EPTRandom Mirror Match Up Percentages Off - Page 3
| Forum Index > SC2 General |
|
SeaSwift
Scotland4486 Posts
| ||
|
REM.ca
Canada354 Posts
On April 11 2011 01:20 DooMDash wrote: It could be broken for some people. I feel like there is actually a problem and I'd like to see if it actually is before people go dismissing it because they are not experiencing it. It would be an actual problem if it WASN'T for "some" people. In randomness, it's normal to have outliers. A few anecdotes from people with sample sizes of 50 won't let you determine for sure if there is a problem. Blizz checks this stuff out all the time with a nearly unlimited sample size. You have to trust them. EDIT: removed argument about map veto because I forgot vetoes don,t affect opponent choice. Don't you think it's more parsimonious to think that Blizzard can write a algorithm that can handle producing randomness out of 4 choices... | ||
|
SovietHammer
United States166 Posts
| ||
|
Zocat
Germany2229 Posts
On April 11 2011 01:26 Noocta wrote: Mirrors are a resultut of map veto sometimes. Each races veto somes maps the other races don't. That's assuming it is: "Look at available maps then decide an opponent" and not "Look at available opponents, then decide a map". | ||
|
NikonTC
United Kingdom418 Posts
*poll* | ||
|
legatus legionis
Netherlands559 Posts
 Binomial probability distribution: There is no conditional probility, the knowledge of what you have spawned as doesn't have any influence on the probability of your opponents race. We define the discrete random variable X as the number of mirror match ups. There are 9 matchups. I include zt and tz. So the probability that we get a mirror matchup is: 3 mirror matchups / 9 possible matchups = 1/3 or 0.333 repeating ofcourse. p = 0.333 Everytime you hit que it doesn't matter what your previous matchup was, the events are independent so we can use a binomial probability distribution to calculate the probability of getting "x" amount of mirror matchups in "n" amount of trials. We will also assume that the order in which you get the mirror matchups does not matter and can happen in any possible order. Your number of trials was 80. Your number of mirror matchups was 35. p(X=35) = (n! / r!(n-r)!) (number of possible ways to get 35 mirror matchups in 80 games) times p^x (the probability of getting a mirror matchup times itself for every occurence of a mirror matchup) times (1-p)^n-x (the compliment of p, the chance to not get a mirror matchup times itself for every occurence of a non mirror matchup) equals the probablity to get 35 mirror matchups in 80 games. I'm lazy so I'll just use this to calculate: http://www.stat.tamu.edu/~west/applets/binomialdemo.html Which gives us a probability of 0.0121 or 1.2%. I'll leave this here, hope I don't make a fool out of myself, feel free to mess around with the calculator and decide what the odds mean. Because for me, I wouldn't raise an eyebrow if I had the same matchup 1000 times in a row. The human mind is very deceptive to thinking they are spotting a pattern where in reality, there is no such pattern. | ||
|
Kalai
Japan17 Posts
Nope, I don't have statistics, but they wouldn't prove anything anyway, since I haven't played 2000 games. And seriously, the percentage of mirror match ups is irrelevant to the thread unless you play random. | ||
|
Varpulis
United States2517 Posts
On April 11 2011 01:43 NikonTC wrote: flipped a coin four times and got heads three times. that's 75% in favour of heads :/ shouldn't it be around 50% heads and 50% tails? I think there's something wrong with universal physics. has anyone else been experiencing this? *poll* This post hits the nail on the head. Sample sizes are really small here. When you've played 1000 games as random, come back and share your findings. | ||
|
NinjaDrone
United States97 Posts
On April 11 2011 01:43 NikonTC wrote: flipped a coin four times and got heads three times. that's 75% in favour of heads :/ shouldn't it be around 50% heads and 50% tails? I think there's something wrong with universal physics. has anyone else been experiencing this? *poll* Many of you completely missed the point of the post. I know my sample size is too small, that's why I asked for statistical information from other random players to increase my sample size. Seriously, you all need to read before you make obvious comments. All I want is how many games random players have had as mirrors and how many games they've had as non-mirrors. | ||
|
Gnax
Sweden490 Posts
Lets say 91% of the player base plays Protoss, 3% plays Zerg, 3% plays Terran and 3% plays Random. Is there still 33% chance that you get a mirror match up when you play as random? | ||
|
Smurphy
United States374 Posts
| ||
|
Doso
Germany769 Posts
| ||
|
Noocta
France12578 Posts
On April 11 2011 01:41 Zocat wrote: That's assuming it is: "Look at available maps then decide an opponent" and not "Look at available opponents, then decide a map". Hum, i don't know about this to be honest. Do someone know or it's blizzard privacy ? | ||
|
Marradron
Netherlands1586 Posts
On April 11 2011 02:03 Gnax wrote: Hmm I am curious about one thing. Maybe some of the math geniuses here can answer. Lets say 91% of the player base plays Protoss, 3% plays Zerg, 3% plays Terran and 3% plays Random. Is there still 33% chance that you get a mirror match up when you play as random? Ofcourse not. Theres more protoss players queueing. Thuss the chance of meeting protoss will be 91 procent. | ||
|
canikizu
4860 Posts
I got like 5 mirror match out of ~60 games or something. | ||
|
Sentient
United States437 Posts
On April 11 2011 01:41 Zocat wrote: That's assuming it is: "Look at available maps then decide an opponent" and not "Look at available opponents, then decide a map". | ||
|
Hopeless1der
United States5836 Posts
- Rolled terran once or twice. - Rolled 3-4 PvZ or ZvP - Rolled ~10 PvP - Rolled ~10 ZvZ - Rolled ZERO TvT Make of that what you will, but goddamn it seems like blizzard is trolling me. | ||
|
shinarit
Hungary900 Posts
| ||
|
KaiserJohan
Sweden1808 Posts
I thought with HT nerf people would leave protoss instead. I like it though, less mirrors the better. | ||
|
REM.ca
Canada354 Posts
On April 11 2011 02:06 Marradron wrote: Ofcourse not. Theres more protoss players queueing. Thuss the chance of meeting protoss will be 91 procent. Actually, it WILL still be 33%. You'll get alot more PvP than the other mirrors but if you add the probability of each mirror together, it will give 33%. (.33*.92)+(.33*.04)+(.33*.04) = .33(1)=.33 | ||
| ||
