Does anyone know on what basis the player selecting system - selects members to the team? Are there any rules at all, is it based on absolute randomness and the team can be assigned to anyone, regardless of experience, KDA, number of games, level, etc.?
Goddamn Heretic1 votesThe game chooses players for a team based on their "MMR" or "match making rank." It is an arbitrary number that goes up a little bit when you win a game, and goes down a bit when you lose a game. Every game you are matched with players with a similar MMR number. so for example: If you win the majority of your games against opponents with a certain MMR, then your MMR will increase until your reach an MMR where the people you play against are of a similar skill level to you, and so you will most likely have a equal balance of wins and losses against players with that MMR. Riot doesn't tell you what your exact MMR is, but it can usually be estimated based on your ranked games.
Neosphoros1 votesDo you know how this MMR is calculated? What if I will have troll or AFK in team and becouse of that I will loose a match? My MMR will be decreased? If I will have a low MMR, it will put me to players with same low MMR or with higher MMR, or it will place me to play against low MMR enemys or maybe we are (as a players) choosen to team with same/similiar MMR? if so is that mean that our enemy will have same MMR as we have?
How it is even possible to calculate anything like MMR (whatever it is) if every time players are playing with different players? If this is a true - all of this creating so many variables that is impossible to calculate anything.
Few minutes ago I have played a match with Xayah on bot lane (against MF and BLitz). She was not even favour adc as her main champs. Her win rate as a adc = 43% (33% as a Xayah), and her Xayah statistics are a much below average when I - as a Heimerbringer have 70% winrate (52% as a support) with stats a far above average... so how it is possible that player selecting system choose us to play together?
Moreover.... enemy adc winrate = 53% (as a MF 67%) with good statistics, and enemy supp winrate = 50% (interesting and weird is fact that he were played as a Blitz and it is not possible to check his stats)
With full respect but it does not look like anything work properly and fair.
Neosphoros1 votesSeriously? Nobody's know how this MMR system works? Are you trying to say that the way how the system works - is hidden? If so... do not you think this is unfair and it is suggesting it is easy to manipulate us - becouse in the same way it could also means that system may works every time in different way or it may not work at all.
Neosphoros1 voteshttps://boards.na.leagueoflegends.com/pl/c/gameplay-balance/cEBcAjEc-unbalanced-gameplay
...is that mean if players from platinum division have the same MMR as silver players? So why silver players are in silver and why platinum players are in platinum division?
By the logic... if two players playing on the same level they should be in the same place so "what" decide that one player is in different division than the otherone... is not the system? So what makas to choose who should be "higher" or "lower" if 2 players have same skills?
Neosphoros1 votesI played over 3000 games, most of them I was playing as a support. I have 51% overall winrate, with some of my favourite champions more than 70% winrate - all of them have statistics much better than people from bronze, silver, gold and even platinum.
There is only one big problem: the difference between the number of games that I'm winning and loosing is always 15 - 25 games (always I have more game that I lost than I won). How to explain it?
Last time I have increased amount of won games to just only -6 to amount of game lost... but now everything comes back to "normality" and I have loose most of games I played. I'm not trying to say that I'm not guilty at all, of course I lost few games by my own mistakes but most of those games were created with people with no experience, that thinking that Leona is only good support, that to have heal potion is obligatory, with adc's that do not know what is all about with Zileans R.
It is look a like, that system keeping me in same place for some reason. Why? It's not about my skills if my favourites champions have very good statistics so... What is all about?
Neosphoros1 votes[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=8l9xVKoY,comment-id=000100000001,timestamp=2018-11-29T19:07:17.598+0000)
- @ Matchmaker. Basically it just Grabs folks near your MMR and trys to build teams with similar mmr. Note MMR and Divisions are not the same thing. Source
For this moment it looks like everyone belives that MMR is something like an universal truth. Nobody knows what is his own MMR, nobody knows how does it work but everyone belives that is working - this is stupid - Is not it like that? There is something like this already... this is called faith, but it is nothing to do with knowledge (also not with religion) but more with, that pink unicorn - exist.
