General Discussion

General DiscussionWin-Loss Streaks and Matchmaking Explained!

Win-Loss Streaks and Matchmaking Explained! in General Discussion
TripleSteal-

    ye okkams the other thing i thought of to
    unnecesary irrational complications make no sense to me

    Jacked

      @story time

      finally someone with a brain.

      MMR is inherently variable. it deviates about its true mmr value thus inevitably creating loss/win streak patterns.

      -vzta-

        If BS were a real thing in the matchmaking system there should be teammates and enemies with the same BS as you, so its the same shit, if you're better than them you have to climb anyways.

        Ce commentaire a été édité
        kanra

          I think this does make sense as I had a 4 game lose streak and when I played today I got good or decent teammates and I won xD (......got carried though, if the game was a bit longer I would have a better kDa xD)

          plaku]

            wow, it seems some people actually are "waking up' and not living anymore in the illusionary world which cookie and the other members of his 'religion' created . lol. dont take him serosly bro he was critic even in my comments about beavhior, and he admited he have no idea about it, 0 knowledge in that field. meanhile imade some researches on it and imstill testing this, till now i have 3 acc with same mmr, and different BS. and i can tell everyone its has a huge in impact in game play, im not going to explain it here in details cause i dont have too, everyone can check this by your own its not that hard to take 1 acc with 6000 BS and aother with 9000-10000 BS and with same mmr lets say 2500. you can tell byyourself the differnces in gameplay and how your team changed and how muc hthat impact your winrate. cookie its coomited rly hard against this cause he wants that his guides ' theories' unprovel aslo, to make sense, he doesnt realize or he dont want too, that bs has agreat impact in games , cause he doesnt know he always played in acc with 8000 and above which is not bad considering, 6000 and below. there his theories to solo carry cant work for shit, iwould like to see cookie himself in an 2k acc with below 6000 bs and see if he can maintain an winrate above 50% . at least.

            Celine

              I skip wall of gibberish

              plaku]

                omly pos 4 -3. as i know its doesnt work like that. its not same as mmr, if your bs is 7000 you will get only teammates with around 7000. and the enemy can be 9000 or 10000 or 3000 it doesnt matter. but fomr wahti know 9000 is average, so anything below that will lead you having worste team than your enemy in most cases, if its 6000 or below your scruwed it is 99% that you have a much worse team .with this sais here is an example. its 2-3 people fighting for mid, i jungler afk 20, mostly of the time offlnaer is a hard carry cause you have 5 carry , if yo u supprting your completely fucked , always your team threw . always team is trashtalking each other sine in draft, and it continues all game, the flame mostly for noneses, like omg retarded wk why quelling blade?. lol. and many other scenarios . who ever experienced this go check your bs right now and get rid of your acc, and most important change attitude , never flame even if you have 4 retareds in team cause they will reprot you even if your 33-2.

                >tfw no gamer girl peeing gf

                  Wow this actually makes sense and explains the constant +24 I get.

                  +24 means my team had a slight advantage over the other? Or is that +26? Because I rarely get +25. It is always +24 or ocassionally +26 with a very rare ocassion of the normal +25.

                  And my last party recalibration was -20. I should not have randomed since I got Puck (a hero I literally never play) and got my stats fucked further.

                  >tfw no gamer girl peeing gf

                    And I can confirm Behavior Score does have a huge impact in match making. I have been all across the board, from 3k - 9k.
                    Currently, I am sitting at 5.8kBS and I personally find it as the worst. At least 3kBS has players that know how to play, but they are very toxic. 5kBS seems to have very VERY passive players or players that make bad choices.

                    Whoever says Behave Score has no impact in matchmaking is blowing words out of their butthole.

                    Jacked

                      Lol just go play mafiosos account which is supposedly impossible to win on because of toxic teammates. Any monkey can win

                      Shou

                        cat ur actually braindead
                        how tf can u say with certainty that u got above specifically higher amts of cancerous players at 3k BS as compared to 9k BS, how can u know for a fact that it isn't ur confirmation bias and specific memories of more traumatic events and triggering players that make it FEEL like 3k BS is more toxic than 9k. HOW CAN U KNOW FOR SURE?
                        WHERE IS THE PROOF?
                        THE ONLY ONE BLOWING WORDS OUT OF THEIR BUTTHOLE IS YOU, HE WHO IS MAKING STATEMENTS THAT ARE COMPLETELY IMPOSSIBLE TO PROVE.
                        kys
                        just a big excuse for ur inadequacies.

