Plz explain the 50% win rate to me

General Discussion
245

That’s likely mostly a result of comps being wonky at least in QM. Expecting the game to somehow balance around all the matchups and counters is not realistic and would make finding a match an absolute nightmare. The game only has the MMRs of individual players to go on which changes each match. It calculates the average and it tries to make the average equal. That said if so many people think the game has forced 50% winrate it does mean that the matchmaker is working since that’s what a working matchmaker would look like.

In the draft modes I expect it’s that they’re barren enough that getting consistently even matches is difficult.

I’m curious though what would a perfectly functional matchmaker look like to you? In an ideal world how would you tell if a matchmaker was working well and making fair matches?

1 Like
246

The usual issue of players is that they take offense to anything that doesn’t give them what they want (a win) and then conclude the match wasn’t “fair.”
And I’m rather verbose in that regard as I’ve written heaps of pages to ‘interrogate’ what people think is “fair” and they will decrey the whole of the earth and the history of sports before they’ll reconsider what they think is “fair” – which ironically, tends to not be “fair”, but gl on getting anyone to notice that.

What people post in reactions to matches tends to only be one instance, rather than the series of matches from other players involve to demonstrate how the match wasn’t “fair”. “Oh I had 50 losses, so the game has to be rigged” That’s not actually a logical conclusion based on what is presented…

Part of the biggest issue for the playerbase is that people don’t play all the heroes and roles as skillfully as they claim they do. It’s easier for them to point out the “zomg noob” type mistakes for other players, but since they don’t watch their own replays, they generally won’t realize when they are the ones making uncharacteristic mistakes. Cuz everyone does; off games happen, sugar imbalance, stress, background distraction, and a myriad of other subtle catalysts do shift the ‘chemical’ balance of a person and shift their own performance a whole lot more than they’ll realize.
maybe those ‘win streaks’ were noted by said player actually pinging and communication with their team, but once they’re in the rut of a lose, they don’t respond or they even argue and insult their team instead. Considerations like that are usually neglected because “zomg I’m forced to lose”

Because players are not consistent, a system to estimate their ability around other players is going to have a wide margin of flex. Part of the issue of understanding that reality is since players have an inflexible understanding of “fair” what they don’t know about matchmaking parameters (why a match was made) is just going to conclude that things are ‘broken’ due to their own ignorance.

That’s part of why people will insist on superstition and hearsay that allows them to curtail their own contributions (or lack thereof) in favor of just looking for anything else to blame. You don’t see people pull up replays of what their potato allies did in the games they won to find out why they are ranked as they are, but they’ll certainly be an expert in expecting the game system to ‘rig’ the games and know that players will do specifics aside from winning, or losing games, across a slew of several heroes.

Those afk raynors that just push a single lane? Yea, sometimes their team still wins in spite of that, or funnily enough, sometimes winnions actually win in those games. The matcher doesn’t check for hit skillshots, successful ganks, anchoring players, first minute afk, or last second return (and turn the game around from not having a bot) and a myriad of other circumstances that players will cite because they don’t know how matching works.

That is also why just about any sort of game that isn’t 1v1 (and even then) will usually have a more vocal playerbase convey how broken/rigged/etc the matching is, but otherwise do little else to suad their own position. (per- or otherwise)

But regardless of how bad or not a matchmaking can be, the funny thing is that those sort of things tend to to be used in ‘real life’ because people usually just try to play the game instead of crying about it. If anyone bothers to look at sports predictions, seeding, ranking and a couple of other stuff (esp for the betting venues) that’s going to see a whole lot more “unfair” than pretty much anything posted online.

The notion of being “forced” to lose generally has to denounce and ignore other details people don’t want to consider, and since they don’t consider them, the appraisal of the match isn’t objective, so the conclusion becomes “unfair”

But unless someone actually puts in 2 bits of thought into considering what they mean by “fair” and put that into some if/then conclusions for an automated system to appraise into a functional average, then they’re pretty much barking at a tree and expecting happiness to fall out of it.