Neosphoros1 votes[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=8l9xVKoY,comment-id=00010000000100020000,timestamp=2018-11-30T22:29:24.847+0000)
That would imply that we know nothing about how Elo systems work. See we know from seasons 1 and 2 Riot uses a Modified Team Elo system, it used to show MMR. What you can do is look for the 50 percentile of player base average. At some point half the players are above or below a Tier so we can set that as a reference point for what riots starting MMR is likely to be. Pretty shure this is what most 3rd party folks do.
Information that "RIOT uses a modified team elo system" it means exactly - that it means everything and nothing in reference to it - speaking about MMR without specific details have no sense. But fact is, that we dont know nothing about this system and only RIOT force us to believe that is exist.
Riots not putting a mystery infront of us, we know the system being used and we know the results from using that system. Bad players fall, good players rise. Consistently if you take a cool at the skill gap between lower to top mmr ever year we can see this is true. Now, for why Riot hides individual mmr? 1. Abuse cases, they don't want folks looking for ways to minipulate their modifications to team elo. 2. They don't want to give competitors their fourmula and modifications for free. 3. It encourages players to play more games. The Division system is more engageing for a large # of players. Theres a reason it started popping up a few years back into pretty much every competitive ladder game.
We do not know MMR and only result we know is that "some" players/profiles are moving betwean "some" divisions... but it is difficult to say that "good players..." when - your position in the table may depend on trolls or afks in your team (in last 6 games I had 2 times AFK, and 2 times troll in team... so, just think how it would look my progresion/regresion if those game would be a ranked games).
All your reasons that RIOT do not want to reveal how the MMR works - are so good, same as they are bad. They are good only for RIOT - and bad for us (players) - becouse we can not verify how it is working - and if it is working.
Moreover your reasons may have nothing to do with the reasons that RIOT may have. Your information about their MMR system - this is only you opinion... but in the same way I could also wrote that RIOT do not want to tell us how MMR works because it is not works fair for every player, and maybe it is working for provide most profits.
You have to be careful when discussing weather or not we can take things on faith, if you question everything and imply that you cant trust some ones words theres no grounds to communicate. For example Is there any way you can prove Neosphoros That we are not the same person using two accounts to have a conversation with himself/herself? If we discredit everything you reply with as dishonest then no, how ever it would be stupid to do so because its honestly a very unlikely situation.
I'm questioning only things that may have effect on me. If i see the things that works in different way that they should - I'm start thinking - "why?".
I can not prove that "you" are not "me" but it would means that I'm very smart guy ;). In other way - denying to myself would be like a triumph of form over content. I am missing out on the fact that it would be a huge waste of time considering that I am very lazy.
Is there any reason to assume riots not using something at its core built around the Elo system? Yes or no? If its Elo we can check it by its results, the results are what we expect from an elo system. Do you have any disagreement on this? If so please list what competitive ladder modle you think riots using if its not built around elo.
Of course that we can. Elo may works only in case of 5-players (the same players) flex games. If there is - solo/duo when the rest of team is choosen randomly (in the meaning of - every time - different player) - it can not be - ELO system. Amount of variables (player position, in-game performing, AFK, trolls, etc.) does not allow to correctly estimate specyfic indicator.
Teoretycly there is a chance to create good system but ELO sytem has nothing to do with it. Moreover I'm able to say that from my observation RIOT system have many different other assumptions.
Neosphoros1 votes[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=8l9xVKoY,comment-id=000100000001000200000000000000000000,timestamp=2018-12-07T03:49:51.069+0000)
Heres what you need to understand.