                        as for u Mafioso ur so autistic I'm not gonna bother talking to u.
                        queue Mafioso's teapot.
                        can someone go photoshop some sort of image to spam abt Mafioso's teapot whenever he makes a post?

                        Ce commentaire a été édité
                        Bungmangler

                          I just checked my score and its 9771.

                          I get nice people. I get super toxic twats too. So apparently it doesn't matter as much as people think.

                          I think the average dota player is toxic and therefore its not a surprise that at least a few people in games are total asshats. Law of averages I guess

                          Ce commentaire a été édité
                          RayZLiENT

                            @Bungmangler & Others,

                            If you read my initial post again, you will find that I have mentioned that your Behavior score is cumulative and what matters is your recent behavior pattern. Hence, you can get toxic players even in 9K (only if your recent behavior has been negative). Negative behavior could be cause of direct reports received from fellow players or through in game filters which monitors in game behavior.

                            Further proof was put out by Valve in their recent update.

                            ========================================================
                            MATCHMAKING BEHAVIOR

                            This update also includes several changes intended to remove undesirable behavior from the matchmaking pool.
                            ========================================================

                            And just for those curious ones out there. Try queuing with around 2000 BScore versus anything higher than 4800 BScore. Your queue time will drastically increase. You only receive LP if you exceed a certain number of reports across multiple games from multiple parties within a window of 2-5 games.

                            @Sanir,

                            Another note I would like to put out there is, if you read my initial post correctly you will see that I have mentioned that the behavior score only affects your game quality indirectly and its effect can be completely negated if a neutral behavior is maintained for as low as 5 games. However, the effect of equal chances of winning is directly enforced. Here's a quote from Valve:

                            ========================================================
                            What Makes a Good Match?

                            The ultimate goal of automated matchmaking in Dota 2 is for players to enjoy the game. The matchmaker seeks matches with the following properties (listed in no particular order):

                            The teams are balanced. (Each team has a 50% chance to win.)

                            The discrepancy in skill between the most and least skilled player in the match is minimized. This is related to team balance, but not the same thing.

                            The discrepancy between experience (measured by the number of games played) between the least experienced player and the most experienced player is minimized. More on this below.

                            The highest skill Radiant player should be close to the same skill as the highest skill Dire player.

                            Each team contains about the same number of parties. For example, the matchmaker tries to avoid matching a party of 5 against against 5 individual players.
                            Players’ language preferences contains a common language. Lack of a common language among teammates’ language preferences is strongly avoided. Lack of a common language across the whole match is also avoided, but less strongly.

                            Wait times shouldn’t be too long.
                            ========================================================

                            Ce commentaire a été édité
                            Cookie

                              Dear retarded god, where the fuck are you pulling any of this shit from?????

                              Either youre just talking out of your ass or you done some heavy research, if you did do it then give us the data.

                              If not, get the fuck out of here with your shitty ideas

                              Data that you can get off playing once on a smurf is the same as guessing, i want real data

                              "
                              "

                                Hmmmmm...

                                rüm

                                  So this means matchmaking Mobility is a myth??

                                  Smurf to the top then

                                  RayZLiENT

                                    @RIKI,

                                    Kid, please tell me exactly what "real data" are you looking for?

                                    Sia

                                      anyway valve matchmaking system actually HAS a way to try balancing things with cmd counts and reports being there. realistically there is no system that can identify people as toxic or nice. it estimates it with thr account's data. now i dont need to know what the system is and how it works since it doesnt matter if we cant change anything.

                                      if u wanna get the same kind of actual behavioural score in both teams every game let valve install an cheapset(?) on your brain and every other player that analyses player's whole personality and mood before searching a match and put players with the same mentality together in a match

                                      SOSLICK

                                        y u have that avatar?

                                        Celine

                                          Because you guys are lame fucks who follow the fashion.

