True Garbage ® Matchmaking

Hordes66·Dec 30, 2019·3 votes·2,248 views

Look riot, the matchmaker is kind of worthless right now and has been for awhile. Something really should be done to address the number of one-way stomps, which is probably close to half of all games. Of the remaining half, half of those will be made up of 2 carries with 8 people doing almost nothing, and the remaining 25% of games might actually be fair and worth playing. This is a pretty low quality game right now. I hope you people come back from your holiday and you wake up to the mess you've made.

17 Comments

Darkdemon653Dec 30, 20193 votes

How is the system supposed to know how people are going to perform in a given game?

Hordes66Dec 30, 20192 votes

I think a better question is how did the system design a match where all 5 members of one team completely and overwhelmingly underperformed all 5 members of the other team. **That **is just statistically improbable and reeks of incompetence. Lets just call this a type 1 match (a match where 1 team is completely blown up in a short amount of time, giving at least the perception that there was no chance of winning at all). - How many type 1 matches are there out of say, 20? If we were going to have real talk about this, we would then move into **causes **of the type 1 match and maybe move into **methods **to reduce the number of them.

The details of 'how' could entail a very deep conversation about all of the tracked variables that go into a winning a match and an individuals performance in those categories. There is certainly more than 1 way to design a match and I would posit that riot has just clearly chosen the wrong one. Now if you'd like to talk about details of the various ways to go about doing that, fine. If you're just a troll, move on.

Kai GuyDec 30, 20192 votes

I think a better question is how did the system design

What Horde is saying here Dark is that they are incapable of answering your question and would you please look at something else.

That is just statistically improbable and reeks of incompetence.

Why people act like Mobas are not inherently snowbally is beyond me. Do they not play this game?

Still even then with a Fair on paper example of Dice being used for a Range you can see large gaps in results between groups of 5. Roll some D10 or D20. Worse cases are 5 to 50 and 5 to 100. Elo and MMR ratings are made to represent Mean averages.

Oh well, Folks always like having an Excuse on why they cant meet expectations. Fuck accountability right?

https://us.forums.blizzard.com/en/heroes/t/this-ranked-system-is-attrocious/29000 https://answers.ea.com/t5/General-Discussion/Ranked-Sucks/td-p/8586513 https://steamcommunity.com/app/570/discussions/0/2632849569459976966/ https://steamcommunity.com/app/570/discussions/0/2632850028527452804/ https://www.halowaypoint.com/en-us/forums/58b8518e005f432381ab99fbcaf931e0/topics/what-happened-to-matchmaking/8ad6e91d-c342-40ec-9e44-fbc0fa1e7527/posts https://steamcommunity.com/app/252950/discussions/0/1736633665865938158/ http://forums.smitegame.com/showthread.php?180117-Ranked-mode-is-a-joke http://forums.smitegame.com/showthread.php?179050-what-a-joke-competitve

Hordes66Dec 30, 20191 votes

You must have missed this part:

Its baffling how anyone can look at a match like the one above and think that this is 'ok'. What's even more baffling is that you can look at a match like that and the first question isn't "how often does this type of match occur?", because that is the only appropriate question in this scenario. If it was 1/100, it could be forgotten about pretty easily. If its 10/20 - well that's a problem.

Kai GuyDec 30, 20192 votes

Note that I provided a large list of multiple games and people whining about the devs MM? Are they all A OK and its Just Riot that goofed? What about the ones that Run Performance adjustments? HOTS has that feature, are the folks whining about MM in HOTS full of shit or did blizzard goof something up?

The argument about stats is they change in relation to skill gaps, but how do you remove noise if you dont check a Win or Loss? It provably takes less effort to Stat pad then to win.

Exactly how do you populate your stats? By individual champions? OK. So then how do you use that For MM when champion are picked AFTER a lobby is made? By arbitrary metrics? Hows that gonna help if some one who plays Soraka if she is compared in Damage to Qiyana ? By Population built statistical Metrics after the game? Hows that usable for MM? Also whys that even sound remotely good given that popularity would throw huge impacts on what individual champions stats are.

The argument "OMG RITO SO DUMB WHY NO USE STATS COLLECTED!" Falls apart when you look at implementation.

Middle of the curve has high chaos but that's just the way it goes. You want to run a 24-7 que a requirement for games to be played at any time people want? There are trade offs that get made.

Its baffling how anyone can look at a match like the one above and think that this is 'ok'. What's even more baffling is that you can look at a match like that and the first question isn't "how often does this type of match occur?", because that is the only appropriate question in this scenario.

