Author Topic: Bad luck with RNG  (Read 1606 times)

Offline Koey

  • Hero Member
  • *****
  • Posts: 2020
  • The one trapped in the Hamster ball.
    • View Profile
Bad luck with RNG
« on: May 16, 2013, 06:49:26 AM »
I must be crazy, but does anyone notice sometimes in their starting hand or when you use trading post you get the same card?
Probably the RNG is not on my side me but I have always noticed non-limited cards getting stuck a lot since alpha. Especially with the trade.
Usually when I have a bad starting hand I have duplicates of say 2 assassinates, 2 firestarter, 2 aberion, 2 rifts.
I also had a case today where I had 20 cards left. Used trading post twice on my Rift and got back another copy of rift twice. I have noticed trading post has this happen quite a bit. (Given I had 2 copies RNG must be not on my side)

Anyone else notice a trend like this, mostly with the trade function?

Its also something I noticed when I use Fill the Grave from time a lot of my cards would be one followed another. (ie. 2 Winds of War stuck together)

Mostly noticing for Limited cards. Was just curious if anyone else noticed it. It almost feels like the deck was built organized nicely and then wasn't shuffled properly. Have been noticing more when trying to build a deck with no unlimited character.

Offline Teremus

  • Really Cool Moose
  • Hero Member
  • *****
  • Posts: 5545
  • Hide your cheese.
    • View Profile
Re: Bad luck with RNG
« Reply #1 on: May 16, 2013, 07:09:37 AM »
Intriguing observation. Who else wishes to add to this?
No seriously, I really like cheese.

Quote from: Poga
I make this game, James.

Offline MrFahrenheit

  • Jr. Member
  • **
  • Posts: 92
    • View Profile
Re: Bad luck with RNG
« Reply #2 on: May 16, 2013, 07:30:36 AM »
I did have one case where I was using the send one card back to the deck and draw another function in the trading post. It was either two or three in a row where I sent back a Wealthy Noble just to draw a Wealthy Noble. Probably just the RNG playing tricks on me but it did have me laughing while I died from lack of removal.

Offline Teremus

  • Really Cool Moose
  • Hero Member
  • *****
  • Posts: 5545
  • Hide your cheese.
    • View Profile
Re: Bad luck with RNG
« Reply #3 on: May 16, 2013, 07:57:09 AM »
A wise man once said: "When something is truly random, it does not look random."
No seriously, I really like cheese.

Quote from: Poga
I make this game, James.

Offline Calamity

  • Full Member
  • ***
  • Posts: 249
    • View Profile
Re: Bad luck with RNG
« Reply #4 on: May 16, 2013, 08:46:43 AM »
Had those situation once or twice in almost every game, but not enough to see any flaws or some sort of pattern within the trading post card shuffle option.
Just a poor asian college student in the middle of an island in the pacific living of Cup Noodle and Spam.

Offline Jade

  • Sr. Member
  • ****
  • Posts: 358
    • View Profile
Re: Bad luck with RNG
« Reply #5 on: May 16, 2013, 02:19:07 PM »
Intrigued by this, so I conducted an experiment.  It's far from perfect, but it gets at the issue.
  • I created a deck with 3 limited Verore cards, 3 Flame Dawn, 3 Warpath, and 31 Conscripts and entered a game with Puffy.
  • As quickly as possible I traded away card until I was left with all conscripts except 1.
  • At this point, I traded the unique card, and recorded what I got.
  • I kept doing this until I only had 20 cards left in my deck (to prevent the draw probabilities from becoming too large) and played 3 games like this.
  • Probability of drawing the same card ranged from approximately 6.7%-10% (2 out of 20-30).  Probability of drawing a different non-conscript card ranged from approximately 20%-30% (6 out of 20-30).

Over the three games I recorded 38 trades.  Only two of the trades got me the same card back.  That's 5.3% of the trades, which is a little less than the expected rate of this happening.  However, interestingly, those two times happened back to back, so I certainly would have noticed this if it had happened in a real game.

10 of the trades exchanged a non-conscript card for a different non-conscript card, or 26%, which is within the expected range.  The remaining 26 (68%) trades fetched conscripts.

Obviously this suffers from lack of data, but it's interesting that I got almost exactly what I would have expected probability-wise.  If people are interested/still distrustful, I could always expand this.

Offline WWKnight

  • Hero Member
  • *****
  • Posts: 2178
    • View Profile
    • dTCG Sanctum - One stop for all your IW needs!
Re: Bad luck with RNG
« Reply #6 on: May 16, 2013, 04:49:33 PM »
Im not interested, but Im still impressed.  Good job Jade :)
I'm a pretty big deal around here.

Feel privileged.

Offline moominpeter

  • Hero Member
  • *****
  • Posts: 548
    • View Profile
Re: Bad luck with RNG
« Reply #7 on: May 16, 2013, 06:55:57 PM »
Every forum I have ever inhabited that was devoted to a game with an RNG sooner or later has a thread about how how the RNG is behaving weirdly. What's actually behaving weirdly, though, is the human brain.

See, it turns out that humans process information in a way that makes random things seem non-random and (some) non-random things seem random. Once you learn how this works, you can choose to ignore the voice in your head that says "Hey, this doesn't seem random" but you can never really get rid of it.