                                          Cookie

                                            i want fucking numbers my dear, i want to see your research method and i want to confirm it myself.

                                            like every other scientific research, it has to be repeatable by another laboratory

                                            and it should give the same results in different conditions

                                            Ce commentaire a été édité
                                            Sia

                                              i was koala for a long time newbie beat it

                                              Jacked

                                                Lul cucki is so silly. I can confirm playing with low behav score is impossible! Team is so bad whenever I play. Other team is always so good! I'm 4K god but my team is bad and I have low behavior score! So There's your proof!

                                                Roflmao rip spurious correlation

                                                Sia

                                                  everyone is a theorist here :thinking:
                                                  im a theorist here :thinking: :thinking:

                                                  noplsnonono

                                                    can confirm, picking a few main heroes for specific roles and sticking to playing them lets you get better at it. Check my dotabuff for proof. Who would have thought people can learn. =))))))))

                                                    For the rest - no, you are just guessing how the mmr works. Behaviour score does match you with similar people. Rest is unknown. Maybe they use hidden mmr and other stats (for example: player_behavior_seq_num_last_report) and not just behaviour_score. Or maybe a bunch of highly trained and underpaid monkeys determine your mmr using tarot cards and astral projections.
                                                    Im pretty sure some people know exact algorithm, but why spoil the mystery.
                                                    But i am fairly shure that in some games you will be the lowest mmr and get 25 and in some the highest and get 24 (or lose 24)
                                                    Playing a few games a day is also a thing, everyone can get tired and start making more mistakes.

                                                    Op? Go sleep a bit please.

                                                    Ce commentaire a été édité
                                                    Sakazuki

                                                      If your teammates flame you at the very start of the game and you sense much cancerous smell, mute them and play your game.

                                                      plaku]

                                                        ahahhahahahahahahhahahahahhahahahhahaahhahahahahahahhahahahahhahahahhaha ' i want fucking numbers my dear, i want to see your research method and i want to confirm it myself.
                                                        like every other scientific research, it has to be repeatable by another laboratory
                                                        and it should give the same results in different conditions'. OMG. ahahahhahahahahahahaahhaahhahahahahahahahahahahhaahahahahahahahahahahahahhahahahahahahahahaahahahahahahah. he thinks the is fucking sicence , i rly cant believei this, ahahahhahahahahahhahahahah. thisi riki im prety sure now he is jsut a kid, or a completely retarded person, he foccused so much in dota mechanisc that must have damage to much is small brain rly, this is not even funny bu i cant hold my laugh, sry god. ahahahahahahahhaahhahahahahahahahahahhahahaha

                                                        Maximus

                                                          Bungmangler

                                                            ^^ lol
                                                            that is funny

                                                            FHigh420HighE

                                                              lul I have 1k behavior score, and have a 90% winrate and 750 avg gpm....

                                                              if this is true I cant wait till I have 10k behavior score and 100% winrate and 1k avg gpm.

                                                              TheMaverick427

                                                                From a game optimization standpoint, having multiple requirements when searching for matches is a bad thing since it will increase search times. If the game just has to search for 10 people within a certain MMR range then that wont take long to find. If they have to search for 10 people within a certain MMR range AND that fall into a certain behavioural trend then the search takes longer.

                                                                The more requirements you add to the matchmaking search, the longer the search times will be. Valve most likely wants to prioritise having short search times which is why I am always sceptical when people propose that the matchmaking system takes multiple factors into consideration when finding a match.

                                                                Jacked

                                                                  the last thing any sane person would want to do is to take the anecdotal evidence of one of the most delusional and mentally retarded posters ever to grace this forum. 43% winrate of 700 games, stuck in normal skill, and blames his team and behaviour score. dont be that guy.

                                                                  oof

                                                                    well technically this is a sort of science, you need to use scientific method to either confirm or deny this hypothesis.

                                                                    >tfw no gamer girl peeing gf

                                                                      MATCHMAKING BEHAVIOR
                                                                      This update also includes several changes intended to remove undesirable behavior from the matchmaking pool.

                                                                      This is proof that behavior score matters.

                                                                      I say it's only time before icefroggy makes this visible on profile without having to console command.

