|
Please keep the QQ to a minimum if you do not like this update. We are happy to hear your reasoning for not liking a ranked system, but no "OMG VOLVO WHY" posts. |
Canada2068 Posts
On January 14 2014 05:48 Sn0_Man wrote:Show nested quote +On January 14 2014 05:37 ILoveCoffee wrote:
Does anybody have any idea what percentiles the MMRs correspond to? I find MMR to be much more meaningful if you have something to refer it to. I recommend reading the OP, its traditionally a good thing to do.
valve says it's likely ranked mmr distribution will be different from normal mmr distribution that's given in the OP. the problem is that there's no way to collect ranked mmrs from every player using some automated method, so ppl are forced to use surveys to collect mmrs, which has problems of dishonesty and being a biased sample (low mmr players might not want to submit their mmrs, so whole distribution might be skewed). here's the one from reddit: https://docs.google.com/forms/d/1rvZthoRONEFRLNEHm56-vf6TQRgldpZsWiFhmURWDtk/viewanalytics it's probably the best one we'll have unless valve releases a distribution.
|
On January 14 2014 05:11 Merany wrote:
So, I have a pretty bad solo ranking (~2500) but I'm really trying to improve and I feel like I'm getting fucked by team comp quite a lot. Like for the past 10 games or so, we are in the lobby, I select a hero (not confirmed yet) and position myself on the map where I'd like to be. I wait a bit to see if other people do the same, maybe try to talk about what we would need and what people want to play. And usually get zero answers. And then, when the timeout expires, it seems like everybody goes back from afk and select their favorites carries without even looking what the team comp is. If I've already confirmed a carry, congratulation, we now have a 5 carries team and we get stomped. If not, that's still a 4 carries 1 support team and the game is hard. And no kidding, this happens 4 games out of 5.
First question is: is there anything I can do against that? I'd really like to up my ranking a bit because I'm sure people don't play like that at higher levels.
And secondly, is there a hero I can pick quickly that's pretty descent no matter what my allies chose and that I will be able to make the most off ? (I've already seen the Lycan thread and it's good but I'd like other suggestions).
Heroes I usually play: Luna, weaver, prophet, enchantress, CM, shadow shaman.
Try not playing all pick. I find I get better team compositions in nearly every other game mode.
|
On January 14 2014 06:07 CountChocula wrote:Show nested quote +On January 14 2014 05:48 Sn0_Man wrote:On January 14 2014 05:37 ILoveCoffee wrote:
Does anybody have any idea what percentiles the MMRs correspond to? I find MMR to be much more meaningful if you have something to refer it to. I recommend reading the OP, its traditionally a good thing to do. valve says it's likely ranked mmr distribution will be different from normal mmr distribution that's given in the OP. the problem is that there's no way to collect ranked mmrs from every player using some automated method, so ppl are forced to use surveys to collect mmrs, which has problems of dishonesty and being a biased sample (low mmr players might not want to submit their mmrs, so whole distribution might be skewed). here's the one from reddit: https://docs.google.com/forms/d/1rvZthoRONEFRLNEHm56-vf6TQRgldpZsWiFhmURWDtk/viewanalytics it's probably the best one we'll have unless valve releases a distribution.
Which is like saying "how good is the average soccer player?" and then explicitly ignoring a comprehensive census of every person who has ever played soccer in favour of a completely unscientific survey of a group of soccer fanatics, and only those who speak a certain language.
In other words, Valve literally told us the percentile of your mmr but people somehow seem to think that the fact that some of those people don't play ranked means that they don't count towards the percentile.
|
On January 14 2014 05:21 Staboteur wrote:
There's no way to make the game not a 5v5, but every one of those heroes is impactful and relatively team-independent. That's slightly less true of Luna - if you're sharing farm in lane life might be hard - but otherwise your current hero pool isn't a problem.
My most reliable hero for wins has always been CM. Especially at low levels, just carrying a TP and counterganking makes a massive difference, and there's never a time where you've got nothing to do, as CM can roam / woods / babysit all very easily. AND your mana aura helps circumvent the problems that would normally arise for people with shitty builds and constantly no mana.
Haha, and I just happened to win a game we were losing 15 - 40 (and a shadow demon afk) with a Luna that I managed to farm god knows how and spamming Eclipse / BKB  Sometimes, I just don't get it...
