Analysis of "forced 50%" win claims. Improvement Suggestions

Competitive Discussion
EDIT: Spoiler for people that don't want to read all this - "forced 50% win" is complete baloney.

Let's lay out some facts here so that we can settle this "forced 50%" claim once and for all.

Let's break it down one step at a time, and Sherlock Holmes the hell out of this with pure logic and brain power and common sense.

And then let's see what improvements we can add to competitive mode.

Let's do this.

THE FACTS

1. You are matched to your opponents based on your hidden MMR(Matchmaking Rating). It is some kind of performance based rating based on your different hero's game statistics. We do not know what the exact parameters are. And we SHOULDN'T know because people will stat pad heroes and abuse it to gain more SR. Some people have already been abusing this and climbing with off-meta heroes and one-tricking.

If you acknowledge this is how MMR works, move onto the next step, otherwise provide evidence on how this is not the way it works right now. My proof is Scott's blue post, it's basically exactly how he explained how MMR works.

2. The match maker will find 12 players within similar MMR range and form 2 teams to fight against each other, which means both teams have equal/50% chance of winning.

BUT there is no "role preference" system or "player hero hours" taken into consideration at all, just 12 players and their MMR range. You will end up with random team comps like 5 DPS mains or 6 mercy mains. It can happen to BOTH your team and enemy team.

If you acknowledge this is how the matchmaker works right now, move onto the next step, otherwise provide evidence on how this is not the way it works right now. My proof is also Scott's blue post. He mentioned 50% chance of winning we can assume they want the 2 teams with MMR that is similar to each other. And based on how fast it takes for us to find matches in competitive, we can assume the algorithm is quite simple.

3. Both your team and enemy team uses the same MMR and Matchmaker

Even if we ASSUME that the performance based MMR might not be a good indication of skill, ASSUME that the matchmaker is very bad at putting together a team that have good team comp. The match quality might be completely crap. But DESPITE THAT, BOTH teams are still in the same position, they both use the same MMR and matchmaker.

If you acknowledge 12 players uses the same MMR and matchmaker and get formed into 2 teams, move onto the next step, otherwise provide evidence on how this is not the way it works right now.

4. Now we get into the matches, you play your matches, you either win or lose for WHATEVER reasons, but it is out of SYSTEM's control.

The outcome is outside of system's control, the result is completely up to how you, your team and enemy team plays, trolls, toxic people, internet DC, derankers, WHATEVER the reason, the MMR and matchmaker has NOTHING to do with how you won or lost

If you acknowledge this is how the outcome of the matches are decided, move onto the next step, otherwise provide evidence on how the MMR and Matchmaker SOMEHOW make JUST you win or lose, when BOTH teams are playing with the same MMR and Matchmaker system.

5. Ok now that you've won/lost a match. Your SR and MMR will both change based on your performance of the heroes you used during that match.

You gain SR by winning, you lose SR by losing. The amount is decided by your performance. We have NO idea how MMR changes because it is hidden.[

If you acknowledge this is how SR and MMR changes, move onto conclusion phase. If you don't. Provide evidence on how this isn't how your SR and MMR changes.

BASIC STATISTICS AND COMMON SENSE AND CONCLUSIONS

1. If you flip a coin 10 times....
The chance of HHHHHTTTTT happening is EXACTLY the same as HTHTHTHTHT.
Each coin flip is INDEPENDENT of the previous coin flip. You CANNOT predict the next outcome of a coinflip/match BASED ON PAST PATTERNS when both heads/tails your/enemy team are in the SAME environment and have the SAME odds as each other every single match.

Therefore observations of patterns does NOT automatically mean the system is forcing you to lose when you start losing after a lucky win streak.

2. If you happen to get into a winning streak, your SR and MMR will increase, matching you into harder opponents each time. However your actual skill might not have improved IN PACE with the increase of SR/MMR.

Therefore when you lose after a streak of wins, you might just be out of your depth. your opponents are too good for your current skill level, 5 other team mates cannot carry you. Which is quite true as one single team member not performing well enough will most likely lose the match.

3. To make a system that actually find specific players to make them lose. It is VERY complex and isn't really worth a developer's time to develop as this feature serves no purpose what so ever.

Firstly, you need to track a player's history, it means extra database space.
Secondly, you need to somehow ENSURE the player loses the next match. I don't know what software options there are that can actually do this, when all players use the same MMR and matchmaker system.
Lastly, even IF they match 5 bronze players with you against 6 grand masters. you might still win. Because you cannot predict the future, what if all the GMs disconnect? what if they just felt like throwing? What if your 5 bronzes are actually GM smurfs?