First question I ask is WHY did this type of match occur. I then Check for uncertainty, for any mismatched players and Duo impact.

Again, if gameplay is inherently snowball then frequency of snowballed games is expected. When has a Moba not been prone to snowball. Allstars, HOTS, HonE, dota2, smite?

Hordes66Dec 30, 20191 votes

Your approach is wrong. The 'why' of individual data points is open to interpretation. Only over a large number of games can trends emerge. The 'why' may be due to short queue times, which is another known variable effecting quality of matchmaking. There is no real talk about matchmaking until you discuss the individual variables that go into making a match and their weight on a statistical phenomenon.

Asides from the fact that you're trolling, if this wasn't a matchmaking phenomenon there would be scores of people analyzing this game right now saying "how do i repeat this so my entire team outclasses the enemy this hard in order to win more games"? The fact is that it isn't possible for any one player or duo to repeat this phenomenon reliably because the matchmaker is completely out of your control. If it was this would have rapidly become a thread about strategies, something these low-quality discussion boards lack in general.

Kai GuyDec 30, 20191 votes

I am saying that snowball is normal in mobas. I am saying that when you use a Range, there are lopsided events even in "fair" individual results. you can see Team A as 5 to Team B's 100. I expect more slanted results in a Moba then in A dice roll because negative performances Drag teammates down and prop enemies up, hence why I claim Snowball is normal in mobas.

The 'why' of individual data points is open to interpretation. Only over a large number of games can trends emerge. Yea? Your presenting a single data point in your opening thread. Just 1 game. Yet, Your recent Kass win certainly is not the norm in your history and when you look over others in the game there is nothing that as a general trend, stands out besides that 1 data point your highlighting as a MM issue or flaw. You know what you do find? Some instances where Riots Duoi-ng implementation is an issue on math quality. You call them 1 sided stomps and blame MM but if **gameplay **is stompy then MM will never fix that. So, WHY things happen is very important to establish.

Questions get asked. You redirect. Dark asked you a question, You told him its the wrong question. Not a reply. I pointed out several issues. You did not reply to them once, just "Nah dudes look at something else".

I want direct reply sand it should not be difficult for you to provide.

How can you use champion stats for MM when champion selection happens post lobby creation?

How can you use universal role stats when champions individual kits have high variance in stat generation ability?

What makes Riots MM special and worth complaining abut over the provided list of other team games complaint threads about MM?

and very simple Yes or no for the last one. Are mobas as a game Prone to Snowball, regardless of developer?

You have to clearly express your basic assumptions if you want to make an argument. Example. I am saying that some Snowball is normal rather then an exception so it should statistically be common.

Hordes66Dec 30, 20191 votes

Snowballs get started from bad matchmaking in the first place. But never mind that, not every game snowballs this hard, so again you're just muddying the waters. How often does a snowballed game look like this, to what degree and what variables increase or decrease the degree of that snowball is what a real conversation looks like.

You and the rest of the boards 'nah' squad are the real impediment to actual discussions on the gameplay boards. There are average variables that go into a total player score that exist outside of champion selection, even with that in mind, players are more likely to use certain champs if they've used them before.

Kai GuyDec 31, 20191 votes

There are average variables that go into a total player score that exist outside of champion selection,

Name them. Please go for it. What stats are better metrics of skill for positive impact on winning a game that are better then actually winning the game?

How do you use then for MM? List the steps. You claim it's doable the show it.

Nothing in Elo or basic MMR cares about gameplay experience. There is no snowball metric.

So you have to answer a basic question. Is gameplay in a moba prone to snowball or not? If gameplay is snowbally in nature MM never fixes that cuz that is Just the gameplay.

Hordes66Dec 31, 20191 votes

You only need look at the tracked variables to get started. Ward score, roam dominance, cs... etc. All of these and more are closely correlated to individual skill level. Applying a variable score as a coefficient to any mmr equation is child's play, child. Furthermore, having played quite a few of these games myself it's clear that team members are frequently not grouped according to similar skill levels, based on just these mentioned categories. Do they make an effort? Probably. Could they do better? Most fxxxing definitely.

I reject outright your rather silly premise that these games are inherently likely to snowball, as the early game trajectory is just as malleable by factors such as turret playing gold. Again, youre just a member of the boards reflexive nah squad'. These factors are so controllable in fact that the average game length had substantially decreased due to meddling.