                                                                      RayZLiENT

                                                                        @RIKI,
                                                                        Kid, I have already given you all the stuff you have asked in my first post. Read it again please.

                                                                        @SemDrengr,
                                                                        Would you mind sharing a screenshot of your BScore and Queue times? And 2.1K Solo, 1.6K Party, I hope to see you get to 8K MMR in a very short time with your new 90% winrate (soon to be 100%). Also, your GPM is 486 and your winrate is 53% (went from 47 to 53 in a very short time, interesting)

                                                                        @TheMaverick427,
                                                                        What you said is absolutely logical and makes complete sense, but you also have to factor in the amount of players we have in the player pool. Like on average we have an estimated 450,000 players online with 90% of them being under 3200 MMR (Slightly outdated), while the average queue time is 150 seconds. This is an incredibly sufficient parameter for the matchmaking to fit in its criteria in prioritized order. Like, if you queue at an odd time in an odd server with an odd language selection, the matchmaking would find it extremely difficult to fulfill most of its required criteria (Mind you, you won't get a game unless it has fulfilled atleast its first criteria #1). Need proof that people abuse these ques to abuse matchmaking? Here's a quote from Valve:

                                                                        ========================================================
                                                                        REGIONAL CHANGES

                                                                        Ranked Matchmaking will no longer be available in South Africa, India, and Dubai. We will update these servers as the unranked populations in these regions grow to the critical mass needed to be able to support splitting the user base into Ranked and Unranked queues.

                                                                        While there are some players that used these servers for Ranked, the low populations made them prime targets for various Ranked Matchmaking abuses. For the time being, players in these regions will need to join Ranked Matchmaking queues in one of the other nearby supported servers.
                                                                        ========================================================

                                                                        @END,
                                                                        "scientific method to either confirm or deny this hypothesis" - Are you being serious? The only way we can know for sure if we have access to the source code and since we only have access to the output of the algorithm, we are trying to reverse engineer the results to correctly guesstimate the actual algorithm. You can only theorize based on your observations and can only apply common knowledge to avoid loops and pitfalls on the algorithm. And also if people get paid to do this, there would be were way more thorough real-time experiments to reverse engineer the code and trust me, Valve would do anything to stop that as they want their pro scene to remain scarce so they have authoritative control over their players and resources.

                                                                        Ce commentaire a été édité
                                                                        Shou

                                                                          cat ur actually braindead
                                                                          how tf can u say with certainty that u got above specifically higher amts of cancerous players at 3k BS as compared to 9k BS, how can u know for a fact that it isn't ur confirmation bias and specific memories of more traumatic events and triggering players that make it FEEL like 3k BS is more toxic than 9k. HOW CAN U KNOW FOR SURE?
                                                                          WHERE IS THE PROOF?
                                                                          THE ONLY ONE BLOWING WORDS OUT OF THEIR BUTTHOLE IS YOU, HE WHO IS MAKING STATEMENTS THAT ARE COMPLETELY IMPOSSIBLE TO PROVE.
                                                                          kys
                                                                          just a big excuse for ur inadequacies.
                                                                          as for u Mafioso ur so autistic I'm not gonna bother talking to u.
                                                                          queue Mafioso's teapot.
                                                                          can someone go photoshop some sort of image to spam abt Mafioso's teapot whenever he makes a post?

                                                                          Sia

                                                                            he meant sources are u serious lol there is no source to ur theory

                                                                            Ce commentaire a été édité
                                                                            🍩🍪Cookie🍪🍩

                                                                              you don't know the source code for the universe but you can still do complex analysis to figure out equations like e=mc^2

                                                                              for all you can know the full equation could be e=mc^2 + doggy biscuits but only way you're knowing that is if you looked somehow outside the universe, but our e=mc^2 works for a good enough estimation even if it were wrong guesstimate.

                                                                              so basically if you did any fucking scientific analysis ever in your life you'd know this.

                                                                              all your theory is a stupid guess without any data or sources to that data

                                                                              i can make any theory i like using no research.

                                                                              goes back to russell's teapot, if you can't deny me then i must be right???

                                                                              the burden on proof lies on the one who makes the claim, not the one disproving it.