What I'm basically saying is.... even if the matches are rigged. there is NO software solution that can guarantee what the FUTURE is. It's against the laws of this universe.

So is there nothing to improve? Everything is perfect?
Far from it.....

Of course there is a lot of things that NEEDS to be improved with the matchmaker and MMR system.

1. Match quality can be improved by a mile by introducing some kind of hero preference system and karma system.

Hero preference system, before queuing, players can prefer the heroes they wish to play or choose the option to flex, the matchmaker will try to form players into teams based on their hero preferences and form teams that can pick majority of the heroes in the roster. This way you can make sure your whole team can play wide range of heroes instead of getting RANDOM mains here and there.

Karma system, pretty simple, add karma score to players you enjoyed playing with, negative karma score to players you didn't like playing with. Matchmaker will prioritise matching players with good karma together.

2. Individual performance based SR might not be good representation of player skill.

How can DPS perform if tanks don't block damage for them and healers don't heal them?
How can tanks perform if DPS does not return fire and healers don't heal the tanks?
How can healers perform if DPS and tanks don't protect them?

This is a team game, your INDIVIDUAL performance is TIED to your team. Either make everyone on your team gain/lose same SR, or introduce solo queue so measuring individual performance make more sense. Right now with solo players and group players playing together, there is NO WAY a solo player's SR equates to a 6 stack player's SR.

3. Separate solo queue and group queue, each have separate rating system.

Solo queue: Solo players only. Current Performance based SR.
Group queue: Duos, Trios, Quads and Sixes. Party of 5 disabled. Team based SR.

(Team based SR means the only variance of SR gain/loss is the difference in SR between the two teams. ALL members will gain/lose the same amount of SR per game.)

If all of the above are implemented, I don't think any player would care how much longer their queue times will be. Players would rather wait longer and have a GOOD match, instead of consistently put into bad matches very quickly.
50% win rate is a myth. I'm currently on 45.5% after 150 matches, and have never ended a season with better than 48% (normally less). Maybe it tries but I pick heroes that are too prone to people throwing matches I should have won, maybe I'm just unlucky. My win deficit last season, should 50% have been real would have put me in Platinum, but it didn't.

The manner in which I'll climb steadily, then hit a wall of 12 or more consecutive losses does cause me to ask questions of the system, as this is a consistent pattern over more than a year. But I don't think it has anything to do with forcing 50%, because, well, surely it wouldn't happen if I hadn't past the 50% mark, right?
while you basement and suggestions are good and well known, I think there is something I want to address with your conclusion:

10/11/2017 05:08 AMPosted by Stebung

BASIC STATISTICS AND COMMON SENSE AND CONCLUSIONS

1. If you flip a coin 10 times....
The chance of HHHHHTTTTT happening is EXACTLY the same as HTHTHTHTHT.
Each coin flip is INDEPENDENT of the previous coin flip. You CANNOT predict the next outcome of a coinflip/match BASED ON PAST PATTERNS when both heads/tails your/enemy team are in the SAME environment and have the SAME odds as each other every single match.

Therefore observations of patterns does NOT automatically mean the system is forcing you to lose when you start losing after a lucky win streak.

Something rated to the statistics. It is better to learn some basic statistics 1-2-3 rather than discussions here. But I want to tell you although "The chance of HHHHHTTTTT happening is EXACTLY the same as HTHTHTHTHT", they are all fall in 5H-5T pattern. Actually it is the most possibilities if you flip coins ten times. What is the histogram chance for 5T-5H? it is about 24.6%. 4T-6H or 6T-4H? It is 20.5%. If you play 10 games, most of time you will fall in these categories, which ppl will find fair. When talking about streak... What is the 10T-0H or 0T-10H chance? Less than 0.1%.
Thus it is not the next time 50% winning rate you should consider; you should think what is the winning/losing streak possibilities. For example, if system working as designed, the 10 game winning or losing streak should be very rare. Average you play 1000 game you will have ONE 10 game streak, winning or losing.
From the feedback of the forum and players, it seems the streak rate is far more than that. If it is true beyond our feeling, there must be something wrong with it.

10/11/2017 05:08 AMPosted by Stebung

2. If you happen to get into a winning streak, your SR and MMR will increase, matching you into harder opponents each time. However your actual skill might not have improved IN PACE with the increase of SR/MMR.