But I agree, I usually do pretty well with CM, even though my playstyle is somewhat limited to harass the lane and making sure the carry can get some last hits. I need to diversify my repertoire :D
And thanks for the other inputs. I've never tried Viper, I'll give him a go!
|
On January 14 2014 06:21 TheMute wrote:Show nested quote +On January 14 2014 05:11 Merany wrote:
So, I have a pretty bad solo ranking (~2500) but I'm really trying to improve and I feel like I'm getting fucked by team comp quite a lot. Like for the past 10 games or so, we are in the lobby, I select a hero (not confirmed yet) and position myself on the map where I'd like to be. I wait a bit to see if other people do the same, maybe try to talk about what we would need and what people want to play. And usually get zero answers. And then, when the timeout expires, it seems like everybody goes back from afk and select their favorites carries without even looking what the team comp is. If I've already confirmed a carry, congratulation, we now have a 5 carries team and we get stomped. If not, that's still a 4 carries 1 support team and the game is hard. And no kidding, this happens 4 games out of 5.
First question is: is there anything I can do against that? I'd really like to up my ranking a bit because I'm sure people don't play like that at higher levels.
And secondly, is there a hero I can pick quickly that's pretty descent no matter what my allies chose and that I will be able to make the most off ? (I've already seen the Lycan thread and it's good but I'd like other suggestions).
Heroes I usually play: Luna, weaver, prophet, enchantress, CM, shadow shaman. Try not playing all pick. I find I get better team compositions in nearly every other game mode.
I have my worst ranked games in all pick. You just get selfish people who last pick doom whe we have one support or pick 3 mids. You can get a bad drafter in CM every once and a while, but as a whole thy are better games. All pick is where the worst of dota players go.
|
On January 14 2014 06:33 Plansix wrote:Show nested quote +On January 14 2014 06:21 TheMute wrote:On January 14 2014 05:11 Merany wrote:
So, I have a pretty bad solo ranking (~2500) but I'm really trying to improve and I feel like I'm getting fucked by team comp quite a lot. Like for the past 10 games or so, we are in the lobby, I select a hero (not confirmed yet) and position myself on the map where I'd like to be. I wait a bit to see if other people do the same, maybe try to talk about what we would need and what people want to play. And usually get zero answers. And then, when the timeout expires, it seems like everybody goes back from afk and select their favorites carries without even looking what the team comp is. If I've already confirmed a carry, congratulation, we now have a 5 carries team and we get stomped. If not, that's still a 4 carries 1 support team and the game is hard. And no kidding, this happens 4 games out of 5.
First question is: is there anything I can do against that? I'd really like to up my ranking a bit because I'm sure people don't play like that at higher levels.
And secondly, is there a hero I can pick quickly that's pretty descent no matter what my allies chose and that I will be able to make the most off ? (I've already seen the Lycan thread and it's good but I'd like other suggestions).
Heroes I usually play: Luna, weaver, prophet, enchantress, CM, shadow shaman. Try not playing all pick. I find I get better team compositions in nearly every other game mode. All pick is where the worst of dota players go.
worst as in attitude? depends on what metric you're judging because the best 'skilled' pub players play almost exclusively AP.
|
On January 14 2014 05:48 Sn0_Man wrote:Show nested quote +On January 14 2014 05:37 ILoveCoffee wrote:
Does anybody have any idea what percentiles the MMRs correspond to? I find MMR to be much more meaningful if you have something to refer it to. I recommend reading the OP, its traditionally a good thing to do.
I read it but it was pre-MMR values. I was just wondering the post MMR values, they did say that the 2 will be different.
|
On January 14 2014 06:36 crms wrote:Show nested quote +On January 14 2014 06:33 Plansix wrote:On January 14 2014 06:21 TheMute wrote:On January 14 2014 05:11 Merany wrote:
So, I have a pretty bad solo ranking (~2500) but I'm really trying to improve and I feel like I'm getting fucked by team comp quite a lot. Like for the past 10 games or so, we are in the lobby, I select a hero (not confirmed yet) and position myself on the map where I'd like to be. I wait a bit to see if other people do the same, maybe try to talk about what we would need and what people want to play. And usually get zero answers. And then, when the timeout expires, it seems like everybody goes back from afk and select their favorites carries without even looking what the team comp is. If I've already confirmed a carry, congratulation, we now have a 5 carries team and we get stomped. If not, that's still a 4 carries 1 support team and the game is hard. And no kidding, this happens 4 games out of 5.
