Dice Analogy to ELO Hell

Comment below rating threshold, click here to show it.

CyraLoL

This user has referred a friend to League of Legends, click for more information

Senior Member

06-05-2011

In the most simple terms of what you fail to understand is that we are not talking about compounded odds.

What you fail to realize is that the odds of any permutation of a sequence is incredible low - while there are many instances of 50/50, any specific one has the exact same probability as a specific instance of 99/1 (which also has many permutations, obviously).

Because its the job of forum trolls to constantly bicker, I'm not going into this too deeply as it would be more or less wasted effort as I'm sure they could find other, even less relevant, wikipedia articles that they've never read and don't understand to link. (PS: Thats not how you model the probability of flipping a coin).

The take home point is that probability doesn't exactly work like the original post assumes it does. It both could (and likely would) take a staggering amount of games to play out with any considerably accuracy, and there is a clearly and definably chance that for a decent portion of players it would never, ever, actually play out even in a thousand games.

While the odds of getting to "your" ELO do increase, that does necessarily mean that you begin to approach it at any given point.

Since I'm not really here to argue the finer points of probability or logic with internet trolls (since they are not going to change their mind short of catastrophic divine intervention to their face in my experience) just skip over their quibbling and move on to this:

The sheer number of additional random fluctuations in a probability simulator as complicated as League of Legends makes an absolute mockery of weighted probability as presented in the original post. The completely staggering amount of viable factors in the equation that decrease the personal-player-weight of the equation make its contribution borderline insignificant in the short run.

For example if even one player receives a telephone call, has a power outage, suffers from massive lag, is having a bad day, decides to troll his teammates due to a disagreement, etc, this is going to likely have more weight than a minor differential in skill of one of the players.

Further, as discussed in my first post, the system itself is already counter-weighted to 50/50 by purposefully assigning a favored and disadvantaged team at (as far as I know) complete random.

TL: DR While the trolls were incorrect anyway, the point their picking on is actually irrelevant to why probability is not especially likely to bring any particularly player to even a good approximation of their ELO. The average player with be within an average interval of their ELO. That is a far cry from the perfect system as reference in the opening post.


Comment below rating threshold, click here to show it.

SallySimple

Senior Member

06-05-2011

Quote:
Originally Posted by DuffTime View Post
Truth.

And the people who complain about ELO hell could be better psending their time playing and improving their skills so that their ELO would rise steadily =p
The second part of this post is so, so true.


Comment below rating threshold, click here to show it.

DuffTime

Senior Member

06-06-2011

Quote:
Originally Posted by sn3aks View Post
yup the trolls who insits match making works are just trolls
Matchmaking is fine, but what she said was true none the less.


Comment below rating threshold, click here to show it.

DuffTime

Senior Member

06-06-2011

Quote:
Originally Posted by MTaur View Post
Just kidding. I haven't started ranked games yet. May hold off a little if they're planning a reset soon, idk. But yes, I know a little background theory in stats, and laws of large numbers ain't be lyin'.
I would get started on Ranked now so that when the reset comes you won't have lost anything if it doesn't work out 8D


Comment below rating threshold, click here to show it.

DuffTime

Senior Member

06-06-2011

Quote:
Originally Posted by CyraLoL View Post
In the most simple terms of what you fail to understand is that we are not talking about compounded odds.

What you fail to realize is that the odds of any permutation of a sequence is incredible low - while there are many instances of 50/50, any specific one has the exact same probability as a specific instance of 99/1 (which also has many permutations, obviously).

Because its the job of forum trolls to constantly bicker, I'm not going into this too deeply as it would be more or less wasted effort as I'm sure they could find other, even less relevant, wikipedia articles that they've never read and don't understand to link. (PS: Thats not how you model the probability of flipping a coin).

The take home point is that probability doesn't exactly work like the original post assumes it does. It both could (and likely would) take a staggering amount of games to play out with any considerably accuracy, and there is a clearly and definably chance that for a decent portion of players it would never, ever, actually play out even in a thousand games.

While the odds of getting to "your" ELO do increase, that does necessarily mean that you begin to approach it at any given point.