3 Likes
247

You’re not helping, you actually making the problem worse. The truth is that the developers know what’s going on but they will not change it because it is tied with a complete rewrite. They always have known the issue (remember the blue post about “it’s okay to try 8-10 times to go a rank higher”? I do).

  • Not matching people in draft/qm based on available characters/preferred picks and slot order (but I think that’s gone since the last time I played). This was a real issue back then when you knew you won with a certain setup too often. Checking profiles after the game cleared many things up in our cases.
  • Secondly, their metrics for player incentives should not based on the good matches alone. I want good people also in my bad matches who are on my player skill level. I see no point in playing a game which throws five dream games towards you only to have five absolute negative experiences where it becomes, at this point, an on-going cycle instead of a quality game.
  • Rewarding players when they perform better than the rest of the team with more mmr.
  • Forfeit/give up option. I don’t want to waste my time in games which are lost by default because you can clearly see contrasting differences when you’re favored by the matchmaker. If every match would be quality wise good (regardless if you win or lose), the game would be in a much better state. In these 15 minutes I can do something more important in life. A big reason why I abandoned HotS is because it simply does not respect my time. Playing WoW, on a game console or doing something productive in RL is more preferable at this point.
1 Like
248

This is what the devs said about matchmaking, 50% chance of winning every game.

Game 1 50/50 chance of winning
Game 2 50/50
Game 3 50/50
Game 4 50/50
Game 5 50/50
Game 6 50/50
Game 7 50/50
Game 8 50/50
Game 9 50/50
Game 10 50/50

This is what the guy in GDC Presentation said “Overall 50/50” and what Forced 50% looks like.

Game 1 Hard game (noob teammates)
Game 2 Hard game
Game 3 Hard game
Game 4 Even game
Game 5 Even game
Game 6 Even game
Game 7 Even game
Game 8 Easy game (pro teammates)
Game 9 Easy game
Game 10 Easy game

This perfectly explains why we have stomps, repeat back and forth losing streaks and winning streaks. Considering hundreds of random factors that could occur in a game still the matchmaker is the only constant across hundreds of thousands of players and matches and it develop recognizable patterns. Anomaly in the matrix…sure…for ppl like Neo, Fan and Leonblack, everyone else stays in the matrix.

Again the goal of the matchmaker is to make “Fun” matches not fair matches and herein lies the crux of the problem.

3 Likes
249

Have you actually ever stress tested your system in low player conditions? Like how it would behave if it doesn’t have enough players to fulfill 2 teams at similar MMR? Cause similarity to MMR isn’t the only thing you take into account. You oversimplified the example too much.

You also failed to mention the uncertainty that plays into matching players. You are not matched by MMR alone. New players with high uncertainty and poor skills do get matched with veteran players, high MMR and low uncertainty. The uncertainty plays a huge role in how the match maker decides if 2 players at are a similar MMR, and it is likely why you get potatoes when you are on a winning streak(as your own uncertainty grows larger as you go on streaks).

You also didn’t mention the constraint of time. Optimization algorithms are tuned to run at infinity, but you always add a time/ loop count/etc constraint to it so it yields a result in a decent amount of time.Have you checked if the match maker produces wild results when under both player count and time constraints?

The funny part is, match making used to work correctly before the game population plummeted in late 2018.

You also need to keep in mind that people do get very upset about getting a match with afkers/potatoes, or a loss streak. This is because people don’t have “infinite” (or 10000 coin flips) amount of time to work through your MMR getting the gist of it and settling at a 50% win rate. This is why people complain, and this is why you need to take measure against trolls and feeders.

1 Like
250

Today I didn’t intended for Storm League mode until Live Patch is released I still want that after updated I go to find remained Nexomania chests that saw will explained more Shards in order to purchased any Fall-event themed skins I must answer for heroes in Fall-event themed skins.

251

News, at 11. Repeat after me: you cannot have a satisfying match maker without critical mass.

252

News, at 11:05. Repeat after me: I know

I was just pointing that out as a conclusion.