First question is: is there anything I can do against that? I'd really like to up my ranking a bit because I'm sure people don't play like that at higher levels.
And secondly, is there a hero I can pick quickly that's pretty descent no matter what my allies chose and that I will be able to make the most off ? (I've already seen the Lycan thread and it's good but I'd like other suggestions).
Heroes I usually play: Luna, weaver, prophet, enchantress, CM, shadow shaman. Try not playing all pick. I find I get better team compositions in nearly every other game mode. All pick is where the worst of dota players go. worst as in attitude? depends on what metric you're judging because the best 'skilled' pub players play almost exclusively AP.
All pick can contain both the best and worstthat dota has to offer. In general, if I want people who are at least willing to talk, I go to CM. If I want entitled assholes who yell at me for their mistakes, I'll go to all pick. It works about 85% of the time, which is good enough for me
|
Ranked is merely a subset of the whole, not a different pool entirely.
While they admitted they didn't know who would choose to be a part of that subset, it is still a certainty that the whole is the same, and thus the values hold up.
If you wish to know how you compare to those that have chosen to play enough ranked to have a visible MMR, you could pretend that the reddit survey was accurate but it clearly isn't so you are straight out of luck. I'd imagine a survey exclusively in chinese on a chinese dota site would give quite different values, for example.
|
On January 14 2014 05:11 Merany wrote:
So, I have a pretty bad solo ranking (~2500) but I'm really trying to improve and I feel like I'm getting fucked by team comp quite a lot. Like for the past 10 games or so, we are in the lobby, I select a hero (not confirmed yet) and position myself on the map where I'd like to be. I wait a bit to see if other people do the same, maybe try to talk about what we would need and what people want to play. And usually get zero answers. And then, when the timeout expires, it seems like everybody goes back from afk and select their favorites carries without even looking what the team comp is. If I've already confirmed a carry, congratulation, we now have a 5 carries team and we get stomped. If not, that's still a 4 carries 1 support team and the game is hard. And no kidding, this happens 4 games out of 5.
First question is: is there anything I can do against that? I'd really like to up my ranking a bit because I'm sure people don't play like that at higher levels.
And secondly, is there a hero I can pick quickly that's pretty descent no matter what my allies chose and that I will be able to make the most off ? (I've already seen the Lycan thread and it's good but I'd like other suggestions).
Heroes I usually play: Luna, weaver, prophet, enchantress, CM, shadow shaman.
Of the heroes you listed, I would highly suggest shadow shaman. His hex and shackle cannot be overstated, but beyond that, dropping your wards on the tier 1's with a fresh creep wave nearly guarantees a tower advantage. That tower advantage leads to a gold advantage and map control which gives your team room to farm and not get ganked.
You can ward here or there, but honestly, most people at that level don't even look at the mini-map so wards aren't the highest priority. Get items that will keep you alive like forcestaff (dagger is usually better, but takes a bit getting used to in lower levels) and if you're rolling in cash, even a scythe of vyse (sheep stick). Shadow Shaman can affect a team fight so much it isn't even funny and it's been my experience when playing in very low level games, people tend to group up and 5v5 a lot. Either that, or they refuse to ever come together and run off farming solo all game.
I could go into much greater detail, but I don't think it'd be as helpful as "pick shadow shaman, push tier 1's asap, give team gold and room to farm."
|
On January 14 2014 06:42 Sn0_Man wrote:
Ranked is merely a subset of the whole, not a different pool entirely.
While they admitted they didn't know who would choose to be a part of that subset, it is still a certainty that the whole is the same, and thus the values hold up.
If you wish to know how you compare to those that have chosen to play enough ranked to have a visible MMR, you could pretend that the reddit survey was accurate but it clearly isn't so you are straight out of luck. I'd imagine a survey exclusively in chinese on a chinese dota site would give quite different values, for example.
