PTCGO Are Coin Flips Really 50/50 in PTCGO?

[sandsmash23]

While I don't usually watch Dark Integral Gaming's content, his recent video Are Coin Flips REALLY 50 / 50 In PTCGO? (Funny) caught my eye for resurfacing a complaint often muttered by the client's disgruntled users. As one might expect, even though the video was meant to lightheartedly bemoan a string of bad flips that the channel owner had experienced, the comment section was largely composed of players trashing the client's random number generator because of gripes that seemed more anecdotal than empirical.

To put the the client's ability to the test, I flipped 1051 virtual coins using specialized decks over the course of just 11 games. Here is the spreedsheet with the tabulated data and screenshots: https://docs.google.com/spreadsheets/d/1Uz17FCRuWsqA4kWPcLoc9KGZDCEBkSRYYL9RWYVWVUY/edit?usp=sharing

The results, 516 heads to 535 tails, demonstrate that the PTCGO coin flip's randomness is indeed 50/50 with more than 95% confidence. Now we all have had those quintuple tails moments where we just cannot believe that the random number generator is truly treating each player fairly. However these situations are far too limited to generate a conclusion about the trustworthiness of the client. There is a 3.1% chance that 5 coins in a row would be tails, and given how much we play PTCGO, that isn't very outlandish to occur once in a blue moon. Beyond this, the change in probability of successive tails only diminishes, as the 1.6% (rounded) difference between 3.1% and 1.6% (6 flips) decreases to a 0.8% difference between 1.6% and .8% (7 flips). Another factor is that people tend to pay more attention to and recall more frequently the events the reinforce preconceived notions, such as that the coin flips are rigged. Despite flipping about evenly over the course of the game, if there was a short string of either heads or tails, one may notice the temporary wonkiness of the coin flips, while ignoring the many other occasions, and by doing so will not get the big picture of whether or not the flips even out over time.

Feel free to curse your luck and bemoan your misfortunes, but keep in mind that PTCGO is not actually "out to get you." If I get around to it, I might test out starting hands on PTCGO to see if they match up to expected frequencies, because I quite enjoyed this little experiment.

DIG video:

 

[MilesEX]

I refuse to believe this! Everytime I play Kyogre it's ALWAYS tails. Like I don't think I've ever had heads, soooo frustrating.

But yeah in all seriousness, it probably is 50/50.

 

[MarshalBry]

I don't 100% understand how computer RNG works more than it being based on the time of a computer (or something to do with time and timers), however at a point where you (of course) cannot rig it.

I could be totally wrong in this as this is just in my experience, but I've found very often the coin will be biased towards either heads or tails, however there's an equal chance of the bias being on either side (ie you can have a few games where you'll flip 90% tails/heads, and then it'll switch over or go "more even").

 

[Nytik]

I could be totally wrong in this as this is just in my experience, but I've found very often the coin will be biased towards either heads or tails, however there's an equal chance of the bias being on either side (ie you can have a few games where you'll flip 90% tails/heads, and then it'll switch over or go "more even").

This is just how randomness works, there's no bias in it. Humans do not really understand how a 50% probability really manifests, in the sense that our expectations are very different from reality. You can read more about it in this article of Mathematical Association of America: https://www.maa.org/sites/default/files/pdf/upload_library/22/Polya/07468342.di020742.02p0021g.pdf

Your hypothesis would be easy to test. One could count the number of runs of heads/tails that appear of a given length, and compare that to the expectation. I really doubt you will find anything suspicious, as the developers have no incentive to rig this and it would be much easier to just use a standard RNG.

Incidentally, people will always try to blame the software (or anything else) when they don't like the outcome. People even complain that unlikely results happen too often on online poker sites. Who would that even benefit? Money just gets passed around from player to player, there is no reason to try to rig that (and statistical analyses again show that it doesn't happen).

 

[WaMike]

Thank you, op, for collecting the data on this. I always chuckle when someone talks about the draw or flip engine being rigged in the game. Too often people see a run of flips that go against them as being rigged when, in reality, it's completely normal to see runs of several heads or tails in a row. 50/50 odds doesn't mean things simply alternate. To demonstrate this I built a simulator to demonstrate 1,000 Timer Ball flips. The results speak for themselves and are consistent with your findings. Seeing a run of 10+ heads/tails can actually happen fairly frequently if you play a lot. Sure, it feels rigged, but that's actually just the nature of random.

Also, I made a similar experiment to simulate starting hands already to see how often someone gets a "bad" start.

 

[Ecko9i6]

Thank you, op, for collecting the data on this. I always chuckle when someone talks about the draw or flip engine being rigged in the game. Too often people see a run of flips that go against them as being rigged when, in reality, it's completely normal to see runs of several heads or tails in a row. 50/50 odds doesn't mean things simply alternate. To demonstrate this I built a simulator to demonstrate 1,000 Timer Ball flips. The results speak for themselves and are consistent with your findings. Seeing a run of 10+ heads/tails can actually happen fairly frequently if you play a lot. Sure, it feels rigged, but that's actually just the nature of random.

Also, I made a similar experiment to simulate starting hands already to see how often someone gets a "bad" start.

Nice job on the analysis, the coin flip simulator really does speak for itself. I can understand the frustration of flipping into tails consecutively, but like you said its normal for something like this to happen.

 

[DragonFang101]

Nice job on the analysis, the coin flip simulator really does speak for itself. I can understand the frustration of flipping into tails consecutively, but like you said its normal for something like this to happen.

How can it be normal if it's supposed to be randomized? Bum bum BUUUUUM!