- Devs pick the starting point, the K factor, the Gains. Folks using Bronze to diamond style ladders pick what percentile of players they want each division to represent. Each system uses its own variables, all picked by the dev. Modifications and adjustments can be made at the devs whims. Its not a universal system, its not like the metric. The way it works is all around variables that are completely arbitrary, so long as they follow some basic principles. Example, You play 2 different video games. you have the same skill in them and the same win rates. In Game A your MMR is 1200 and your a silver player. A new win will get you 300 MMR. In game B your mmr is 1800 and your a bronze player. A new win will get you 80 MMR.. In both games you in 35% of the top because they also magically have the exact same # of players. (Ranked better then 65% of the games playerbase.) The numbers alone are not important because they don't mean anything w out the rest of the system.
So... it simply means that system not working fair for each of us because a system factors may be not connected with players skills at all, but with some elements that are not included to gameplay.
- I mentioned it in my last response. In team games you have to mod it at the very least to account for weight. the players the system things will carry more weight have a bigger stake in the outcome of a match. This also needs to be in relation to the K factor, Each player needs their individual K factor sperate from the team. The more uncertain accounts need to get up or down faster then the folks with high # of games. Those are the players who bring the most negative impact to a matches balance.
So question is what is the "K" factor?
Does not matter what - it is difficult to say that system working fine... at least, fine for me (fine - in the meaning of "fair" - fair in the meaning that playing with created in specific way (by system) team I will have the same chances to win as enemy team do... last game I lost 5:36 and becouse of that have the chance to be promoted in the table with the proper speed or be promoted at all).
- This information is old and may be outdated, previously in a FAQ riot listed their expectations for their Matchmaker to be 150 to 300 games for you to hit your "true Matchmaking Rating". Now in A 5 man basic Elo/MMR that's just modified for K factor and weight like I talked about, its going to be in the range of 400 to 600. A bad player that really hurts their team? that player is bad in every single game they are in. You see them once, you always occupied 20% of your team slots. Odds favor you as a player that the enemy team will have Baddies that lose more then your team because your team has you. (typically folks argue HA THEN THE ENEMY TEAM HAS BETTER ODDS FOR THE BEST PLAYER IN THE GAME! At which point...they are admitting skill matters and they don't consider themselves good enough to consistently be the best player in a match. This is where I kindly thank them for making my point for me. End of the day, skill matters in a skill game and law of # favors good players in MMR systems) If you don't have the ability to commit the time to play a high # of games then you don't have the time to counter RNG. If a player said matchmakers unfair because he only played 5 games all season, Face palm.
It suggest that team can be build to just "fill" difference between my high skills (or experience if you like), rest of my team and whole enemy team like... if for example (hidden) overall team MMR (as experience) is 2000 it means that enemy team should have exactly the same to compensate end score - 2000:2000. Fallowing... easiest way for system will be to build a team with members with 400MMR point each (in every team)... but what if one of those players have for example MMR=1600 (teretycly) ? Rest of the team members will have like 4x100? But...
What if in specific moment of time there will be no "equivalent" enemy player to compensate a difference between most experienced red team player and blue team player?
What if even both teams will have overall MMR (ex) = 2000, but one will be build like:
a=1600 b,c,d,e=100
and the enemy team will be:
f,g,h,i,j=400
sometimes one player is not enough
- Team MMR systems have been tested and found to be workable. They have pros and cons. Some of their cons for players (need large # of games) are pros for a company. The pros for players are it enables matchmaker to be more competitive. Bad players sink over time good ones rises up. They over time tend to get pretty accurate in their measurements of a player. They are superior to systems that 100% RNG build teammates because that does not make for fair teams. They are superior to systems that only Track wins because skill is a 0 factor when all that's required to top the leaderboard is having the most time to play the game. A con is that luck tends to be a factor when your near your skill range cuz More impact from teamamtes, the pro to that is if you increase your skill you start to see yourself climb faster. Keeping games in a range of acceptable MMR helps keep players from experiences gross mismatches in skill. Some folks cheat this by using smurfs/buying accounts. But the system will place those guys where they belong over time as well.
