This matchmaker is garbage

Dope Solo·May 31, 2019·3 votes·1,330 views

https://imgur.com/a/eNcDCh7

This isn't even a real game. You just log in, get 4 terrible teammates and spend the next 15-20 mins losing.

22 Comments

DuskDaUmbreonMay 31, 20192 votes

#KDA

#IS

#NOT

#MATCHMAKING

Dope SoloMay 31, 20192 votes

Didn't say it was. But the matchmaker is true garbage.

DuskDaUmbreonMay 31, 20192 votes

Then don't use a fucking picture of their KDA as justification for why matchmaking is bad

Dope SoloMay 31, 20192 votes

In case you didn't notice genius, there's a lot more in that picture than just kdas. Interesting that that is what you chose to focus on though.

Unker139May 31, 20192 votes

I'm trying to see why you think that is an example of bad matchmaking.

They had two Carries but both of them also have games that are not so great in their history.

Your jungle fed but it also looks like it might be a first time with a champion.....and normal games are where everyone wants you to learn new champions.

You also had an afk at level 4.

So 4 v 5 with a new champion as one of the 4 and a couple of enemies that had good games.

What could the matchmaker have done to prevent this from being lopsided?

Dope SoloMay 31, 20192 votes

For starters it failed the most basic job of a matchmaker - to create 2 teams of approximately equal skill.

ZephyrDrakeMay 31, 20192 votes

[{quoted}](name=Dope Solo,realm=NA,application-id=3ErqAdtq,discussion-id=A4T02FAs,comment-id=00010000,timestamp=2019-05-31T01:38:17.119+0000)

For starters it failed the most basic job of a matchmaker - to create 2 teams of approximately equal skill.

In normals? The mode in every single game where you never get matched with people of somewhat equal skill? You have shown nothing at all that says this is "bad" matchmaking

Dope SoloMay 31, 20194 votes

Yes in normals. Why not? It wasn't this bad last season. What happened?

Unker139May 31, 20195 votes

[{quoted}](name=Dope Solo,realm=NA,application-id=3ErqAdtq,discussion-id=A4T02FAs,comment-id=00010000,timestamp=2019-05-31T01:38:17.119+0000)

For starters it failed the most basic job of a matchmaker - to create 2 teams of approximately equal skill.

What makes you think they weren't. It is possible to have a lopsided game between two approximately equal teams.

Dope SoloMay 31, 20194 votes

Approximately equal teams don't feed that hard in the first 5 mins.

Unker139May 31, 20191 votes

[{quoted}](name=Dope Solo,realm=NA,application-id=3ErqAdtq,discussion-id=A4T02FAs,comment-id=0001000000010000,timestamp=2019-05-31T01:57:41.097+0000)

Approximately equal teams don't feed that hard in the first 5 mins.

When you have an afk AND someone playing a champion for the first time? How do propose the matchmaker identity and adjust for that before the game even starts?

Dope SoloMay 31, 20191 votes

The afk was just the cherry on top of a poorly made match. You'd have to watch the replay before jumping to conclusions though.

Kai GuyMay 31, 20194 votes

Again. As the guy has asked you over and over now. How was MM to know that your jungler would first time his pick. AND your AFK would afk. BEFOR lobby was made?

MMR is used to pull all 10 players into a game, how would MM know if your teams that it thinks are fair because of MMR # would have a first time pick or an afk prior to the lobby being made?

You need to address this. Its a very easy question.

Dope SoloMay 31, 20192 votes

Yeah thats bullshit and clearly misses the real issue. For starters that new jg was filled to jg by the matchmaker. Played 4 games tonight - all 4 were full of feeders with zero chance of winning the match. Not just full of feeders, but up against a team that were nearly solid. https://imgur.com/a/u0oRtSj

https://imgur.com/a/LL5NuHF

https://imgur.com/a/JPtXAaB

https://imgur.com/a/osxGi6e

It's every time I turn on the game now, most of them are like this. There are several posts about the matchmaker quality every day now. Its time for the issue to be addressed and the apologists can sit down and shut up. Again, this is a completely different feel from last season and the difference is BAD. We just went through a period where riot tried out another bright idea called positional matchmaking that failed spectacularly in case you forgot. Apparently, they aren't done ruining the game yet.

Kai GuyMay 31, 20192 votes

Positional Ques died. For the record far as i know Positional MM is still in place.

You can call players Apologists. You can insult riot you can be a dick to random users.

You have a very simple question asked again and again. How exactly is it that MM is able to predict your teammates behaviors prior to a lobby?

Dope SoloMay 31, 20192 votes

You're kidding right? Is this some kind of joke? Are you unable to understand a simple concept like correlating relative skill levels to in game choices and behaviors? Is that what you're really saying? Because you're essentially suggesting that choices made between players of different skill levels can't be predicted by a statistical algorithm, choices like warding and positioning between bronze and gold players as examples. What **else ** do you think the matchmaker would be predicting lol?

