General Discussion

General DiscussionMMR in percentage

MMR in percentage in General Discussion
Zus

    So relentless and a Few other people came up with that. I think relentless said only 10% of Dota 2 players are 3K or something, today I heard a 5k players say the average amount if Dota players are 3K and 5k is 2-3%ish. Does anybody know the real percentage?

    Crudude

      I have no Idea, but I saw in a few places that the majority of DOTA players are stuck around 3k - 3.5k

      BenaoLifedancer

        dotabuff has some numbers, ask them to post an update (they did once before like a year ago)

        Crudude

          If you think about it, the way the system works, it is going to change based on new players joining dota. The new players will go lower and lower.

          Theoretically everyone in 5k, 4k, 3k are always improving, and thus the calibre and quality of 3k is actually going up because they are keeping each other down (+24/-24). So what a 3k skill player was a year ago would have probably only been 2k right now

          Този коментар е бил редактиран
          Seoulmate

            The problem with getting dotabuff's numbers is it is only taking into account the people who have signed in through dotabuff. Many players who play the casually don't use sites like this and are typically lower mmr's.

            THICC BABY SHUM

              dotabuff ppl told me that 75% of ppl are normal skill 15% are high skill and 10% are very high skill
              Seems legit

              Relentless

                YNIT you are referring to the stats for games played, not players.

                The Very High players play a disproportionately large number of games. This is in agreement with Valve's original stats on MMR distribution.

                Crudude is correct. There appears to be some amount of MMR deflation. The value of MMR points has been drifting up over time so that while the distribution remains the same % wise players at a give MMR level are better now than players with that same MMR and percentile rank a year ago were at that time.

                To reiterate - a player, having improved significantly over a year ago remains the same MMR despite actually being better at dota because the entire community is better. The pros are also better now than in years past. This trend will continue for a few more years, with less impact each year as the community matures.

                The only way to gain MMR is to directly outplay and defeat higher ranked players moving them to lower MMRs. This is why it is denotatively impossible for everyone to move up. Only 1% can be in the top 1%. Only 10% can be in the top 10%.

                Този коментар е бил редактиран
                Rocket

                  Is there definitely mmr deflation? Seems to me that people who feed mmr are more likely to quit and thus mmr for active players will suffer from inflation.

                  TripleSteal-

                    Would be interesting to see the percentiles, but I think it can't be extracted from stram API, only the percentiles for games can (and that is what YNIT/Relentless alrdy told about).
                    So I don't think dotabuff can do it, although I'm not sure tbh.

                    Mokujin

                      there is data from 8 months ago with the normal distribution

                      Този коментар е бил редактиран
                      Mokujin

                        i'll find it when i finish my game

                        Dan

                          If you guys are talking about this?

                          5% 1100
                          10% 1500
                          25% 2000
                          50% 2250
                          75% 2731
                          90% 3200
                          95% 3900
                          99% 4100

                          Dan

                            It'd be nice if valve would release some actual stats for ranked matchmaking or maybe some sort of live chart which updates over time showing the average MMR, Although i guess the reasoning behind not making it public is the fact it'll make bad players feel like shit.

                            Guess that's why we don't have the Very low, Low, Normal, High, Very high brackets.

                            Този коментар е бил редактиран
                            Relentless

                              Yes, that is the data Valve provided. It shows the broad structure of how their MMR curves are set for unranked games. Originally they thought ranked might be different because they thought lower MMR players might be less likely to play ranked. But turned out to not be true. An even distribution of players down to sub 1k MMRs plays ranked. So Valve has not updated their distribution of MMR percentiles.

                              While lower MMR players play ranked matches, they are much less likely to be interested in publicizing their scores. So data we might be able to garner from dotabuff or other websites that provide stats services is significantly biased toward the higher MMRs.

                              Relentless

                                Valve did have some statement with a graph, once if I remember correctly, explaining how a players MMR will plateau over time.

                                Maybe I'm just remembering this graph. http://blog.dota2.com/2013/12/matchmaking/
                                They show differing paths to reach the same stable MMR over time.

                                Този коментар е бил редактиран
                                Reese

                                  5% 1100
                                  10% 1500
                                  25% 2000
                                  50% 2250
                                  75% 2731
                                  90% 3200
                                  95% 3900
                                  99% 4100

                                  "Note that this distribution is from normal matchmaking. We don’t know yet what the distribution will be in ranked matchmaking, but we expect it to be different."