"Workable" in business process can mean absolute something different than player may think and have nothing to do with player skills but only with company profits.
Look, There is no easy way to really judge some ones skill 100% accurately in advanced of them playing games. So we cant have a 100% accurate system. I like MMR because I find that systems built to be self correcting and rewards skill are good for a competitive ladder. I was around when Mobas did not have good matchmakeing. I used to play a lot of Dota allstars back in wc3 days. There was 0 matchmaking outside of 3rd party programs that used elo as their base. It was a feature often included in a decent banlist.
...but if the team loosing after 10 min 1:20 it is obvius that that something is not right
...and if "I can" increase my win-ratio from 51% to 75% in just only one day, but after that loose severall games in a row that will decrease win-ratio to just only 27%
...and if will you connect that with AKFs, troll
,,,it simply suggest you that there is nothing like "equal" chances for every player, but the system trying to keep you in the same place as you were last time
Try to imagine that about 2 months ago (when I was less bussy) I was moving from from 1st position in s4 to last position in s5... please do not try to say it is normal.
I am not claiming I know how riots MMR system works. I am saying that there's enough information about MMR systems that you can check Riots statements vs Basic expectations from MMR. You can look at the result at the end if it matches expectation. Its like Solve For X, 7+ x = 10
RIOT statement is like:
"There is MMR system" :D
I'm saying that some of the things your asking for have no value and require a complete reveal of the entire system, and that's a unrealistic expectation for a company.
Maybe, but it means that we can not believe that system working fine (fair) and only thx to our hard work (efford, skills, experience) we may achieve our target
Neosphoros1 votes[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=8l9xVKoY,comment-id=000100000001000200000000000000000000,timestamp=2018-12-07T03:49:51.069+0000)
In games that claim to use MMR systems, can you or can you not see a gap in skill between the bottom percentile of players vs the top? its a very simple yes or no and I will hold you to answer this.
LOL is the only game I'm playing. Can you explain please what do you mean: "gap in skill between the bottom percentile of players vs the top?"
I told you to give the system you think they use if its not Elo/MMR. I will ask you again to please do so.
ELO system is specified and build for chess - so it is even stupid to trying to use it for different games moreover if chess is a 1vs1 type game and LOL is 5vs5.
MMR as a MatchMaking Reating it could be a general name for any system so it simply possible to say that ELO is MMR but system build in LOL (if it exist) should be something absolutely different than ELO - it contain much more variables than ELO which using this system for other (LOL) games would be something irrational
Next question. What company's that use MMR/ELO/Trueskill Publicly share their full algorithms? (I can save you some googling and let you know that Dota2, hots,leauge, and smite do not.)
I do not know which company do... like I wrote before - I'm playing only in LOL.
True is that ELO is described and known and is possible to check his formula. If it was not possible, it could be considered a form of manipulation.
It is not imposible to create "any" MMR system. It is easy to create one but it is difficult to create a good one. Every company can have differentone but it does not mean it working fine.
I truly want you to respond to those.
Look, i am not trying to be a dick , im not trying to belittle you, I don't have insider information on riots system. Im spending the time to post because of one simple thing. You do not seem to understand how team MMR systems work. my friend I don't mean any disrespect by this.
Maybe - I do not. Explain me please.
- You think knowing your MMR number would show the system works or is working well. that watching it go up or down is some how more fair or accurate then seeing yourself climb ladder up or down. (to a small extend your not entirely wrong, leagues do sometimes have folks misplaced and misrted. But it does not negatively interfere with how matches are built so at the worst some players need to play some extra games and other will end with a higher rank then actual represents them if they quit playing ranked.)
Exactly. Only by knowing your MMR number, and formula you can verify if it works properly or not.
- You made the admittedly common mistake of adjusting a 1v1 elo to be team elo with improper compensation for team play. You just subbed Player for Team in the adjustment math.
Explain me please how system will behave if I will have a AFK, troll in team? What will happened with my MMR?