Kai GuyMay 31, 20191 votes

https://na.op.gg/summoner/userName=Dope+Solo

Your Ranked MMR is at a level where Bad players are expected to exist. every mismatched new account should be moving positively or negatively thru Silver MMR. Its not shocking or suprising to find games with shitty teammates or enemys.

Normals are much less accurate in MM.

Your still evading. Any player can pick something subpar as long as they have it unlocked on their account for any reason. There is a massive amount of players with recorded history of champions they play only 1-3 games on and never played consistently and this is across multiple MMRS not just bronze or gold.

Again. Far as i know, Positional MM is still in place.

You still have a simple question to answer. I want a blunt reply with no ambiguity. From your response what seems to be implied is riots using a statistical system off in game subjective data. I want to make this clear. Do you want to claim that Riots using a Statistical system and not MMR?

Edit. Because MMR is not a subjective system. Its probability theory and wins or losses and who your vs. That's it. So your Stats falls apart when the claim is a system that's not built off that.

Dope SoloMay 31, 20191 votes

The MMR **is **a statistical system. Obviously it gives each player a score based on their performance and an array of in-game behaviors. It is the mismatching of scores or misappropriation of scores that leads to mismatches in gameplay - which do exist, and quite frequently which is easy enough to observe. The quality of that matchmaking is subject to a performance grade, reflecting riots actual ability to create a quality match and this is the topic of this post.

Suggesting otherwise would require that random in game elements make matches impossible to predict regardless of score which is on its face ridiculous and would defeat the purpose of having a matchmaker in the first place. What is your level of education?

Kai GuyMay 31, 20191 votes

If its made around the concepts of Arpad Elo then you are incorrect. His system Is not statistical. Nowadays is mostly logistical distribution and tracking wins/losses + who played who.

Rn=Ro+K(W-WE) new rating after a game is RN. Ro was rating used prior to make the game. K is uncertainty. Where on the ladder you are and/or how many games you played in the que. W is match results. Win or loss. WE is if there are system expectations for unfair games built off the probability theory of class intervals.

NONE of this takes stats from the game. It never subjectively rates anything ever. Its not a statistical system of in game information. I own Arpad Elos book on the topic. I can literally provide you quotes from him if you need.

So, just to be shure. What Your claiming is that riots lying and does not actually use a MMR system? MMR is an Elo derivative, that's a fairly well established understanding.

Dope SoloMay 31, 20191 votes

Derivative is a very broad term. I would expect that the system you describe does exist as a component in the actual more inclusive and modern system, which is proprietary and undisclosed. Think about all of the information that is collected, displayed and recorded over the course of a match. You think it just floats away? You think positional matchmaking doesn't have variables present in that simplistic equation you reposted? This has been a fairly unstimulating chat, and I'll have to be going to bed.

Kai GuyMay 31, 20191 votes

Fair enough its a wide term But if you break the core principles its can not be treated as a derivative of Elo. That's pretty simple and fair for me to claim.

I am saying that Yes. Those stats mean nothing with out context of each match. Building context for all 140+ champions in relation to MMR in relation to offsets of skill gaps that accounts for stat padders/bad players who just spam skills?

Karthus is my go to example. If you cast his R every single time its off CD and they have 5 champions up. You 100% of the time will do more damage from that ability then karthus who just uses it when smart to help teammates or secure a kill. KDA? Your KDA drops by 50% if you go from 1 to 2 kills. Less kills are offset by less deaths. Dramatically. 3-1-2 is 5 to 1 kda. 15-8-15 is 3.75 to 1.

If its not off KDA because of that. Then Deaths Kills and Assists are valued individually but... you still reward players to be selfish and try to not die.... plus longer games produces better odds of higher impact stats if you play selfish all game. Stats as a whole are impacted by game time.

So... a subjective stat system man? Yea? Its... Tracking Positioning trends, kda, damage, objectives, vision,healing,sheilding,cc,towerdamage,minion cs, jungle cs, ETC ETC ETC ETC ETC

Every little detail. Then runs that... thru a filter of did some one misuse their abilitys to pad stats. Then run that game thru a filter of... how long was it and what champions are being played.

Vs the system that's made around Wins. Losses. Uncertainty. Assigned Rating.

This is a massive amount of work for no deceniable gain.

You Can make stats shave off some Loss or add bounes gains and long as it sticks to MMR your system works fine.

You cant Remove the impact of wins and losses by forcing a result from prior to the lobby.

You seem to make the claims that its off in game stats not objective measurements off wins+losses. And also that its trying to impact the results of a game prior to creation. And that its doing this from tracking in game statistics?

By all means don't let me misrepresent you here. I want you to provide clear opinions. I am providing my understanding of what your sound like your representing. When this is unfair call me out.

Or... feel free to bail out because your lack of knowledge on the topic at hand and blame some proprietary system that may or may not exists. Your claims current imply riots not using MMR. If I got that wrong you need to address it. I can support my stance when its about the principles of Arpad Elo used. You got anything besides dipping out on the conversation?

Edit. I don't mind waiting for a few days. No rush. Get your sleep mate. Being well rested is important in life.