Since I'm not really here to argue the finer points of probability or logic with internet trolls (since they are not going to change their mind short of catastrophic divine intervention to their face in my experience) just skip over their quibbling and move on to this:

The sheer number of additional random fluctuations in a probability simulator as complicated as League of Legends makes an absolute mockery of weighted probability as presented in the original post. The completely staggering amount of viable factors in the equation that decrease the personal-player-weight of the equation make its contribution borderline insignificant in the short run.

For example if even one player receives a telephone call, has a power outage, suffers from massive lag, is having a bad day, decides to troll his teammates due to a disagreement, etc, this is going to likely have more weight than a minor differential in skill of one of the players.

Further, as discussed in my first post, the system itself is already counter-weighted to 50/50 by purposefully assigning a favored and disadvantaged team at (as far as I know) complete random.

TL: DR While the trolls were incorrect anyway, the point their picking on is actually irrelevant to why probability is not especially likely to bring any particularly player to even a good approximation of their ELO. The average player with be within an average interval of their ELO. That is a far cry from the perfect system as reference in the opening post.
I love reading your posts xD

Well informed, and entertaining to boot =D


Comment below rating threshold, click here to show it.

psymunn

Senior Member

06-06-2011

not to bog down on symmantics but:

Quote:
Originally Posted by CyraLoL View Post
What you fail to realize is that the odds of any permutation of a sequence is incredible low - while there are many instances of 50/50, any specific one has the exact same probability as a specific instance of 99/1 (which also has many permutations, obviously).
nothing about his post indicated that he failed to realise this. you said the odds of 50/50 coin split are the same as the odds of a 99/1 coin split. this isn't true. if i said 'the odds of rolling 7 on two dice are the same as rolling 3' you'd tell me i was wrong. yes, the odds of rolling a 4 and a 3 on a dice are the same as the odds of rolling a 1 and a 2, but, again, this isn't what your original post was discussing. it doesn't make the rest of your posts any less valid.

while ELO 'hell' can create a lot of noise when working out your rating, a 1600 ELO player will have a trend toward 1600 ELO. the occasional dropped connection does happen, but even then 4v5 wins are possible. a good player who picks champions that help team composition can, and does advance in ELO. and this can happen a lot quicker than one might assume because a player isn't a dice and their preformance isn't actually random.

i'd also argue that the effects of a good player become more noticeable at higher levels, rather than less noticeable. a single feeder can overshadow the efforts of even the best player. in a game like LoL where being bad is worse than being absent, there certainly will be more 'noise' at lower levels


Comment below rating threshold, click here to show it.

DuffTime

Senior Member

06-06-2011

Quote:
Originally Posted by psymunn View Post
not to bog down on symmantics but:


nothing about his post indicated that he failed to realise this. you said the odds of 50/50 coin split are the same as the odds of a 99/1 coin split. this isn't true. if i said 'the odds of rolling 7 on two dice are the same as rolling 3' you'd tell me i was wrong. yes, the odds of rolling a 4 and a 3 on a dice are the same as the odds of rolling a 1 and a 2, but, again, this isn't what your original post was discussing. it doesn't make the rest of your posts any less valid.

while ELO 'hell' can create a lot of noise when working out your rating, a 1600 ELO player will have a trend toward 1600 ELO. the occasional dropped connection does happen, but even then 4v5 wins are possible. a good player who picks champions that help team composition can, and does advance in ELO. and this can happen a lot quicker than one might assume because a player isn't a dice and their preformance isn't actually random.

i'd also argue that the effects of a good player become more noticeable at higher levels, rather than less noticeable. a single feeder can overshadow the efforts of even the best player. in a game like LoL where being bad is worse than being absent, there certainly will be more 'noise' at lower levels
I agree and I disagree. =P

"
nothing about his post indicated that he failed to realise this. you said the odds of 50/50 coin split are the same as the odds of a 99/1 coin split. this isn't true. if i said 'the odds of rolling 7 on two dice are the same as rolling 3' you'd tell me i was wrong. yes, the odds of rolling a 4 and a 3 on a dice are the same as the odds of rolling a 1 and a 2, but, again, this isn't what your original post was discussing. it doesn't make the rest of your posts any less valid.

while ELO 'hell' can create a lot of noise when working out your rating, a 1600 ELO player will have a trend toward 1600 ELO. the occasional dropped connection does happen, but even then 4v5 wins are possible. a good player who picks champions that help team composition can, and does advance in ELO. and this can happen a lot quicker than one might assume because a player isn't a dice and their preformance isn't actually random."

