5% of players deserved punishments in 2014? BULLSHIT

Jamohs Zes·Nov 14, 2014·7 votes·1,360 views

3 ranked matches in a row 2 trolls on my team...5%???? Playing the game says otherwise.

19 Comments

DregenNov 14, 20146 votes

5% is 1 in 20 players. Each game has 10 players. 1 of them is you. That means after 2 games, you've played with 18 different players. The chance of meeting a player who deserves punishment in 2 games is thus roughly 60%. 3 games = 75%. 4 games = 84%.

UNDENIABLE LOGIC

The Storm DragonNov 14, 20143 votes

I don't think you understand how statistics work, bro.

DregenNov 14, 20142 votes

I don't think you understand how statistics work, bro.

There's a reason you're only in gold 5.

disregardableNov 14, 20143 votes

Troll response is troll. That's not really how statistics works. Regardless, it's a fair point conceptually. I just usually find someone to report every game, not every other game. That's B5 though.

The Storm DragonNov 14, 20141 votes

I don't think you understand how statistics work, bro.

There's a reason you're only in gold 5.

The reason I'm in gold 5 is because there's no reason to play ranked after reaching it, since I have no intention to go pro and plat/diamond rewards are cosmetic (I mean, I guess a champion skin is also cosmetic, but it's more substantial than an icon or a border). But also, I was talking to the OP anyway.

TrylobyteNov 14, 20142 votes

There was originally a longer post here, but then I learned I was measuring from the wrong variable (though my math after that was correct) and I'm going to let the statistics major handle things. Suffice to say, the odds of having a troll in your game are pretty good, but not as high as people like to think.

ValyrianBladeNov 14, 20142 votes

Even your initial math is wrong.

The expected value of the number of trolls in a match is 0.5 (10 × 5%). That doesn't mean there's a 50% chance of a match having a troll, not does it mean that you will see a troll every other match.

You say that the equations are correct and that's how math works, but both statements are wrong. Btw this is coming from a statistics major.

If 5% of players are "trolls" then you assume the number of trolls in a match of 10 players is a binomial distribution (where n = 10 players and p = 5% chance of each player trolling). Of course this isn't perfectly accurate since knowing the first player is/isn't a troll actually changes the probability of other players in the same match being a troll due to mmr matching, but you and everyone else in this thread have assumed independence here so I will too.

Thus, the probability of seeing k trolls in a match with n=10 players and p=5% of one player trolling is (10 choose k)(0.05)^k(1-0.05)^(10-k)

Substituting k=0 for the probability of 0 trolls yields: (1)(0.05)^(0)(0.95)^(10)=0.598736939... Or roughly 60% as the original post here indicated and everyone seems to think is wrong.

The reason it's not simply 105% is because it's possible to see more than 1 troll in a match. K=1: 10(.05)^(1)(.95)^(9) = 0.31512470486... = 32% K=2: 45(.05)^2*(.95)^8 = 0.0746347985... = 7% K=3: 120*(.05)^3*(.95)^7 = 0.010475059... = 1% K=4: 210*(0.05)^4*(.95)^6 = 0.0009648... = 0% I could go on but it should be clear that k=5, 6, 7, 8, 9 will continue to get closer to 0%. K=10 = 1*(.05)^10*(.95)^0 = 0.00000000000009765625 if anyone is interested in the chance of all 10 players being trolls.

In case anyone wants to know about the choose function at the start, it's basically a fast way of counting how many ways you can order something. There is only 1 order that makes all 10 people not trolls (the first is not, second is not, third is not, etc...). There are 10 ways 1 person of 10 is a troll (the first is and no one else, the second and no one else, etc...). There are 45 ways 2 can be trolls (1 and 2, 1 and 3, 1 and 4, ... , 1 and 10, 2 and 3, 2 and 4, ... , 2 and 10, 3 and 4, 3 and 5, ... , 3 and 10, 4 and 5, ... You get the picture). While these solutions can be counted fairly easily, There are 252 ways that 5 are trolls, and it's not realistic to list them out. The choose function tells you how to quickly calculate it as: 10!/(5! * (10-5)!) = 10!/(5!)^2 = 10987...21/(54321)^2=3628800/(120)^2=3628800/14400=252. Where ! Represents the factorial function and means to multiply the number by each whole number less than it down to 1, and we'll say 0! = 1 is a convention (at high school math level you'll probably be told this but in university you'd be taught why 0! = 1).

Anyways your math lesson for the day.

Edit: and wow it seems these forums treat an asterisk, the common symbol for multiplication because times looks too much like an x, as code for italics so all of my formulas look like garbage. Whatever, I'm too lazy to fix that.

TrylobyteNov 14, 20141 votes

Actual Math

Eh, I haven't had statistics in a long time. I also explicitly state that my calculations are quick and fast, and that I assume that the player could very well be the troll (because they certainly can be, though that would make their chances of having a troll in the match 100% wouldn't it?). I did figure out the correct way to do the math later on (the way you did it, by player) but by that point I figured nobody cared any more and I didn't bother to correct it. I've done so now.

Sailor MintNov 14, 20141 votes

The sample pools are also different.

While it's 5% overall. It could easily be 10-15% of ranked players and <2% in ARAM and coop vs AI.