1 Like
253

how can we talk about “higher skilled players” or mmr if i had a match with valla, lt. morales, gremane, raynor and MURKY in our team against a party which has reghar, uther, falstad, abathur aaaaaand… BUTCHER. it was a mass murder from the start. butcher had 450+ meat by the end of the match. is this “higher skilled players” we talk about? oh come on. how did math work here? i know i am not the only one who had these types of mathces but god damn it, at least make it not that obvious if algorithms said “we need u to lose, guys, sorry”

1 Like
254

QM is clownfiesta mode: stop complaning about it.
If you want proper comp, go UR or SL.

255

Then what do you match them on? The game currently tries to find people of matching roles for each team and tries to average out the MMRs to be equal. I asked what a perfectly functional matchmaker would look like to you not what a good one wouldn’t be.

So you’d want the variance between MMRs of players in the game overall to be narrower rather than averaging out the MMRs? Understand that this does significantly increases the time it takes to find a match because you need to find people in the right roles on both sides within a narrow mmr range.

This is easy to explain how this happens. The goal of the matchmaker is to make all matches as even as possible which means a 50/50 chance of winning or losing. The game assesses this buy comparing MMR numbers and tries to make both sides even based on that since it’s the only way a computer can possibly compare something as nebulous as player skill. If someone goes on a prolonged winning or losing streak that sends up a flag within the game that this player’s number might be wrong. In that case the game gives an, at first, temporary increase or decrease to that number to see if it had that player’s MMR number wrong. If the trend reverses the game undoes that temporary change since the streak was an anomaly and the player is where they ought to be. If the trend continues then that new number becomes the baseline and the game gives another temporary change to the MMR. The end result of that is that most players will end up hovering around 50% win rate since eventually players’ mmr will eventually adjust to the point where they are stomping opponents or getting stomped since their MMRs are too low or high respectively then get pulled back to where they belong. There isn’t really much of an alternative to this other than an MMR system that does that big boost/decrease on a game by game basis resulting in very few win/loss streaks at all but that doesn’t seem as enjoyable

That can help a bit and they really shouldn’t have scrapped performance based matchmaking

Irrelevant to matchmaking and would overall make the game much worse and more susceptible to trolling and griefing. The devs have also said they never plan to do this so don’t hold out any hope for it.

1 Like
256

UR and SL are QM with knowing the map beforehand.

1 Like
257

Just letting you know I’m a fan, Xenterex.

258

actually it doesn’t “perfectly explain” it just suits bias.

The ability to want to find patterns can be a form of bias that overlooks other details at hand; for instance, the conclusion is drawn afterward if something is a pro/newb player, but in the context of matching uncertainty (ie, the game doesn’t know the difference between a pro/not) people overlook that in favor of concluding they found what they wanted to find afterward.

That’s part of why you just post a string of expected ‘patterns’ rather than indicate the around-info I posted that people tend to ignore when they look at these sort of things.

That’s part of why the conclusion also just accepts that exceptions are exceptions and doesn’t ‘perfectly’ tie then into the explanation – which is part of why it isn’t ‘perfect’ – it just ignores figures that don’t suit the confirmation of the bias.

So if a ‘perfect’ explanation doesn’t actually explain it’s explanation ‘perfectly’ then calling it such is just a concession for wanting to claim it is that. And if peopel just want to claim something, and accept it, then they can also claim ‘it’s fine’ (the matching) and accept that with the same degree of ignorance.

The functional issue is that people seeing “fair” in one light, and expecting the system to follow a definition it doesn’t have. But since humans, ie players, actually have free will – a concept this sort of stuff tends to overlook entirely – they gets thrown out the window for these “perfect” explanations.

2 Likes
259

I think you are just belittling people, whatever people are experiencing are just biases, like its your job to instill doubt in people. You are reminding me of the documentary Merchants of Doubt about doubt mongering on tobacco.

First of all, this is based on a presentation at GDC “Overall 50/50”, not bias as you claimed time and time again.

Game 1 Hard game (unfavourable)
Game 2 Hard game
Game 3 Hard game
Game 4 Even game
Game 5 Even game
Game 6 Even game
Game 7 Even game
Game 8 Easy game (favourable)
Game 9 Easy game
Game 10 Easy game

It’s really about the matchmaker making favorable and unfavorable conditions for each team. Newb teammates for hard games and pro teammates for easy games are just 2 examples of making favorable and unfavorable conditions. The matchmaker have all our data, it’s not hard to create all sorts of conditions.