Kai GuyDec 31, 20191 votes

Ward score, roam dominance, cs... etc. All of these and more are closely correlated to individual skill level

Only, Not every Ward placed is relevant to a game so that stat is really easy to pad. CS at the cost of map pressure is bad And not every champion will want to roam at the same frequency due to play patterns. So... Hows this more significant then Winning games consistanty? If you say its done champion to champion then its impossible to use for MM as champion selection, as I keep mentioning, happens After a lobby is made.

With out usable metrics for MM your still gonna need to run a MMR system which is at its core a mathematic concept that does not give a shit about gameplay experience. Seems like your aware of this becuase

Applying a variable score as a coefficient to any mmr equation is child's play

Ok. so then Why do you see IDENTICAL QQ complaints about MM for games that do this? Why is there the same low quality "fix your shit MM! Stop giving me ass teammates! Ranked Sucks. MM IS GARBAGE" for games with Statistic Rating adjustments or Performance Rating adjustments?

I already mentioned it, there are Devs other then Riot who do this. They get the exact same fucking thread after thread of players whinning about teammates that and they can often be copy and pasted between fourms and boards with no differences in content.

I reject outright your rather silly premise that these games are inherently likely to snowball,

For me I view it as just a core part of the genera and the logical view on gameplay. This is built, as I mentioned, from playing numerous Moba games.

Fine. Q and A it. In league are the comeback mechanics so strong they outweigh the benefits of a lead?
When you get ahead vs some one of Equal or lesser skill it does not give you an edge? You don't think the gains and gaps generated for gold, Exp, and Map pressure gained from consistently being ahead of your counterpart are an advantage towards winning the game?

I don't see how you can look at The power spikes come from Level ups (ability improvement and stat gains), Items (Gold), and Objective Buffs (stats and map pressure) and Losing towers makes the map more dangerous to the team who lost the tower. So its weird for me to see folks behave as if the momentum is even for both teams.

If comeback mechanics are not strong enough to compensate and the majority of players are equal skill. why the hell would the team in the lead throw it so badly they allow the enemy to recover?

What in the gameplay makes you think building momentum will not snowball matches at "equal skill"?

Hordes66Dec 31, 20191 votes

I'll say that your rather limited and useless critique of 'ward score' could have been differentiated between minutes of vision granted, placement and whether or not that ward spotted anyone. In addition, there are hotspots on the map that are known to be more valuable. All of these are independently trackable and contribute a certain value to the overall score of a player's understanding, among others. You didn't expand on any of that in typical 'nah squad' fashion. There simply isn't enough time in the day to refute all of your illogical fallacies, and trollish misstatements of fact. Nor for your cherry picking of irrelevant data from other games that, while they may have their own matchmaking complaints, does absolutely nothing that could be considered a systematic evaluation of their claims.

Kai GuyDec 31, 20191 votes

Vision is an Easy stat to pad. Do you think vision is more significant then winning games?

"'ward score' calould have been differentiated between minutes of vision granted, placement and whether or not that ward spotted anyone. In addition, there are hotspots on the map that are known to be more valuable"

Safe wards add time. Spotting folks is gonna be RNG human behavior because if you don't get ganked there is no one to spot. Hot spots change with game state.

With out context of individual game how do you sort noise?

With out subjective grading then the above stats are very easy to pad with low significance on game outcomes. And good play can be punished because arbitrary benchmark values don't match requirements for the individual match.

Your ignoring once more, OTHER DEVS ALREADY DO STAT ADJUSTMENTS MMR AND GET THE EXACT SAME RAGE THREADS ABOUT MM.

Stat adjustment on MMR systems is done and you see the exact same posts like yours. "Garbage ® Matchmaking" Is a common complaint in team games. What makes league special that it would change?

Not a single comment about the inherent nature of the gameplay eh?

What in the gameplay makes you think building momentum will not snowball matches at "equal skill"?

This is a direct question I'd like a direct answer on.

NaymlissDec 30, 20193 votes

Everyone was a similar rank in that game, what was garbage about it?

Hordes66Dec 30, 20192 votes

If the matchmaker is broken, what makes you think ranks are appropriately assigned?

GreenlightxDec 31, 20191 votes

the cs average in the game was 4.6, this tells me that the teams were more or less close to each other, kills get you out of bronze, farms get you out of silver, idk what gets you out of gold still figuring that one lol

Hordes66Dec 31, 20191 votes

At least you're looking at relevant information. Can't say that for the rest of the 'nah squad'.