don’t read any of those “conspiracy theory” posts
this is just straight new account sr info
this is also why if you look at the forums
there used to be tons of posts of my main acct was hardstuck
etc
and they buy alt
level it and place considerably higher than original account
because they “learned” the game on that account and probably que’d for comp before they should have
tons of reasons for this type of placeing
so a bronze did hear about it from cliff terio’s video and tried it
lol
nice
and you wonder why players even take anything remotely semi-seriously anymore
literally it feels like teammates are enemies on ladder
hardstuck doesn’t mean bad matchmaking system
it won’t stop your climb, only slow it down
here is a scenario that shows the weakness of ranking systems
you are 1850sr. ten thousand GM players decide to throw until they get to 1750 and win when they get to 1950. and they keep doing this over and over and over. Now here’s you, playing normally…your SR will fluctuate violently, but not when compared to the other 1850s just like yourself
Elo, Glicko are relative metrics. You need to see a list of all players and their ranks, and trends of ranks to 100% accurately find out what your true place in the ladder is. Short of that, 1850sr plus or minus 200 is accurate enough.
if you think you are being shafted, add 150sr to your current score, and that’s the best you can hope for.
bearing the burden of people who get to 2350 DR by playing 15 games?
no, your first competitive game you have an SR of 2350 (I think/pretty sure)
so those players that the system has little confidence in should be placed with other players who the system doesn’t know that much about.
I play a lot, and I don’t mind sorting out new players every once in a while. But the match maker should do it’s best to consider SR certainty when making teams and matches.
i’m pretty sure it’s 2500 but that’s not really a big difference
i just played on a new account and got 103 sr for a win, new accounts get pretty fast in the proximate Sr they belong to anyways
unless someone is dead bronze or gm 20 games in they are probably max 200 SR off their real skill rating (assuming they play all games at their real skill level and don’t just mess around in low elos or sth)
awesome, now can we address the endless stream of new accounts that well established 2500 players have to filter through?
so either keep inexperience players segregated, or start everyone with zero SR 
actually starting people right in the middle is the least inaccurate you can start them so 2500 is fine
I totally agree. But then keep them away from well established players.
usually lower than lvl 100 players get more likely matched with <100 lvl players but i noticed that you get “normal” matches once you win a few games in a row
and honestly the difference between lvl 120 and 8439 is probably 0
correct, I get about 75 wins per season, every season. Recently I have been seeing A LOT of <100 in my games. I hope they cut this out in the next month or so.
This is hands down a stupid idea. Segregation of players on anything other than skill level makes it impossible to actually compare them skill wise. It breaks the math. If new accounts are only playing new accounts then they will be sorted on a separate curve than the regular player base and their rank will be innacurate.
The matchmakers has NEVER sorted on player rank or amount of games in ranked as it breaks the math. Then ONLY sorting of new players is in quickplay for sub level 25 and the only reason to do it is because new players get angry when they see higher level players as it feels unfair even if that gold border might be a legit silver. Even then anyone even plat or high gold who’s leveled a second account knows that the wait times will skyrocket if you are even average at the game on a new account and it will eventually put you with regular accounts because it can’t find enough low level accounts that are at your skill level.
Ranked matchmaking is a hard thing to do, and for it to work various conditions need to be met that limit what you can do without skewing data. Any separate player pool/population will not be directly comparable unless we can assume the skill level distribution will be roughly the same. We can make this assumption for regions and it will hold true for the vast majority of the player base with only top ranks potentially having skew if there are significant differences in the top players. We CANNOT assume a population divided on play time is in ANY way representative or along the same skill curve and so we CANNOT have a separate ranked pool without skewing the data.
no, they will eventually be filtered in as they play more games. I’m not saying a level 25 should never play a level 300, I’m just saying there should be some checks and balances in there to separate them a little.
certainty of rank is always calculated when using Glicko etc… it’s one of it’s great features.
if i’m 1800 and I play a ton, my real rank is 1700-1900. if I’m 1800 and have only played 10 games, my real rank is 800-2800
certainty and rank are one and the same
I understand how the certainty works. But problem is you have a different curve between low level accounts and the full population. This will mean a separate low level rank mm will end up with those players being sorted only amongst themselves and they will not be accurately ranked. Once they enter the general population they will be placed too high or too low which will cause more games before they reach an accurate rating. This also has potential to kill new player experience as they will be in a separate pool with just new players and alt accounts which will have a huge skill gap near the middle. Not only that if new players end up getting boosted depending how the curve works out with them only playing each other, then once they are added to the full pop they will drop like rocks which will reduce retention.
There is NO good reason to separate low level accounts in ranked. It reduces accuracy, increase time to find true rank, will place new players with smurfs, and will skyrocket queue times for players during a key time for player retention.
I already said, they will start to be integrated once they play more games. Why should I have less accurate games at their expense?
I say a new player’s MMR is wild enough that they can duke it out amongst themselves in the beginning without too much being lost?
And besides, if you’re hardstuck 1800, you think a new account should let you play with established 2300s? no good imho
it’s not separate, they are mixed in. A level 25 can be matched with a level 300, but only if the queue is really long. I’m not asking for strict segregation, only a soft one.
if you’re hardstuck 1800, you think a new account should let you play with established 2300s? this is the reason
not for the established 2300s
they can handle an extra minute, I don’t think it will increase queue times by more than 10%. which is 30 seconds for a 5 minute queue
New players are usualy willing to cooperate so tell them what to do, ask which heroes they can play and build comp around their strongest hero.
the problem is that I don’t know what rank they are, not if they are good or bad
A level 25 in their placements on their first account is almost certainly not playing at 2300 SR. This is frustrating for the people on their team, but keep in mind since you yourself are not a fresh player, you have only a 5/11 chance that a given one is on your side. You should see more opponents of that caliber than teammates.
Start everyone in Bronze and let them work their way up. I think that’d solve a lot.
yes, this is the situation I’m in. but what about the people at 2600 who have to filter out all the GMs that made a new account?
it’s not so much about good or bad, it’s more about who has to bear the burden of ranking new players? It shouldn’t be the best obviously, and I think the worst players should bear that burden before the midranked players have to.
starting people in the middle is better for the whole system…unless you reside in the middle lol