thanks for your time and consideration.

 


  • 0

08 August 2017 - 05:36 PM

#103

Startropic1

    Junior Trainer

  • Startropic1

This does not take into account flips initiated by opponents cards.


This is a flaw in your data, particularly instances requiring YOUR coin to be flipped.

I rather doubt that the coin flip algorithm factors in the card that initiates the coin flip. This would require a lot more code, and thus more room for error, than simply coding for the coin itself whenever it is needed. Coin flipping is a very simple thing to code, and such programs have existed nearly as long as computers have existed. Trying to reinvent the wheel would be a waste of time.

A more interesting question would be wether there are separate coin flip algorithms for the two players or just one that they share. This would make a big difference in regards to how one should examine the data. It's also more playable than what you're suggesting.
  • 0

29 March 2018 - 04:43 AM

#104

jcgss77

    Expert Trainer

  • jcgss77

My hat off to Sakura for an EXCELLENT bit of research here.  You did very well, and your findings are very interesting and expected.  Thank you, even though this is a very late thanks.  This thread is by far pin-worthy.

 

I did not read the entire thread, so if I missed anything I am about to mention, I ask for forgiveness beforehand.

 

As I am not complaining of a broken RNG or anything like that, I have noticed what does seem to me to be things to look at, and even for those who have much more patience with this game than me, to put into implementation and record for the sake of the truth.  So here they are:

 

-RNG during peak hours and off-peak hours in each format-I have experienced consistent card-jamming in Standard during loaded up servers, which is why I did not play Standard much, if at all. (I say did because I have not played a Standard match in probably 3 years I do believe).  I personally flip mostly tails during sleep and burn recovery, but experience "making up" flips when my opponent makes recovery flips from statuses which I inflict, so it does even out.  I do believe there is an "evening out" mechanic in the RNG, as truly pure RNG is not possible in the world of coding.

 

-using different coins from different eras-I find that I have horrible imbalance with what I consider major trending in coin flips-runs of heads and tails-with older coins from the older sets.  I have a red Pikachu coin which I have very good fortune with, I do flip more heads than tails most of the time, from a Worlds code card, which I incorporate in many of my decks.

 

-I also experience when I play in Legacy matches, the tendency for runs of flipping the same side repeatedly.  Again, I do believe there is an "evening out" mechanic which brings the flips closer to 50/50 in play in the RNG.

 

-I have experienced in both Standard and Expanded a tendency, which does not happen frequently, but when it does happen, I will get hands full of energy cards, while obviously possible, but is very unlikely, especially in games back to back.  Again, this does not happen frequently.

 

Now please read this post with a grain of salt, as I am definitely far too lazy and never play more than 3-4 games a day on the PTCGO, if I play at all, to do an actual scientific study, recording data.

 

So, if anyone else is interested in furthering this excellent example of work, I recommend the following trials:

 

-have a deck in each format, with attack flips, status flips, and ability flips (as much as is possible, obviously)

-use the same coin in each deck while playing at different times, using extremely old coins, newer coins, then new coins

-don't just record all flips, but different categories of flips-opening flip, attack flips, ability flips, recovery flips

 

Maybe I will go ahead and do a similar study, and if I do, I hope to be even 25% as thorough as Sakura.  If I do, I will definitely post the results.

 

So the purpose of this post is to encourage others to do likewise, and to share what is experienced.  And if you do, thank you!


Edited by jcgss77, 29 March 2018 - 04:45 AM.

  • 1

20 September 2018 - 06:00 AM

#105

ufa007

    Rookie Trainer

  • ufa007

nice post  

[url=http://www.ufa007.com/ufabet]ufabet[/url]  

[url=http://www.ufa007.com/ufabet]สมัคร ufabet[/url] 



  • 0

06 October 2018 - 03:56 AM

#106

kathdyqq

    Rookie Trainer

  • kathdyqq



so good [URL="http://www.ufa007.com/ufabet"]พนันบอลufabet[/URL] :DY



so good [URL="http://www.ufa007.com/"]ufabet[/URL] :DY


  • 0