|
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. |
On December 17 2013 08:29 wuhan_clan wrote:Show nested quote +On December 17 2013 07:35 Sn0_Man wrote:On December 17 2013 07:33 DrPandaPhD wrote:
High KDA supports like Enchantress / Wisp
Low KDA supports like Dazzle / Chen
Way easier to achieve high KDA with Storm compared to Chen Well ur not playing 4/4/0/0 chen with dagon often enough i see. Huehuehue... KDA is a dumb term, KDR is what people usually mean. Assists are incredibly irrelevant in dota since gold is given as an AoE not to "people who damaged" anyway. How can you say assists are incredibly irrelevant? It's probably the MOST relevant. Your kills participation (kills + assists relative to team's total kills) is a pretty good indicator of skill more often than not. Low assist count usually means you are a bad support who isn't active on the map. Low assist count is maybe acceptable on Treant, since your living armor casts won't be reflected in any stats. In your KDA vs KDR comparison, the person who went 5-5-0 might have an even KDR but I would deem him less valuable than the person who went 0-5-10.
not neccessarily true, wherein the problem with using kdr/kda for mmr lies
i just think its wrong to give you someone an attempt of actual "player skill" instead of a simple ranking system
btw the system seems really bugged i can randomly see the mmr stats of some people (not all but quite a few) if i just check their dota profile
|
On December 17 2013 08:37 teddyoojo wrote:Show nested quote +On December 17 2013 08:29 wuhan_clan wrote:On December 17 2013 07:35 Sn0_Man wrote:On December 17 2013 07:33 DrPandaPhD wrote:
High KDA supports like Enchantress / Wisp
Low KDA supports like Dazzle / Chen
Way easier to achieve high KDA with Storm compared to Chen Well ur not playing 4/4/0/0 chen with dagon often enough i see. Huehuehue... KDA is a dumb term, KDR is what people usually mean. Assists are incredibly irrelevant in dota since gold is given as an AoE not to "people who damaged" anyway. How can you say assists are incredibly irrelevant? It's probably the MOST relevant. Your kills participation (kills + assists relative to team's total kills) is a pretty good indicator of skill more often than not. Low assist count usually means you are a bad support who isn't active on the map. Low assist count is maybe acceptable on Treant, since your living armor casts won't be reflected in any stats. In your KDA vs KDR comparison, the person who went 5-5-0 might have an even KDR but I would deem him less valuable than the person who went 0-5-10. not neccessarily true, wherein the problem with using kdr/kda for mmr lies i just think its wrong to give you someone an attempt of actual "player skill" instead of a simple ranking system btw the system seems really bugged i can randomly see the mmr stats of some people (not all but quite a few) if i just check their dota profile
I don't think you understand what I was saying and if you read your 2nd sentence, you'll find it makes zero sense. What I'm saying is that K-D-A is a more relevant factor than strict KDRatio. But I'm not saying that either should be the definitive factors when determining MMR. I was just refuting someoene's claim that assists are irrelevant, which in fact the opposite is true. They are extremely important to nearly every hero in most roles.
|
I don't get to play my ranked calibration games 'cause apparently the network is down. But I'll be pleasantly shocked if I'm at 2k lol
|
United States12280 Posts
On December 17 2013 08:28 teddyoojo wrote:Show nested quote +Let's look at an example. My mean rating is 6000 and my uncertainty is 500 so my MMR would be 4500. I just finished a game where I am heavily favored to win. My team gains 300 points to be distributed between our players. The system adds 50 uncertainty before the game and my uncertainty drops to 480 so I am at 530. My team's total uncertainty is 2000 so I get (530/2000)x300=about 80 points. My new MMR would be 6080 - (3x 530) = 4490. A net decrease of 10 MMR."
is that actually true?!?!?!?!?
That's from the Playdota post, right? That guy is mistakenly applying the Microsoft TrueSkill system to Valve's, when they have said so often that they use a modified Elo system. The "mean rating" IS the MMR value, and the uncertainty extends along both sides. He's also wrong in assuming that uncertainty reaches zero. It's not this way in SC2, and Valve probably has good reason to ensure it never reaches zero as well.
Valve said that it would be possible for MMR to increase after a loss or decrease after a win. This could happen in one of two scenarios. The first is when you're in your calibration phase where individual performance matters (to what degree, we don't know, and we don't know how much impact it has after that period, if any). The second is, depending on how they handle uncertainty, potentially similar to how the guy described, only without the 3*sigma TrueSkill part. Individually, you could have your MMR fall, but your team's average MMR would rise after you win.
|
the whole rating thing doesn't really work for dota anyways. I was randoming my last 100ish games before they added ranked, losing a good bunch with heros i never play. I think because of that my initital rating was lower than it should have been. I had a lot of easy games in ranked because i just pick a hard carry or mid hero that can at least semi-carry and win most of my games, if i pick anything else it is a gamble if my teams carry can farm or not.
currently at 3500 mmr
|
Well well My solo rating is at 3704 gotta improve that ! #roadto5K !
|
On December 17 2013 18:14 Yarbleck wrote:
Well well My solo rating is at 3704 gotta improve that ! #roadto5K !