So, how does it work? Well, when something is truly random there will be times when something hugely unexpected happens. For instance, when flipping a coin you would not expect to flip heads 5 times in a row (this has a 1 in 32 chance of happening over the course of any 5 flips), but if you spend a lot of time flipping coins you will have some crazy streaks of flipping the same thing repeatedly 5, 6, 7, or more times in a row. These crazy streaks register as odd, but actually they are an expected outcome of a random process. Because the human mind is primed to look for patterns (likely for evolutionary reasons, since pattern recognition helps in survival), we automatically reinterpret random events to make them appear non-random where possible, and so with truly random things we can almost always manage to find what look like patterns by paying attention to the crazy streaks.

Back when Blizzard released Warcraft 3 (before they joined Activision and became the company everyone loves to hate) they actually released it with a RNG that wasn't random (or at least, that didn't do what it said it did), but it ended up feeling really random. They did this by preventing highly unlikely combinations of events. So, for instance, one hero character had a 10% chance to score critical hits (with 1 rank in the relevant ability). Rather than actually give the unit a 10% chance to score criticals, though, blizz made it so that every time the unit critted its chance to score another crit dropped substantially, while every time it failed to crit the chance of the next attack critting raised a bit. Over the course of a full game, it would always score very close to 10% of its attacks as crits, and unlikely events like critting multiple times in a row basically never happened. The ability wasn't actually doing what it said it was doing, but it felt like it was--in other words, the non-randomness of it felt more random than it would have had it actually been random.

TLDR; As Teremus said, "When something is truly random, it does not look random."

Offline Koey

  • Hero Member
  • *****
  • Posts: 2020
  • The one trapped in the Hamster ball.
    • View Profile
Re: Bad luck with RNG
« Reply #8 on: May 16, 2013, 07:50:56 PM »
Intrigued by this, so I conducted an experiment.  It's far from perfect, but it gets at the issue.
  • I created a deck with 3 limited Verore cards, 3 Flame Dawn, 3 Warpath, and 31 Conscripts and entered a game with Puffy.
  • As quickly as possible I traded away card until I was left with all conscripts except 1.
  • At this point, I traded the unique card, and recorded what I got.
  • I kept doing this until I only had 20 cards left in my deck (to prevent the draw probabilities from becoming too large) and played 3 games like this.
  • Probability of drawing the same card ranged from approximately 6.7%-10% (2 out of 20-30).  Probability of drawing a different non-conscript card ranged from approximately 20%-30% (6 out of 20-30).

Over the three games I recorded 38 trades.  Only two of the trades got me the same card back.  That's 5.3% of the trades, which is a little less than the expected rate of this happening.  However, interestingly, those two times happened back to back, so I certainly would have noticed this if it had happened in a real game.

10 of the trades exchanged a non-conscript card for a different non-conscript card, or 26%, which is within the expected range.  The remaining 26 (68%) trades fetched conscripts.

Obviously this suffers from lack of data, but it's interesting that I got almost exactly what I would have expected probability-wise.  If people are interested/still distrustful, I could always expand this.

Thanks for doing that. I guess its most likely just bad streak.

One thing that I was curious is how the card was shuffled back in the deck. (ie. Shuffle card back in then draw from top. Or draw from top of deck then shuffle the card back in) - The first will increase the possibility of drawing the same card. While second decreases the possibility of drawing the same card.

Offline Teremus

  • Really Cool Moose
  • Hero Member
  • *****
  • Posts: 5545
  • Hide your cheese.
    • View Profile
Re: Bad luck with RNG
« Reply #9 on: May 16, 2013, 09:17:01 PM »
Very well done Jade.
No seriously, I really like cheese.

Quote from: Poga
I make this game, James.

Offline grahaminator

  • Sr. Member
  • ****
  • Posts: 350
    • View Profile
Re: Bad luck with RNG
« Reply #10 on: May 16, 2013, 09:24:22 PM »
If you really want to test this you need to do it in a highlander deck with Varyus in the command zone. It would take a lot of work though since you would have to notate what all in in hand / play / graveyard and then get percentages based on what is left. This would change every turn as more cards change zones from your deck. Not sure how the randomness is programmed but I remember doing a simple programing project where I did math based on a random function combining it with the exact timestamp. Like moominpeter said sometimes you get things that seem like patterns but they really arent. There is probably a complex shuffling agorithm in place. In real life I plile shuffle 60 card stacks and then mix it up with some mash (weave) shuffles, overhand shuffles, and them some more mash before giving my deck to my opponent.

Offline grahaminator

  • Sr. Member
  • ****
  • Posts: 350
    • View Profile
Hypergeometric Distribution
« Reply #11 on: May 16, 2013, 09:41:40 PM »
Also if you want to find the chance of getting a specific card you can use the statistical spreadsheet function for hypergeometric distribution:

Calcuations based on the function HYPGEOMDIST(a,b,c,d) where
a= Number of a specific card you are looking for
b= Number of cards drawn
c= Number of a multiples of a single card
d= Deck size

Example:

=HYPGEOMDIST(1,5,3,40) which would return 0.30111336

So in plain english this is telling you that you have a 30% chance of drawing exactly 1 of a specific card that you have 3 of in your deck within your opening hand in a 40 card deck. Those numbers can be adjusted based on extra cards drawn and what is left.

If you wanted to see what is you chance of drawing at least 1 in your opening hand provided deck and sample size stayed the same you would use:

=1-HYPGEOMDIST(0,5,3,40) which would return 0.337550607

Offline mace

  • Jr. Member
  • **
  • Posts: 60
    • View Profile
Re: Bad luck with RNG
« Reply #12 on: May 17, 2013, 03:01:49 AM »
I had it happen with twice a card turned in there was only 2 and got same card traded in.