I believe even NotParadox said that in general 3 out of 10 games is unwinnable, you should just let them go and focus on the other 7. He said something like that which fits the above model nicely.

1 Like
260

What you think of me as ‘belittling’ people is you not liking that someone disagrees with you.

Of the replies here, I’ve posted details people overlook, explain why I call their conclusions a confirmation of bias (everyone is biased, but how they use it matters) and in turn you ignore what I wrote, go off on red herrings instead and then repeat yourself.

Ignoring what other people post isn’t them belittling others, but if they ‘feel’ offended, they’re rationalize an excuse to ignore what other people so long as they’re convincing the other person won’t agree with them.

Persisting in this behavior is pretty much just more examples of confirmation bias.

What notparadox concluding is a bit apart from what you want to say, but since it suits your context, you find it ‘agrees’ with your premise and use it as an example of the above cycle.

Part of the issue of the interpretation of matching is that it assumes human-like qualities the system does not have. However, because the conclusion suits what people want to say of the system (circular reasoning) they choose to forgo details that don’t suit that conclusion and settle on brief exceptions to the standard they want to profess.

If you’re under the impression that someone not agreeing with you, explaining their conduct, and why point A doens’t match with claim Z, then that’s you choosing to take offense where none was given, and just concluding an incorrect conclusion because it’s easier to revel in ignorance statements that have what they want, rather than look into details that could influence their perspective.

Back and forths like this are part of why I know people neglect details; they do it in reading the stats for the games, they do it for the replies people post; the conduct of one thing demonstrates the issue of the other.

And if you, or any one else don’t realize that, then that’s part of the flaws of the ‘bias’ that tends to result in similar dissatisfaction people take to the matchmaking. They have something that doesn’t agree with their expectations, don’t look to become better informed on ‘how’ something works, and then take offense that they aren’t getting what they ‘want’ and instead look for something else to blame.

And yes, your ‘rebuttal’ for what I’ve written is to just skirt details and shift blame instead.

My ‘job’ isn’t to instil doubt on people, it’s to point out that they’re are perspectives they aren’t considering, and because they ignore details to suit a forgone conclusion they already want, they feel offended that someone doesn’t agree with them rather than take a different, or even more ‘honest’ approach and actually take something into consideration that they haven’t.

2 Likes
261

I have given up talking with people here as well. The people who offer “solutions” to the problems are just sock puppets for the developer team and do not contribute anything to the problem but their own bias. Not seeing MMRs should have been a red flag for years but it’s better for some people to be apologetic foot-kisser because they profit from it in some form (smurfing, boosting, luck).

You can argue all day long but it’s a fact that the game is on a life support for questionable decisions the team has made and this is the result of it. Matchmaking is the #1 problem in the game and has been for most of its lifetime. It will NEVER have a third wind unless they rework the matchmaking.

1 Like
262

You are not even talking about the game. You are just making bias lectures on bias.

1 Like
263

This is flat out false and untrue. A challenge of the MMR system is when employees are dishonest about the system (regardless of it being intentional or simply uninformed) Leaving out information that cuts against your argument is another way of creating a lie. Politicians due it all the time. Half truths are getting old both in this game and in reality.

If the MMR was simply W/L based it would actually be way better and even fair. The fact of the matter is it is not. Depending on where the “system” thinks you should be will determine how many points you get for a “win or loss”. This should never happen if you are going to make the argument that over time the statistics will work themselves out.

Simple case and point of what should be a 50% win rate and MMR break even. Say for simple math your MMR is 1000. You start a new season and get unlucky losing 10 in a row, but then win 10 in a row. You should be right back to 1000. However, each loss gives you -150 pts. Each win (as they came after) now give you +75. So instead of a 50% win keeping you at 1000 MMR, you are now at a MMR of 250, significantly lower than the 1000 you should be at.

