Matchmaking system issues : the theory that explains it all?

Give a grade to players for their “skill”, from 0 to 10. It’s a secret grade so Blizzard won’t let you see it… :shushing_face:

On a good day :sunny:, your team has 5 players with the same skill level. GG.
On a bad day :cloud_with_lightning_and_rain:, your team has 5 players with very uneven skill levels… BG ?

How could it be ? :thinking:

A party of 5 players with the same grade (5+5+5+5+5=25) may be matched against a party with uneven grades (10+6+3+3+3=25). The grade of the parties will match (25), therefore the Matchmaking System (MMS) will put them against each other in the same game, but the grade of the players won’t match at all… This could explain everything about the confusing, frustrating, bad games you experiment.

But also, if this was to be true, it would be such a shame from the devs to have allowed such an unbalanced kind of MMS…

And DO NOT confuse this “skill” grade with the bronze to master ranking, it’s something else entirely.

I’m pretty sure it’s close to reality. What do you think ?

This is how the game feels to me a lot of the time. It’s a decent theory but the variables are very uncontrolled. Like some players may have higher grades on different roles but you can’t know what role they will be playing except in QM. How would the system attempt to MM players with these variables? Is the score averaged across roles?

I mean have you seen other Blizzard games? Overwatch has a terrible system and that’s supposed to be performance based. Overwatch has fewer variables than HotS too.

It’s a good theory and even if it’s not close this is what the MM feels like a lot of the time.

1 Like

I thought you quitted Hots or did you change your mind.

Who decides this “skill rating”?

The company already worked on it for 4 years and couldn’t come up with a good system (performance based rating).

I think you are correct. It doesn’t really matter how we call it, MMR or skillrating, it’s basically the same thing: a single number that should represent someones “skill”.
To have an even match between two teams, both teams need to have the same average MMR. That absolutely means there can be a big gap in skill between the players - just the overall team rating/mmr is the same.
Especially in QM and Unranked Draft this is the case because there are no restrictions. Ranked Draft should be a lot more even because there is a rule that people can’t play together when their rank difference is too big.

Personally i don’t need perfect matchmaking to enjoy a game. Perfect matchmaking doesn’t really exist anyway. QM needs to be this way because otherwise we could not play with all of our friends.
If you prefer tighter matchmaking then you should try ranked mode.

1 Like

I have quit Hots, the very day I said I would. It feels better this way. This is also my very last post.

I think its possible, but how would you explain one team working together like clockwork and another being a mess?

Now lets make the rating bigger to showcase variaty, so it’s not 1-10 but 1-4000.
Let’s say that small differencies doesn’t mean much skill/knowledge-wise on the bigger scale so ±200 is basically the same.

Now imagine someone (for example a rating of 2500) wants to play. How many 2300-2700 ppl are in the queue at the same time? How long should we make the player wait? What happens if he wants to play with their friend(s) that have different rating?

This rating is obviously constantly changing, so you might have higher/lower rates in the moment than usually.
You can also have days when you perform above/below your estimated performance.

You also have 89 subjects that influence these ratings/grades, that you can choose from but treats you with all on your average rating in mind.
Plus there are some random factors that can give you (dis)advantages (map, allied and enemy comps).

And with these tons of variables (with the decreased population) imagine getting matches that won’t provide “bad mm” complains while also maintaining to get ppl quick games, because ppl usually don’t like to wait.

For all the topics that crop up over and over again it doesn’t take some magical super theory to “figure out” the match making. Here’s the click-bait secret:

Garbage in = Garbage out.

Matching is ‘bad’ because the players are bad.

  • Players don’t have map awareness (to make informed plays based on what the enemy team can be seen doing)
  • The wrong roles facecheck bushes
  • Players spend a lot of time chasing or doing the back-and-forth waiting for something to happen (reactive rather than proactive)
  • They have divergent tactical preference and
  • Don’t communicate their intent, they just expect everyone else to be psychic

There’s a bit more to the list, but that’s already longer than people are willing to read, which is part of the awareness issue.

So long as people have something to blame (such as “matching”) then they overlook reflecting on their own flaws, realizing the communication gap to others, and generally just keep on repeating the same ol and expecting something to magically be different because they demand it to be so… and then do nothing to facilitate the change.

Here’s a spiffy replay analysis that demonstrates key qualities on why players are bad.

For context: this is a replay of FAN playing with Jun (two of the top players) Since 40 minutes is obviously too long for anyone to reflect on their own actions, let alone those of someone else here’s a few key bits.

a) left side’s battle plan uses two healers; one for each lane on braxis. The idea is that they have a stronger 2v2, so the enemy side has to rotate more people top, so then the bot lane gets value.

b) Fan acknowledges some of his own mistakes.