- You think individual games matter to the system.
Should be.
4.Your ignoring information the devs provide and trying to figure it out from scratch. Rather then check their words vs reality to see if there's inconsistency's, your assuming as the potential exists for falsehood we need to discount all information provided as if its untrue.
Yes. In other way player can be punished for whole team - that he even do not know.
- Your acting like knowledge of systems like this is not available to be studied. That we can only use personal observation of league to figure out if its an mmr system. The theory behind team MMR systems has been tested and math used in the tests is publicly available.
No it is not. Information about (LOL) MMR formula is hidden.
Neosphoros1 votes[{quoted}](name=Kai Guy,realm=NA,application-id=3ErqAdtq,discussion-id=8l9xVKoY,comment-id=0001000000010002000000000000,timestamp=2018-12-05T09:18:58.334+0000)
I do not want you take this personally, but you are not showing a clear understanding of how MMR systems work. im going to pull a few quotes from you.
I know how "systems" works... I do not know how LOL MMR system work (this specific system).
general assumption of the (traditional) MMR system is to evaluate specific player with his game 1vs1 and it based on ELO that comes from chess system. System ELO does exist also in case of team games but there whole team can be assesed as a whole element not as a single players - which is mean that in the case of randomly slected players - it make no sense at all.
How to assess a player for example: if he lost a game becouse his teammate was AFK. if MMR depends from won and lost games (as you said) or loosing a game?
First knowing your MMR is irrelevant largely. The actual MMR numbers don't matter too much because devs get to pick what a starting MMR is, what the K factor is. when and where the K factor changes. Plus the modifications I keep mentioning are also a factor.
If it is about system that will have to assess us to place us in right division - everythings matter.
Try to imagine - Few minutes ago I have lost a game only because my team was feeding. in 15 minutes we were loosing ~20 : 1. TOP have lost his lane in 10 minutes (with whole game score 1:10), our jungler did not know how to play at all loosing game on 2:10... and it is because TOP have lost, jg have lost, mid lost - enemy was able to attack my lane (bot).
How this system working if in 10 minutes "equal" players represent absolutly different experience level?
It is because my teammates brake the game... I will have to assessed by system in negative way because I lost game?
Funniest thing is that even I had not receive any honor point after this game - my honor level went up... for what? is this some form of apology for "unsuitable" allies...
But good players or bad players moving up and down, why is that not something we can check? MMR systems have a starting point that is the same for every account ( ish… to be fair devs can totally change this number if they desire because they are the devs. If they care about competitive integrity they wont do it mid season however. Doing the starting number as RNG hurts the system and players so there is no justifiable reason for that) The system takes points from a player on losses and adds points on wins. If the modifications don't introduces extra points then we know that the starting MMR score is going to represent roughtly 50 Percentile marker of a games ranked ladder. Around that point we will see half the players are going to be above half the players will be below.
You writing your comments by trying to create some paradigmes. Nobody knows how it works - but you do. Even RIOT did not explain how MMR exactly works. Moreover... your comments may suggest that you are someone more than just a normal player. You are not even trying to undermine anything that you have learned. With full of your strenght you trying to defend whole RIOT/LOL rights - which is weird for "normal player". If your knowledge is the right one - it will be the same as confession of RIOT "worker"... and if you are - you are answers can not be objective in other way you are know not more than me, and your opinion can be wrong.
Neosphoros1 votesTry to put those player against themselfs:
--------------------------------------top--------------------mid--------------------------------jg---------------------------adc--------------------sup allies--------------------------1/9/1;----------------4/11/2;-----------------------5/4/5;---------------------4/9/2;---------------3/5/4 enemies-------------------12/3/9;------------11/6/12;--------------------17/5/12;-------------------6/3/4;---------------2/0/10
I think that is not difficult to say that there is something wrong with this system (MMR) if based is on player win-ratio (which have to also means - player skills and experience)