I am curious, how good is this considered? I think I am roughly around this area too from all the people that I have been asking while queing. From what I see, 5000+ = GrandMaster, 4000-5000 = Master, and 3000-4000 = Diamond? Of course this is just rough estimate but am i correct in assuming this?
|
No one really knows the percentages yet, valve said themselves the numbers they gave don't apply to ranked.
|
On December 17 2013 18:37 SheaR619 wrote:Show nested quote +On December 17 2013 18:14 Yarbleck wrote:
Well well My solo rating is at 3704 gotta improve that ! #roadto5K ! I am curious, how good is this considered? I think I am roughly around this area too from all the people that I have been asking while queing. From what I see, 5000+ = GrandMaster, 4000-5000 = Master, and 3000-4000 = Diamond? Of course this is just rough estimate but am i correct in assuming this?
what I know is that pro/pubstar are at 5k+ merlini is 5300 dendi is 5600. (from their stream)
so 5k+ is god like top 0.0X%
4-5k is top 0.1%
I am just making % up from what i saw on dotabuff reddit etc..
|
Oh wow, what is this? I loved my good old solo queue games. They never failed produce even games with skilled players. I just tried the new solo ranked matchmaking just to realize that it is fucking terrible! No solo queue, the skill difference between players is just terrible and the overall skill level seems to be alot lower than what I used to play before.
I just had a Captains Draft with literally noob players that never played their drafted heroes before and feeded like shit. A mid pudge that lost mid so horribly that I cant even describe it. I won my Lone Druid solo offlane vs trilane easily with good farm and 3-0.
I am just pissed right now...
|
On December 17 2013 20:05 DennyR wrote:
Oh wow, what is this? I loved my good old solo queue games. They never failed produce even games with skilled players. I just tried the new solo ranked matchmaking just to realize that it is fucking terrible! No solo queue, the skill difference between players is just terrible and the overall skill level seems to be alot lower than what I used to play before.
I just had a Captains Draft with literally noob players that never played their drafted heroes before and feeded like shit. A mid pudge that lost mid so horribly that I cant even describe it. I won my Lone Druid solo offlane vs trilane easily with good farm and 3-0.
I am just pissed right now...
they take your old mmr and match you with players around your skilllevel - a single game is not enough sample size to judge the ranked mode. Also Pudge is supposed to lose mid, he will almost always have less cs than any standard ranged mid.
|
So i need some people to go ranked cm with me on eu. Where should i post this? How do we arrange a good party to go ranked? Do we need a new thread for finding ranked partners?
|
Czech Republic18921 Posts
On December 17 2013 20:05 DennyR wrote:I just had a Captains Draft with literally noob players that never played their drafted heroes before and feeded like shit. A mid pudge that lost mid so horribly that I cant even describe it. I won my Lone Druid solo offlane vs trilane easily with good farm and 3-0.
You got matched with players at approximately your skill level. That's how the system works.
|
On December 17 2013 20:25 idonthinksobro wrote:Show nested quote +On December 17 2013 20:05 DennyR wrote:
Oh wow, what is this? I loved my good old solo queue games. They never failed produce even games with skilled players. I just tried the new solo ranked matchmaking just to realize that it is fucking terrible! No solo queue, the skill difference between players is just terrible and the overall skill level seems to be alot lower than what I used to play before.
I just had a Captains Draft with literally noob players that never played their drafted heroes before and feeded like shit. A mid pudge that lost mid so horribly that I cant even describe it. I won my Lone Druid solo offlane vs trilane easily with good farm and 3-0.
I am just pissed right now... they take your old mmr and match you with players around your skilllevel - a single game is not enough sample size to judge the ranked mode. Also Pudge is supposed to lose mid, he will almost always have less cs than any standard ranged mid.
I mean he lost mid, got solo killed ALOT! And I know how this system is supposed to work, but the matches I played with the new system (3) were just terribly onesided and the overall skill level was just waaaay lower than what I am used to.
Edit: just to clarify, I was high-very high the last time you could check your mmr
|
I'm not done with my first 10 matches in ranked mm, but I think all games so far (6-8) have been close, even and tense; unlike before, when the conclusion was rather obvious after 20 minutes in most games. The downside is, this makes losing a little tougher, especially when you have a lineup that peaks earlier than the opponents, but your promising pushes get screwed by somebody diving into their base and dieing when you destroy a tower/rax instead, or someone getting picked off needlessly alone a few seconds after the team decided to "push mid in 1 minute". But in general, better manners and less feeding.
|
solo 4300 
|
I think they removed information about uncertainty and probable rating in dota_game_account_debug
....
low_priority_games_remaining: 0
solo_calibration_games_remaining: 1
recruitment_level: 0
secondary_leaver_count: 0
last_secondary_abandoned_game_date: 0
....
|
On December 17 2013 21:52 PetrBlaha wrote:
I think they removed information about uncertainty and probable rating in dota_game_account_debug
....
low_priority_games_remaining: 0
solo_calibration_games_remaining: 1
recruitment_level: 0
secondary_leaver_count: 0
last_secondary_abandoned_game_date: 0
....
Have you played your first ranked game already? Because I still have it, maybe you can only see it if you have played your 10 games though
|
On December 17 2013 21:55 Jetaap wrote:Show nested quote +On December 17 2013 21:52 PetrBlaha wrote:
I think they removed information about uncertainty and probable rating in dota_game_account_debug
....
low_priority_games_remaining: 0
solo_calibration_games_remaining: 1
recruitment_level: 0
secondary_leaver_count: 0
last_secondary_abandoned_game_date: 0
.... Have you played your first ranked game already? Because I still have it, maybe you can only see it if you have played your 10 games though
they removed it with last patch
|
I'm so fucking tired of having people blaming mid for failing their lane... like at 4 min our dazzle/pa lane was getting wrecked and asking "magnus why no gank come gank come gank". Of course they fed like hell because bristback was snowballing and despite magnus doing dazzle went 0-21...
I understand that it feels bad to see that your lane is snowballing the wrong way, but don't put the blame on others. I was playing earth spirit so I tried to help but even then it was too late.
|
|
|
|
|
|
EPT
