in order to gain ranked rewards at the end of the season you would have to play at least 100 ranked games.
also* for your first 100 games players would be highly prone to moving up and down the ladder freely, im talking whole divisions at a time.
what exists, vs whats actually true and an actual measure of skill. my method would bring it about. http://i.imgur.com/dEoSdpZ.png
Sorryififail4 votesWhy? Literally no reason for that
Very Hard Engage1 votesWhy? Literally no reason for that
0 reason for promotional matches and having to win above 75% of games, but ya know, riot must have its log scale. and not bell curve.
put it this way, with 100 games of moving up and down people would quickly fall in/out or abolish the whole "elo hell" climb. no more "getting stuck with trash team mates" because the people around you are supposedly competent players. and if you lose it should be your wrong doings, not your teams.
if say you end up right where you started in plat after 100 games with a 50% win/loss, great, you're plat, you think you can climb from say bronze to plat in 100games? NOPE, promotional matches!
or how about this, you lose more than you win, at say 30-70, well you'll quickly find yourself in bronze after your first 15games. and then you can proceed to try and win your next ~80 and see what happens.
or you win a lot? enjoy diamond and winning against diamond players, no more of that nasty +8LP - 25LP nonsense.
LadyRenly4 votesthen all the people who deserve to be plat wouldn't because of all the terrible people that belong in bronze would drag them down(such as myself, im terrible)
terrible idea
Very Hard Engage1 votesthen all the people who deserve to be plat wouldn't because of all the terrible people that belong in bronze would drag them down(such as myself, im terrible)
terrible idea
except they wouldn't even be listed because they haven't even played their 100 games, it would be strictly for placement. NOT SEASON, BONUSES, no border, no rewards, nothing, they aren't even placed.
now as for "everyone" literaly everyone, playing their 100 games at the same exact time, and odd non plat players being in the group, remember, players like you would quickly find themselves playing with gold, then silver, then bronzies after just a handful of games, finding it difficult to make it back into plat groupings without a very high win streak.
and as for potentially lower ranked players being placed higher than where they belong, remember, you're still focused on the log curve of what riot wants you to believe is skill, vs what skill actually is, a bell curve with high amounts in the middle and slight curves upwards again at either end. you would see many more plat-gold players rather than bronze players, this would actually normalize the ranking curve, instead of forced scaling upwards due exclusively to the current win/loss climb that exists.
Kinifesis1 votesMy first ranked season I didn't play 100 ranked games, cap should be lower I think.
Very Hard Engage1 voteswhich would stand to reason where ever you were placed wasn't quite accurate, and there was probably still room for error at the time.
and you think thats fair for you?
Sir Yamazuki1 voteswhich would stand to reason where ever you were placed wasn't quite accurate, and there was probably still room for error at the time.
and you think thats fair for you?
Why not? Some people just want to maintain their rank.
Very Hard Engage1 votesim sorry, silly me for thinking people here could do math, let me explain in detail what you aren't getting at, currently, people aren't rated by their skill level, instead its solely by number of games played and how determined you are to reach the top. what im suggesting gives players a chance to achieve their actual skill rank quickly and easily. my apologies, sorry that rustled the current diamond+ players jimmies.
Z Statistic1 votesWhat is wrong with an exponential distribution as a basis for skill tiers? That's not reality (reality is a highly skewed distribution that resembles something of an F-Distribution), but if it were, so what? The only thing that's magical about a normal distribution is how it's the result of the CLT. that's it. Everything else? It's just another distribution. In fact, solo queue's distribution in Season 3 had a significant contaminating distribution that was about as normal with similar variance but a larger mean.
But of course there is a strong correlation between time invested and how high you climb. Do you think Diamond players just wake up one day and make it to diamond in twenty games without any goal of self-improvement? Obviously SOME people are anomalies, but that's not how most people work. Diamonds have to work hard and invest a lot of time into the game in order to become good at the game. That's like saying the only thing that dictates whether you get into the Olympics is how much time you spend playing a sport. Well, yeah, if you're at that level, you're almost certainly going to have thousands, if not tens of thousands, of practice time. But that in and of itself does not make you Olympics material or Diamond material.
What you're saying is that the Elo parameter K should have an insanely larger base value because you think that higher initial variance leads to a better outcome.. You're also saying that the initial condition of Elo should be plat. Here is what would happen:
You can mathematically prove that increasing K does not change the limit of the E(Elo[a,b]) when a=b (where a and b are the normal and real elo of a player). Increasing K only helps if a!=b and only by reducing the number of iterations required to be within epsilon of the aforementioned limit L (which in real terms is what the system would call "confidence".) However, you can also reduce the number iterations required by simply having the base elo be reasonable close to the player's real elo. You do this by setting the base elo equal to the median of the population's distribution.
