|
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. |
United States47024 Posts
On January 08 2014 04:05 wuhan_clan wrote:
But I still think that when someone is just a little bit better than their current MMR, the climb up is a slower process than people are expecting. A lot of variables have to work out in your favor and when your own contribution isn't reliably the main catalyst, a climb of say 500 elo may be slow.
The idea that you're even looking at "a little bit" of MMR means you're not understanding the system.
8-10 game winstreaks are very common among players at all levels, meaning that a change even as "large" as 200-300 MMR doesn't represent a significant difference in skill unless you're at the extreme top or bottom end of the ladder.
MMR differences are by and large only significant over large ranges. Even a couple hundred MMR can represent a lot of random fluctuation, so you really just shouldn't be reading that precisely into it.
As an aside, this is arguably why League of Legends switched from having visible MMR to having "leagues" that roughly represent a range of ability. The majority of people try to extract more precision out of numerical MMR systems than they're actually capable of, so they start treating MMR differences that are not statistically significant as representing significant skill differences.
|
United States12248 Posts
On January 08 2014 04:15 TheYango wrote:Show nested quote +On January 08 2014 04:05 wuhan_clan wrote:
But I still think that when someone is just a little bit better than their current MMR, the climb up is a slower process than people are expecting. A lot of variables have to work out in your favor and when your own contribution isn't reliably the main catalyst, a climb of say 500 elo may be slow. The idea that you're even looking at "a little bit" of MMR means you're not understanding the system. 8-10 game winstreaks are very common among players at all levels, meaning that a change even as "large" as 200-300 MMR doesn't represent a significant difference in skill unless you're at the extreme top or bottom end of the ladder. MMR differences are by and large only significant over large ranges. Even a couple hundred MMR can represent a lot of random fluctuation, so you really just shouldn't be reading that precisely into it. As an aside, this is arguably why League of Legends switched from having visible MMR to having "leagues" that roughly represent a range of ability. The majority of people try to extract more precision out of numerical MMR systems than they're actually capable of, so they start treating MMR differences that are not statistically significant as representing significant skill differences.
Yeah. If you're blue/pink and I'm brown/orange in a game and the rating difference between us is 20 and everyone on our team screams "LET BLUE MID", what does that really mean? A few minutes ago you won your last game and that's why you're 20 rating above me. Are you really that much better after a single game? The tables could just as easily have turned (say you lost your last game instead). Most of the time the gap isn't going to be pronounced enough to matter.
Now, if the gap between us were 1000 rating and 100 games later we compare ratings again and the gap remains 1000, that's probably information we can use.
|
The moral of the story is that you're better off just ignoring the colors and taking positions and match ups your are comfortable with. Trying to game the MMR system to make sure the bestest player is in mid doesn't really do anything but creat conflict in a setting where working together is preferable.
|
I actually do much worse when I'm the high-mmr player on a whole since to my experience it feels like the MMR balance expects me to take a snowbally hero mid and pubstomp, and thats just not what I'm good at.
I can respect why you'd want your impact player in an "impact position" like mid, but sometimes I struggle to be that sort of an "impact" player.
PS i fuxing hate playing support in 6.79 cuz early game theres nothign to fucking do you either leech from your carry or you roam which ends up in retarded 3v3 mid shitfests where ur spending all ur support gold buying smokes so that you can run into their support pair also smoking around.
|
3 games out of the last 5 didnt count, well played volvo.
|
for some reason the loss is counted but not the win y volvo
|
All my games counted but there was a delay on some, they might be there in a bit ..
|
On January 08 2014 05:09 Erasme wrote:
for some reason the loss is counted but not the win y volvo
It's valve's secret plot to decay mmr so that blizzard feels better about their hilariously broken system.