During team fights, the players end up getting split, and left side eventually loses (spoiler, but then again, learning from mistakes, like losing, is how people can get better)

From the blaze’s perspective, he sees his dps heroes dying for being out of position. From the dps perspective, their tank is chasing kills he’s not going to get, not anchoring plays, not leading rotations, so they have to be in a different position to compensate.

So one side sees ‘mistakes’ from their allies, and tries to compensate. The other side sees ‘mistakes’ from their allies, and acts based on what they see.

For the ‘side’ that is [you], the other is in the ‘wrong’, and that’s the reason they lost. Doesn’t take a gm level game to see the same reaction regardless of mode: people lack experience, awareness, and a willingness to reflect on their actions compared to the perspective of the rest of their team, but so long as they find fault with some else, [they] must be right. (and also not picking a hero that suits how they actually play ofc)

However, a core ‘truth’ in all this is that if anyone is willing to know a little morea and look a little harder, then their actions will be different because they have a bit more information.

Tme and again, people fault an automated system that doesn’t know if people have map awareness, anchor plays, communicate with their allies, and a myriad of other not-performance reflected metrics that do influence how a game is won, or lost.

Players don’t know what they don’t know (how to watch the mini and read plays) and so long as they lack the willingness to critically review their own actions (learn from mistakes) they aren’t going to change; they have a higher dependence on ‘being lucky’ than on just making that ‘luck’. So the key part of where people can “improve the matchmaker” is by being a better, more informed player than they were the previous time that they played. Looking over comment sections, ama reactions and the like, a lot of negative reactions stem from myths, superstition, hearsay.

That’s why a myriad of games (in similar genres) all suffer “bad matching” Sure, there are ways they can improve things (anything can be better, that’s the point) but a fundamental defect stems from players who don’t strive to be the thing that will be better. And life-hack spoiler: those same flaws apply to real life too.

Lacking awareness, bad communication habits, unwilling to learn from mistakes via self-reflection? That affects how people drive, their customer practices, ability to improve at other skills/disciplines and even leadership/responsibility in political office: all highly visible consequences in current events so while I could use other examples, those are visibily accessible enough for people to ‘see’ the connection.

The 40 minute video is 40 minutes because there isn’t just one super mistake that causes the game to flop, it’s a series of small mistakes that add up; players see “stomps” in their own games because they aren’t aware of the little mistakes. People demand some dramatic flaw to blame (matching isn’t “fair”) because they don’t have the patience to assess collective details. It’s so much easier to tl;dr and be stagnant than it is to foster a bit more effort, broader perspectives, and increase personal awareness. If people aren’t aware that something is possible, then of course they won’t consider it a possibility in their assessment of… ‘life’.

But so long as they have an attention span shorter than a goldfish, they aren’t going to retain much more than the emotional whiplash of remembering something ‘bad’ happened, and not know why.

MMR is not a ‘fixed’ value; people are supposed to improve and constantly challenge the assessment of their ability. If you are ‘better’ than your weight of the system, then of course it’s going to be ‘mismatched’ because [you] aren’t a dank stagnant pool of rot (the consequence of still water in nature.)

Or at least [we] aren’t supposed to be. THAT is the magical explanation of mmr: people expect a static system to be ‘fair’ when the whole thing is supposed to be dynamic. When people don’t realize their disconnected expectations influence their reaction to “issues”, then they don’t apply themselves to changing that.

Tl;dr, people that look to fault something want an easy scapegoat and will generally argue themselves blue in the face instead of checking a myriad of ways they can improve what they can actually influence. If they knew they could influence stuff, they wouldn’t be looking for things to scapegoat cuz that’s time/effort they can use for other reactions they probably didn’t even consider because they weren’t aware those were even options.

And that’s the whole cycle: if people aren’t aware stuff are options, then of course they don’t take them.

1 Like

:roll_eyes:
That was true a year and a half or two ago. Rank did not represent skill directly.

Currently, rank reflects your skill level minus any leaver penalties.

Now, that skill level will practically always be different from your QM/URD skill level that the game uses to matchmake.

But the fact you’re lacking some basic facts makes the whole post rather moot overall, in it’s conspiracy type thinking.

You do get how MMR is averaged to approximate two teams with close to 50% winrate each. And the longer the queue timer goes (at least in QM) the wider degree of skill the game allows.

So say at <1 minutes it would be players within a 3 skill range, 4-6. After 2 minutes perhaps 3.5-6.5, after 4 minutes, 2-8. Et cetera.