True.

"i'd also argue that the effects of a good player become more noticeable at higher levels, rather than less noticeable. a single feeder can overshadow the efforts of even the best player. in a game like LoL where being bad is worse than being absent, there certainly will be more 'noise' at lower levels"

Sometimes true, sometimes not true. The best of the best players can single-handedly roll a team with a good carry. It's the reason they always end up high elo.

They win the 4v5's and they can win them even with the bads and the feeders. That doesn't mean it's easy, but they might pull some real skilled moves, get the clutch ace when the enemy is pushing, and push down the base for the win. It won't happen every game, but it's the frequency at which it does happen that makes the difference.


Comment below rating threshold, click here to show it.

CyraLoL

This user has referred a friend to League of Legends, click for more information

Senior Member

06-06-2011

Something to keep in mind is that I'm not nay-saying the system - match making couldn't be made much better than it is easily.

ELO Hell is merely the level of murk from which incredible bad players and trolls permeate and create an excessively random outcome in terms of game victory.

It is not at all impossible to get out of this level of murk - it is merely that your level of skill while in this murk is going to be reflected in your match outcome to a considerably less degree, and according to probability, it is fully possible that you will not see your own skill play you out of the murk-level for hundreds of games - or ever (though this case becomes unlikely under thousands of games, one cannot reasonable expect a normal person to play that much, especially in ELO Hell conditions - and even than there is a significant possibility of getting boned.

A lot of people point to the vastly superior matchmaking system of WoW Arenas, the at the end of the day, that system would have failed epically just the same if there had been a solo queue option for 5v5 Arenas (in fact, that was vaguely painful to type as I imagined it). Point is, solo queue matchmaking at entry level is Hell.

At the end of the day, you can say that a 1600 player trends towards 1600, and this is theoratically true. It is not, however - in many cases - practically true. Especially in the short term. It is pretty simple probabilty that could tell us that with the volume of the noise of ELO Hell (Hell to which I'm going for that pun) that a 1600 player may find themselves in ELO Hell and holding. In fact, far more 1600 players would likely be stuck at 1100 than would have risen to 1800 (which, under standard probabilty circumstances, would be more likely).

PS: As far as probabilty goes, I'm done feeding trolls. If you don't like my version, go to college and study it. Come back in four years. We'll discuss it over tea than. Emphasis on the four years part. Ie: Long after I've forgotten this exists.


Comment below rating threshold, click here to show it.

BlindPhaydo

Member

06-06-2011

Quote:
Originally Posted by LoLasagna View Post
ಠ_ಠ

http://en.wikipedia.org/wiki/Binomial_coefficient

Chance to get 50/50 is like 10^27 times more likely than 99/1.
Haha, glad someone pointed this out. Sad that a lot of people intuitively believe CyraLoL's way of thinking is correct.


Comment below rating threshold, click here to show it.

ReLaTiviT

This user has referred a friend to League of Legends, click for more information

Senior Member

06-06-2011

Quote:
Originally Posted by CyraLoL View Post
This is sorta true, but makes assumptions about probability that are just not necessarily true.

The common example of why this does not actually work the way you think it does is that say for example you flip a coin one hundred times. The chance of getting 99 tails and 1 heads is actually the exact same as getting exactly 50 heads and 50 tails. While the system you mention is a weighted average, thus more likely to trend, it still far far far from statistically impossible for someone to move vastly the wrong direction in a hundred games.
Wrong, if you were to flip a coin 100 times you have a much higher chance of rolling 50 heads/tails than 99 tails 1 head.

50/50:
The first 50 flips DO NOT matter, even if you got 50 heads/tails you could balance it out.

99/1:
The first 1 flip DOES NOT matter, even if you got 1 heads/tails you could balance it out.

So the difference is 49 freeloading flips, 0.5^49.