|
Man when you see you have a similar rank with Sheever you know MMR is fucked. Im not good but im not that terrible 
|
On January 08 2014 04:15 TheYango wrote:Show nested quote +On January 08 2014 04:05 wuhan_clan wrote:
But I still think that when someone is just a little bit better than their current MMR, the climb up is a slower process than people are expecting. A lot of variables have to work out in your favor and when your own contribution isn't reliably the main catalyst, a climb of say 500 elo may be slow. The idea that you're even looking at "a little bit" of MMR means you're not understanding the system. 8-10 game winstreaks are very common among players at all levels, meaning that a change even as "large" as 200-300 MMR doesn't represent a significant difference in skill unless you're at the extreme top or bottom end of the ladder. MMR differences are by and large only significant over large ranges. Even a couple hundred MMR can represent a lot of random fluctuation, so you really just shouldn't be reading that precisely into it. As an aside, this is arguably why League of Legends switched from having visible MMR to having "leagues" that roughly represent a range of ability. The majority of people try to extract more precision out of numerical MMR systems than they're actually capable of, so they start treating MMR differences that are not statistically significant as representing significant skill differences.
Oh I get how ELO works and natural fluctuation due to uncertainty in a small sample of games.
I quoted 500 MMR. But thats besides the point. The point is, what exactly is a significant difference in MMR? And at what difference in MMR can there be noticeable difference in skill? After how many games can we see reliable trends not due to natural fluctuation? These are questions with no exact answer but the numbers obviously start to matter at some point. Maybe the tier system you mentioned is more suitable than a single concrete number.
Also, is this Blue>Orange, Pink>Brown spectrum real? I know it doesn't matter but has this actually been confirmed?
|
On January 08 2014 05:15 SkelA wrote:Man when you see you have a similar rank with Sheever you know MMR is fucked. Im not good but im not that terrible
sheever plays quite a lot also its easier to see the mistakes when you're watching than when you're playing :p
dont forget that when most people watch her its because she stacked with singsing or some equally skilled player which put her against stronger opponents
|
The color issue appears to be pure speculation based on some annicdotal evidence and some pro streams. It appears to have some relationship to how match making works and the person with the highest MMR appears to get blue, but that may not be the case.
|
anyone know how captain is decided in a ranked game? i'd like to know
i mean is it random?
sometimes I get the "become captain" button and sometimes I don't.
when I get the button does it mean I have to click it first (i.e. someone else also has the button and can out-click me to it)?
what's happening when I don't get the button?
|
the color thingy probably only works (or matters) at the highest end. 30k people could see dendi getting blue/pink everytime
|
Press "be the captain" button first.
|
On January 08 2014 05:24 evanthebouncy! wrote:
anyone know how captain is decided in a ranked game? i'd like to know
Whoever clicks the "become captain" button the fastest when everybody has loaded and the picking phase starts.
|
On January 08 2014 05:27 Sn0_Man wrote:Show nested quote +On January 08 2014 05:24 evanthebouncy! wrote:
anyone know how captain is decided in a ranked game? i'd like to know Whoever clicks the "become captain" button the fastest when everybody has loaded and the picking phase starts.
so i'm guaranteed to always have the button? is this true for both solo queue and party queue?
like sometimes I don't se the button
|
That would be because you are too slow in that case.
As in, yes the button is always there in modes with a captain.
|
On January 08 2014 05:15 SkelA wrote:Man when you see you have a similar rank with Sheever you know MMR is fucked. Im not good but im not that terrible
Maybe you are bad at movement, imo the hardest part of dota. Being at the right place at the right time and creating opportunites. Pieliedie is really good at that part. Commentators never pick up on it because he just "is there", there is nothing flashy at all about it. But it requires tremendous game sense to be able to move like he does.
If you queue into Sheever then maybe she just is better at other points than you while you are better at landing spells. Also judging her from playing with Sings stacks is pretty unfair because when you play vs vastely superior opponents its hard to not feed. Try stacking with 6k+ people and see how well you do.
|
As someone who has played against 5.5k people in the TL in house, you wil be in awe how poorly you will do. You will walk out of position for half a second and just die before you can even figure out what happened. If anything, you should respect sheevers ability to hang with the heavy weights.
|
|
|
|
|
|
EPT