True, however
1. Nobody knows how normal mmr was transfered to ranked mmr. They could have added some kinda of factor to increase the difference between lowest and highest rating. It is odd that pros are as high as 6xxx when already 4100 is the 99% percentile.
2. The whole (as in all registered players) may be a terrible thing to compare you with since there will be millions of accounts with only a few games.
I guess that a possible explanation of the skewed curve might actually be all the inactive low rating, low amount of games accounts. If we got numbers from ranked mmr we could at least compare us with people who play equally as much.
|
United States12240 Posts
On January 14 2014 07:09 Kemy wrote:Show nested quote +On January 14 2014 06:42 Sn0_Man wrote:
Ranked is merely a subset of the whole, not a different pool entirely.
While they admitted they didn't know who would choose to be a part of that subset, it is still a certainty that the whole is the same, and thus the values hold up.
If you wish to know how you compare to those that have chosen to play enough ranked to have a visible MMR, you could pretend that the reddit survey was accurate but it clearly isn't so you are straight out of luck. I'd imagine a survey exclusively in chinese on a chinese dota site would give quite different values, for example. True, however 1. Nobody knows how normal mmr was transfered to ranked mmr. They could have added some kinda of factor to increase the difference between lowest and highest rating. It is odd that pros are as high as 6xxx when already 4100 is the 99% percentile. 2. The whole (as in all registered players) may be a terrible thing to compare you with since there will be millions of accounts with only a few games. I guess that a possible explanation of the skewed curve might actually be all the inactive low rating, low amount of games accounts. If we got numbers from ranked mmr we could at least compare us with people who play equally as much.
Because even when 4100 was the top 1%, that's still out of millions of players. Pros are that much better and are closer to the top 0.001% where they could easily maintain a 1000 or 2000 rating gap over other so-called "top" players.
|
On January 14 2014 07:09 Kemy wrote:
1. You don't know if they didn't add some kinda of factor from normal matchmaking to ranked matchmaking to increase the difference between lowest and highest rating. It is odd that pros are as high as 6xxx when already 4100 is the 99% percentile.
Well, if Pro's were only the top 1% of players there would be over 70,000 professional players.
Unsurprisingly, there are not 70,000 professional dota players.
MMRs follow a bell curve, so as you get closer to the "fringes" (aka top), the amount of MMR you need to increase to go up one percentile also increases exponentially.
So in short it is not even a little bit odd, but people certainly don't understand statistics very well.
On January 14 2014 07:09 Kemy wrote:
2. The whole (as in all registered players) may be a terrible thing to compare you with since there will be millions of accounts with only a few games.
If you want to know how you are compared to your friends, compare your mmrs. If you want to know how good you are compared to the world, we have that data. If you want to compare yourself to another specific subset of players, you are kinda out of luck in this current system.
The point is to play dota not to circlejerk over numbers lol.
|
On January 14 2014 07:09 Kemy wrote:Show nested quote +On January 14 2014 06:42 Sn0_Man wrote:
Ranked is merely a subset of the whole, not a different pool entirely.
While they admitted they didn't know who would choose to be a part of that subset, it is still a certainty that the whole is the same, and thus the values hold up.
If you wish to know how you compare to those that have chosen to play enough ranked to have a visible MMR, you could pretend that the reddit survey was accurate but it clearly isn't so you are straight out of luck. I'd imagine a survey exclusively in chinese on a chinese dota site would give quite different values, for example. True, however 1. Nobody knows how normal mmr was transfered to ranked mmr. They could have added some kinda of factor to increase the difference between lowest and highest rating. It is odd that pros are as high as 6xxx when already 4100 is the 99% percentile. 2. The whole (as in all registered players) may be a terrible thing to compare you with since there will be millions of accounts with only a few games. I guess that a possible explanation of the skewed curve might actually be all the inactive low rating, low amount of games accounts. If we got numbers from ranked mmr we could at least compare us with people who play equally as much.
It does make sense that you can reach 6k but 99% is 4,1k. 99% isn't actually that good at all, 6k is a rank very few people have, it's % is minuscule.
|
On January 14 2014 07:17 Sn0_Man wrote:Show nested quote +On January 14 2014 07:09 Kemy wrote:
1. You don't know if they didn't add some kinda of factor from normal matchmaking to ranked matchmaking to increase the difference between lowest and highest rating. It is odd that pros are as high as 6xxx when already 4100 is the 99% percentile.