                                  This is normal matchmaking rating in 2013

                                  Dan

                                    Thats from reddit though so the average is probably more like 2.5k i guess.

                                    TripleSteal-

                                      this surveys give non-representative data cz the sampling is not random; there are errors due to correlation between chance of participation in the vote and actual mmr of a person, so the results are biased.

                                      Light.k

                                        dota used to be able too see how many people are actually in-game.(not idleing, but in server)

                                        I've count few 4.5k avg game, using how many people are actually playing combine with Live game pages.
                                        4.5k was around ~2%.

                                        And another 5.3k avg game was ~0.3%.

                                        Two problem,
                                        1. Not all people are playing ranked game.
                                        2. Not all live games are ranked game.

                                        TripleSteal-

                                          by the way i think the data from NMM Steam postedis still viable, cause relative difference in skill probably stands the same, while the current data for RMM would be far from ideal due to massive amount of smurfs and not everyone playing RMM. These errors were less important before RMM was released cz everyone played same NMM, and people used smurfs much less than they do now.

                                          Mokujin

                                            Sure, the mean is most likely skewed + 1σ (or more) towards higher mmr than the 'real' average because of the biased sample but you can still get an overall look and more or less put boundaries on where the real values can be

                                            Този коментар е бил редактиран
                                            Mokujin

                                              like if you adjust the y axis scale so that the percentile values on the graph match the existing skill bracket values
                                              e.g. normal high and very high percentages, you can get a good guess i think

                                              Този коментар е бил редактиран
                                              Andy

                                                Maybe i'm the only one who wants this, but I think that dota should also get divisions (I am not talking about dotabuff, I mean the actual game ranking) like leagues. I think that's one of the biggest pros of leagues if you compare it to dotka.

                                                Von Darkmoor

                                                  Ranked.... Who the fuck cares their just as bad as in normal thats why i quit playing it rather play a good captain and have a retard ruin in it then go next than have a captain ruined and also lose point to make me even more pissed off serves absolutely no point game is game by game anyways and if you got no brain and no skill or idea of what to do just go play allpick ranked unranked doesnt mather both is equally braindead.

                                                  (My opinion on it.)

                                                  Този коментар е бил редактиран
                                                  TripleSteal-

                                                    ^the mmr you achieve in this braindead ranked tends to accurately represent your skill though, it's just mathematically obvious.
                                                    also, new new ranked ap mode is not that far from cm in terms of balance, and pick stage takes less time. imo this mode is one of the greatest things that were introduced in dota in the last several years.

                                                    sano

                                                      "Originally they thought ranked might be different because they thought lower MMR players might be less likely to play ranked. But turned out to not be true."
                                                      Can you provide a source please
                                                      You love to spew information with no evidence at all except your opinion

                                                      nami

                                                        Why does Sano hate Relentless so?

                                                        Anyway, there's too much analysis regarding MMR. If you're good, you go up. If you're bad, you go down. Knowing where you stand in the entire populace doesn't really do anything for you except make you feel better (cuz majority of players are <3k or some shit like tat).

                                                        sano

                                                          Person A states X as if it was a fact.
                                                          Person B asks for evidence regarding X because there are no clear reasons to believe X.
                                                          Person A provides no evidence, but still claims X is true as if it was obvious.
                                                          Person B asks for evidence again.

                                                          Clearly person B hates person A.

                                                          Relentless

                                                            Should I give him a chance to find it himself? Or wait while he stews and rages for a month and then show him? He won't accept evidence anyway, he doesn't want to know he is wrong.

                                                            Този коментар е бил редактиран
                                                            sano

                                                              Show me 1 post where I raged at you. Also show me 1 post where you provided evidence regarding your claim about MMR distribution.

                                                              jus chillen

                                                                you are raging inside, in your SOUL

                                                                PROTECTHIMFROMWHATHEWANTS

                                                                  SANO CHESS MASTER

                                                                  Zus

                                                                    Plis do ULTRA HIGH SKILL for 4,5k -7k dotabuff!!!

                                                                    Flyingpigs

                                                                      Sano the soul rager