                                                                              basically what i'm saying is: your idea about the matchmaking system is not to be held true unless proven true.

                                                                              if you were trying to reverse engineer and guesstimate the results, you'd show us your method and research data and we can confirm/deny them using our own analysis.

                                                                              Ce commentaire a été édité
                                                                              Sia

                                                                                it's over 9000!!

                                                                                SOSLICK

                                                                                  @END,
                                                                                  "scientific method to either confirm or deny this hypothesis" - Are you being serious? The only way we can know for sure if we have access to the source code and since we only have access to the output of the algorithm, we are trying to reverse engineer the results to correctly guesstimate the actual algorithm. You can only theorize based on your observations and can only apply common knowledge to avoid loops and pitfalls on the algorithm. And also if people get paid to do this, there would be were way more thorough real-time experiments to reverse engineer the code and trust me, Valve would do anything to stop that as they want their pro scene to remain scarce so they have authoritative control over their players and resources.

                                                                                  NO! That is not what you do to draw any sort of factual conclusion for pretty much anything in life. How you "reverse engineer" if BS has any affect on MM/whatever you are trying to prove is RECORD DATA and then analyse that data.

                                                                                  You have no data, all you have is your MEMORY of your personal experience. Where are the pages of games where you took everyones win rate and then saw if your theory in the OP was right?

                                                                                  Ce commentaire a été édité
                                                                                  RayZLiENT

                                                                                    @SOSCLICK,

                                                                                    "Factual Data" - I have indeed done my homework in coming to these conclusions that I have mentioned earlier. You can dig through my game history to draw similar conclusions. I have infact analyzed every single game and behavior score changes that have occurred. Moreover, I have even dropped my score below 2000 which gave me a 20+ mins queue time. Now talking about RWR and game impact, I have tried to max out my performances (highest game impact) and have seen the quality of my teammates drop significantly after and within a certain threshold and go right back up when I have intentionally performed poorly. Behavior Score Marker has been an incredibly important factor as that clearly dictated how the attitudes of my teammates summed up. What you are talking about can only be done if a proper scientific report has to be generated (only if we get paid to do it), however the study was done with a similar intention, but with a more informal approach. So instead of whining about how this sucks, go ahead and do a similar study and post back the results.

                                                                                    Ce commentaire a été édité
                                                                                    🍩🍪Cookie🍪🍩

                                                                                      so you're basically admitting that you're a moron?

                                                                                      you basically did no credible research that can't be re-created and just drew conclusions from a sample size of 1 based of your own experience rather than real data analysis.

                                                                                      as predicted

                                                                                      Ce commentaire a été édité
                                                                                      SOSLICK

                                                                                        Then show me the data. If you dont write it down and put it onto paper/excell then you are susceptible to all the flaws of humanity that the scientific method attempts to overcome. Selective memory, confirmation bias, etc...

                                                                                        You are also going into your theory the predisposition to prove your theory correct.

                                                                                        dunce

                                                                                          sub 70% arc winrate garbage 123

                                                                                          Jacked

                                                                                            u guys do realize the assumption these behavior score people make is that the retards end up more often than not on their own team rather than the enemy team. that itself is enough reason to debunk them as retards

                                                                                            Jacked

                                                                                              a much simpler explanation for a low behaviour score and the association with toxic behaviour, and low win rate, is that they are toxic fucks themselves. end of story

                                                                                              Jacked

                                                                                                its a spiral effect. AND if you are stuck where u are, you are more likely to create scenarios in your team where there's a lot of friction and blaming and toxicity. then of course you're gonna think something is up with your teammates.

                                                                                                Sia

                                                                                                  friction was one of the best albums out there kappa

                                                                                                  Ce commentaire a été édité
                                                                                                  RayZLiENT

                                                                                                    @Cookie,
                                                                                                    Sample size of 1, so you are also admitting you are a moron?

                                                                                                    @SOSCLICK,
                                                                                                    Ok. Add me.

                                                                                                    @Jacked,
                                                                                                    Dude you are 1.3K solo, how do you even call others retard?

                                                                                                    Sia

                                                                                                      jacked's 3.5k lul