Well, if Pro's were only the top 1% of players there would be over 70,000 professional players. Unsurprisingly, there are not 70,000 professional dota players. MMRs follow a bell curve, so as you get closer to the "fringes" (aka top), the amount of MMR you need to increase to go up one percentile also increases exponentially. So in short it is not even a little bit odd, but people certainly don't understand statistics very well. Show nested quote +On January 14 2014 07:09 Kemy wrote:
2. The whole (as in all registered players) may be a terrible thing to compare you with since there will be millions of accounts with only a few games. If you want to know how you are compared to your friends, compare your mmrs. If you want to know how good you are compared to the world, we have that data. If you want to compare yourself to another specific subset of players, you are kinda out of luck in this current system. The point is to play dota not to circlejerk over numbers lol.
Why is each of your posts in such an aggressive tone? This discussion is silly. Valve could easily publish ranked mmr percentiles and that's what I (and some others in this thread) would like to have. This is not something you have to argue about so fiercely. Even if you mock my statistics knowledge, I would still like that because comparing myself to somebody who just played his first 10 games just doesn't mean anything at all. If that would be interesting I could stop playing and watch myself getting into higher and higher percentiles because there will be way more new players with a worse rating than me who stop playing than ppl overtaking me.. Yay, maybe in a year or 2 I will be top 0.5% without doing anything!
|
? you don't change percentiles without changing ur ranking. You don't start at 0 lol.
Apparently u'll be insulted if I point out that this is showing again how little most people understand about MMR and why they shouldn't make dumb comments lol.
"What percentile am i?"
> Relax, You're doing fine
|
Canada2068 Posts
On January 14 2014 09:59 Sn0_Man wrote:
? you don't change percentiles without changing ur ranking. You don't start at 0 lol.
Apparently u'll be insulted if I point out that this is showing again how little most people understand about MMR and why they shouldn't make dumb comments lol.
"What percentile am i?"
> Relax, You're doing fine
what if he's at 4100 mmr (1%) atm and in the space of a few years, 10M smurf accounts are created that start from avg mmr (say 2500). then won't he have gone from 1% -> 0.5% as he claimed?
![[image loading]](https://dl.dropboxusercontent.com/u/90663256/TL/mmr.png)
i think you shouldn't call his comment dumb when you come up with statements that are demonstrably wrong like "? you don't change percentiles without changing ur ranking. You don't start at 0 lol."
|
On January 14 2014 09:59 Sn0_Man wrote:
? you don't change percentiles without changing ur ranking. You don't start at 0 lol.
Apparently u'll be insulted if I point out that this is showing again how little most people understand about MMR and why they shouldn't make dumb comments lol.
"What percentile am i?"
> Relax, You're doing fine
Let's assume
- another 10 million players download this game, play 10 games and leave again.
- 10 games is not enough to find your "real" rating. MMR system may adapt not fast enough for you to get there or you need more time to adapt to the controls etc.
If I'm already in the top 1% percentile before these ppl join chances are almost none of those will/can surpass me in 10 games alone because their mmr will not adapt fast enough to their skill or their skill takes a few more games to materialize but the playerbase has increased by 10 million which will put me in a higher percentile of the new playerbase.
That's why active playerbase (with 150+ games to get in ranked mode) is more important for me than the whole playerbase which includes all these people who quit before they got where they belonged.
Am I making a mistake here? I should just go to bed I guess..
|
amusingly, 10 games is enough to reach 5K mmr based on a LOT of anecdotes of very highly ranked players who have used smurfs and then complained that their 6 (or less) win smurf is higher ranked than their 300+win main.
So unless the system is actively "gamed" the examples above basically don't happen.
I suppose that I should admit that yes, the vast majority of new players will likely be below your MMR at the point that they join so if they don't actually improve eventually you would trend to a higher percentile. Assuming your rating is reasonably high.
|
How do you climb ranks as a support character assuming solo? I can gank, make good saves, pull, and then the team will do very dumb things like 4 v 5'ing, getting hooked over and over, lack map awareness. This is around 3800 too (or was). Long 5-9 game losing streaks are common.
What are you supports doing in your in own solo games?
|
|
|
|
|
|
EPT