There are less people playing ranked but they are known to be much more toxic.

TrylobyteNov 14, 20141 votes

The sample pools are also different.

While it's 5% overall. It could easily be 10-15% of ranked players and <2% in ARAM and coop vs AI.

There are less people playing ranked but they are known to be much more toxic.

This has more of a factor on it than any mathematical errors, I think. Co-op has very, very few toxic players (there's no 'reward' for being toxic in bots) while Ranked seems to almost encourage that sort of behavior (since being toxic has definite 'rewards').

DregenNov 14, 20142 votes

Well you made the math sound really complicated, but admittedly since you have the degree and I don't you get the right to flaunt it. But the basic assumption is that in a pool of all eligible ranked players at a certain mmr, 5% of them are trolls. Assuming they are randomly picked from that pool, each player picked has a 95% chance of not being a troll. The chance of that happening 10 times in a row is .95^10 = .5987. That means there is a 59.87% chance a match will contain 0 trolls. It;s thus easy to calculate the chance a match will have at least 1 troll, by subtracting 59.87% from 100%. That gives us 40.13% chance a match will contain at least 1 troll (there may be more than one troll, but that is the chance there will be at least 1.)

Note that when i say troll, I mean "a player that deserves punishment". This means the troll is either an intentional feeder, a constant afk'er, verbal abuser, or anything else that might be reportable and punishable. It doesn't mean that person will "troll" a particular game he's in, which is why it feels lower than 40.13% in reality. If we assume the troll only displays reportable behavior in half of his games, and that we do not notice trolls that are on the enemy team, then it starts to feel like a more reasonable number when you take into account actual ranked experiences. It's not unheard of for a troll player on your team to actually go 12/2/7 and carry the game, and you'd never know he was a troll.

Looking at that number is honestly depressing, and makes me not want to play ranked anymore :p.

ValyrianBladeNov 14, 20142 votes

Your math was entirely fine, correct, accurate, and to the point. Someone else claimed you were wrong and tried to correct you (they've since edited their post) which is why I went math-HAM on them :p

You're also right that they don't behave like that every game :p. So it might not be that bad.

OhBoyItsaMegamanNov 14, 20143 votes

Toxic players have a higher tendency to foster toxicity in their games. They also have a lower threshold and tend to see reportable behavior where there is none.

There is also the possibility that some of the most toxic players are also the ones who play the game more, so you're seeing them more often than the friendly casuals even though there really aren't that many of them.

KellenGraceNov 14, 20141 votes

It's because you're in Bronze. I feel like even at Silver, the bullshit just became infinitely less of a thing, and in Gold, I hardly get it at all. Whenever I play with my Bronze friends, though, it's there all over again, and it's awful. Try to improve your game and move up, and you'll be playing with people who aren't total shitheads.

AloneInTheCrowdNov 14, 20141 votes

Rigged item 3070

JJI744Nov 14, 20141 votes

Actual (hopefully) correct math.

5% chance, 10 people per match.

Odds each person NOT being toxic=1-5%=95%.

Odds of everyone not being toxic: 95%^10=~60%.

Odds of at least one person being toxic: 1-60%=40%.

Odds of having at least one toxic player two matches in a row: .40^2=16%. Quite possible.

Note: Toxic in this case means a player behaving poorly enough to receive some form of punishment from Riot.

Personal note: I feel like I get a lot less toxic players than this indicates. That MAY indicate that there are certain ELOs or game modes that have higher/lower toxicity levels. EDIT: I play almost exclusively ranked solo-queue, around high Gold-low Plat ranking.

AetherKingdomNov 14, 20141 votes

Maybe you haven't separated doing bad from trolling

SalidfingersNov 14, 20141 votes

Dated 2014, Jan.

In October 2012, Riot Games revealed that League of Legends had 12 million daily active players , meaning at least 12 million people a day logged in and played the game. With 70 million accounts, it was easily one of the most popular games in the world. Today, Riot Games announced that it has more than doubled the 12 million figure, revealing it has reached 27 million daily active users, the Wall Street Journal reports.

Riot also said that it broke its record of peak concurrent players, up to 7.5 million from the 5 million it reported last March. In terms of monthly active users, it went up to 70 million from the 32 million it reported in 2012. If that didn't drop your jaw consider that a recent report from SuperData estimates that LoL, a free-to-play game, brought in $624 million in 2013. Though LoL is only heading into its fourth season, last week , Riot Games' Vice President of eSports Dustin Beck said that he expects it to match traditional sports in its longevity. "It's a sport like just like anything else,” he said. “Baseball has been around 110 years. I'm not saying League of Legends will be around that long, but we see the shelf life of LoL as hopefully decades."


5% of  27 million daily active users = 1,350,000 punishments 

I don't know if my information is accurate, but its something

Jamohs ZesNov 15, 20141 votes

First ranked match of the day and no joke all 4 people on my team should be punished.

jungle: never connected.

top: bitched about ping and talked shit the entire 12ish minutes.

mid: leaver after 2-3 minutes.

sup: has so little idea of how to play the game and is so negative that they don't know how to use the spoils of war stacks from the relic shield and then bitches at team when they bring it up. Then spends the rest of the game bitching at everyone else to "play" while she just stands there doing nothing at all.

5%???????