If we were to return to pure elo like in season 2, then what you would be doing is simply giving ranked rewards to everyone because Plat would be the new bronze. If we were to keep the LP system, then all you would be doing is requiring someone play 10x more placement matches.
If you want placements to require more matches than 10, then just say so. Do not dress this up to be more than what is.
Very Hard Engage1 votes[{quoted}](name=Menea Calais,realm=NA,application-id=3ErqAdtq,discussion-id=bOAvWXhX,comment-id=00030000,timestamp=2015-11-10T02:39:23.617+0000)
What is wrong with an exponential distribution as a basis for skill tiers? That's not reality (reality is a highly skewed distribution that resembles something of an F-Distribution), but if it were, so what? The only thing that's magical about a normal distribution is how it's the result of the CLT. that's it. Everything else? It's just another distribution. In fact, solo queue's distribution in Season 3 had a significant contaminating distribution that was about as normal with similar variance but a larger mean.
But of course there is a strong correlation between time invested and how high you climb. Do you think Diamond players just wake up one day and make it to diamond in twenty games without any goal of self-improvement? Obviously SOME people are anomalies, but that's not how most people work. Diamonds have to work hard and invest a lot of time into the game in order to become good at the game. That's like saying the only thing that dictates whether you get into the Olympics is how much time you spend playing a sport. Well, yeah, if you're at that level, you're almost certainly going to have thousands, if not tens of thousands, of practice time. But that in and of itself does not make you Olympics material or Diamond material.
What you're saying is that the Elo parameter K should have an insanely larger base value because you think that higher initial variance leads to a better outcome.. You're also saying that the initial condition of Elo should be plat. Here is what would happen:
You can mathematically prove that increasing K does not change the limit of the E(Elo[a,b]) when a=b (where a and b are the normal and real elo of a player). Increasing K only helps if a!=b and only by reducing the number of iterations required to be within epsilon of the aforementioned limit L (which in real terms is what the system would call "confidence".) However, you can also reduce the number iterations required by simply having the base elo be reasonable close to the player's real elo. You do this by setting the base elo equal to the median of the population's distribution.
If we were to return to pure elo like in season 2, then what you would be doing is simply giving ranked rewards to everyone because Plat would be the new bronze. If we were to keep the LP system, then all you would be doing is requiring someone play 10x more placement matches.
If you want placements to require more matches than 10, then just say so. Do not dress this up to be more than what is.
you're still working on the assumption that wins/losses are equally weighted in both directions, whereas with what im proposing would give players the chance to reach their normal median more easily, reach where they belong faster and potentially climb higher with a shorter number of total games played.
its not about being exact, its about whats fair, the current system can only produce a scaling log system, where for every person who climbs up, someone else must lose that rank. yes i admit, more people would be placed higher, but guess what, they deserve it. they shouldn't sit in bronze/silver because their slight skill increase isn't enough to carry them through, but instead they feel degraded by being forced to play with people who dont match their skill level most of the time, and literally have nowhere else to go because it is the bottom.
im sorry you want your precious EPEEN, but guess what, people misconstrue rank for skill, and thats outright wrong, i proposed a fair way for skill to be measured, it would work, but because you people are soooo brainwashed by riot into believing you're the best player in league because you've obtained diamond, when in reality theres tons of others with your skill level.
stop putting your faith into a bad system, wins/loss =/= skill.
rank should be based on skill, not matches played.
you want rank, you want to say you're better than someone else, well the current LP system doesn't grant you that, only the extreme bottom half sees rank as skill.
all the challenger players even say it, "once you get to high plat everyone is pretty much the same" "maybe some very slight differences, but mechanically everyone is all there" qtie, bjerg, nightblue, dyrus, i can go on.
but guess what, all those plat players, are in plat+ not because they're better, but because they won more games than someone else who is lower than they are.
dont you dare say this system is good and fair for everyone. its only fair to those who play 1000+ games every season. if the system was more liberal in movement, then i could start to see and understand it, but its just so unforgiving to everyone and so biased towards the people who play more.
what will it take for you to understand the simple fact that some silver player could be just as good as a diamond player, but just doesn't play the game. and because that silver player plays so infrequently he will never enjoy diamond rewards? im not saying myself, but someone could be.
GigglingAngel1 votes100 games? If every game were to end at 20 minutes, that would still be 2000 minutes. That would mean that, IF every game was able to be ended by exactly the surrender time, that would still require over 33 hours of play.
Seeing as most games are 30-40 minutes, you can realistically multiply that amount by around 1.5-2x.
No thanks. Not interested in staying in a cesspool of toxicity for that long.