This is why any argument about statistics and “eventually working it out” are flat out false. It does not work itself out because the system does not allow a simple win/loss to determine your MMR. It catered to people who were losing rank because they were pulled by friends, to keep them there. This right here ruined any notion of the system being “fair” and eventually correcting for any win/loss streak. It punishes those who are learning and their rank drops, and when they start to win the MMR tries to keep them low. Or rewards those were pulled and keeps them high even while losing. It makes a “guess” after such a mathematically non-significant number of games where it thinks you are then tries to keep you there buy ensuring that if you start to improve you get less points for wins and more for losses and vice versa.

This is why the best strategy years ago is make a smurf and team up with people for the first 50 games. After that does not matter how much you lose… The MMR bonus for win loss keeps you where the system thinks you should be not what your play merits.

It is also sad how an outlier is constantly used to say… see it works. Usually, a pro player who smurfs. This is so the account has very few games. Which is nothing like someone who has played for a while and improved. Anyone with any statistical knowledge would not use rare outliers to make generalizations for a population at large. Yet Blizzard does it all the time. Sadly, instead of just being honest they manipulate statistics to say what they want.

There it is across the board Blizzard has stopped listening to players and only listen to themselves and are defensive and defending systems they created that do not work. Instead of collaboration they simple become defensive and dig in more. A great example is Diablo 3 immortal… I mean that could not be Blizzards fault you all must not have phones.

1 Like
264

You first response to me was based on my post of 246.

The very first line I wrote is

What I explain in that post is very particular to this game, but it also applies to other games because that sort of behavior is the sort of thing people take from one game to the next to the next and so on.

Of what you have replied to me, you have gone on to demonstrate the very things I claim, refuse to acknowledge the problem of you trying to ignore what I’ve written despite 'proving it, and then go on to try to blame me for the problems you’re exhibiting.

Yes, I am talking about the game, when you decide to finally notice and stop denying relevant information, you might actually have a point. However, when people cripple themselves and their ability to have a discussion because their bias, then getting a ‘lecture’ on bias is a valid response.

Maybe you should set aside your bias for 2 thought and consider why that is relevant before doing more of the same.

Most of the issues for the game stem from player perception.

The ‘life support’ aspects of the playerbase generally come from misinformation, bad expectations, and looking for something to blame.

As an example, if you look over the comments section (horrying to suggest, I know) of carbot videos and announcements, a sample of those are players that just assume the game is ‘ded’ stopped updating entirely, and a number of other misconceptions.

This isn’t about being a stickler for details of “well technically it is ded” as the extent of their information is outdated by years; like operate selective hearsay and don’t bother to check with a basic google search. And that’s been problematic for the game since it’s inception; people had certain expectations, refused to consider otherwise, and are content to be misinformed about stuff.

Now that isn’t to say the dev team isn’t to blame, or couldn’t have done something to correct misconceptions or the like (esp since what playerbaes a game gets is a resource the devs have to use, or they will lose it) but the same sort of players in that rut just carry those sample problems to other games they visit so they end up in this floating ambience of dissatisfaction and feeds almost entirely on nostalgia because they only ‘remember’ when stuff was good, rather then learned how to figure out how to make it be ‘good’ for themselves.

HotS could actually get a ‘third wind’ because the basic gist of what keeps the game down is speculation and rumors. However, a key flaw of hots is that most of blizzard’s playerbase come from pve games, so having a game where the majority of content is pvp doesn’t jive with them. That’s part of why most of the issues are complaints about superficial aspects based on ignorance, and not more quantifiable qualities.

HotS gets compared to “done” games like their rts series that have hundreds of unit assets, lots of existing spell animations, and loads of content that they can play alone, or with user-created variations.

If HotS was maintained closer to a WoW expansion and had “done content” released in large waves, with regular pve solo content, then it’d be a lot easier for players to dissect and just play the game instead of looking for fluff to fault.

Now part of the concern can be for getting the initial effort into making that shift (and it’d take years of background updates to make that) but so long as blizz doesn’t want to regulate the game to ‘ded’ and "shut down the servers’ and some players keep thinking is going to happen ‘any day now’, it can happen.

1 Like