Therefore when you lose after a streak of wins, you might just be out of your depth. your opponents are too good for your current skill level, 5 other team mates cannot carry you. Which is quite true as one single team member not performing well enough will most likely lose the match.

you need to consider what is MMR created for? Shouldn't MMR tells the true SKILL LEVEL beyond the game win/lose? if MMR works as intend, even you lucky enough get 10 win streak, internally your MMR shouldn't change much. That is to say, one should NOT get lose streak immediately after win streak with MMR system. If it does, that means the MMR system is not working as designed purpose.
Actually IMHO, evaluating players performance with some kind of computer algorithm is a doomed failure. You can check my previous posts for this. One word, the team game is too complicated for single player performance value evaluations.

10/11/2017 05:08 AMPosted by Stebung

3. To make a system that actually find specific players to make them lose. It is VERY complex and isn't really worth a developer's time to develop as this feature serves no purpose what so ever.

Firstly, you need to track a player's history, it means extra database space.
Secondly, you need to somehow ENSURE the player loses the next match. I don't know what software options there are that can actually do this, when all players use the same MMR and matchmaker system.
Lastly, even IF they match 5 bronze players with you against 6 grand masters. you might still win. Because you cannot predict the future, what if all the GMs disconnect? what if they just felt like throwing? What if your 5 bronzes are actually GM smurfs?

What I'm basically saying is.... even if the matches are rigged. there is NO software solution that can guarantee what the FUTURE is. It's against the laws of this universe.

[/quote]
Actually, it can be done. You can track the player drop line rate easily. Matching you with high dropping rate player even you all have similar MMR, you will find how easily to destroy one's gaming experience.
I don't think Blizzard intends to do this neither. However, their MM algorithm sometimes WILL DO. Just think what is the quickest way to drag one from 100% winning rate to 50% at dedicated MMR range? The programs do not have feelings, they just want to accomplish the tasks assigned to them...


2. If you happen to get into a winning streak, your SR and MMR will increase, matching you into harder opponents each time. However your actual skill might not have improved IN PACE with the increase of SR/MMR.

Therefore when you lose after a streak of wins, you might just be out of your depth. your opponents are too good for your current skill level, 5 other team mates cannot carry you. Which is quite true as one single team member not performing well enough will most likely lose the match.


That is the only part you are wrong about. I don't know if you have never truly climbed a significant amount (range) or are just full of ideal wishful thinking.

The reality is that YES someone ~500 SR more than you is likely better at the game (not always , if they just don't play enough to drop).

But, if you stomp 3 in a row and your are hard carrying your team. You do not face harder opponents...you face the same SR level of opponents only YOUR TEAM gets WEAKER.

Now this isn't the system out to get you. This is simply the system TESTING YOU and YOUR POTATO TEAMMATES. If YOU win then your MMR goes up and you start seeing a new pool of players. If your teammates do poorly their MMR probably still goes down even if their SR goes up (or their MMR goes nowhere). Then these people come to the forum complaining about <20 SR on a win, and make up conspiracies re: tanks and supports.

Now if YOU LOSE that game (or couple games) your MMR probably goes down and you still get a new pool of people to be tested with AGAIN.

This system works well testing skill. It is roughly accurate in a way that is quite cool considering the complexity of the game.

But it is this testing phase that pisses people off. People don't like losing. And they hate being made an underdog because they think it is forced losses. They think of each game as super important and don't think in batches of games changing their MMR and SR just roughly reflects their MMR. Which is how some with > 50 % win rate drop and others with < 50% winrate climb.

My point is that if you consistently TRY YOUR BEST. You will get to a roughly accurate SR. You will WIN and you will LOSE. But eventually you will get to a point that you don't really get long streaks and that is YOUR SR. Until you IMPROVE some aspect of your play.
Something rated to the statistics. It is better to learn some basic statistics 1-2-3 rather than discussions here. But I want to tell you although "The chance of HHHHHTTTTT happening is EXACTLY the same as HTHTHTHTHT", they are all fall in 5H-5T pattern. Actually it is the most possibilities if you flip coins ten times. What is the histogram chance for 5T-5H? it is about 24.6%. 4T-6H or 6T-4H? It is 20.5%. If you play 10 games, most of time you will fall in these categories, which ppl will find fair. When talking about streak... What is the 10T-0H or 0T-10H chance? Less than 0.1%. Thus it is not the next time 50% winning rate you should consider; you should think what is the winning/losing streak possibilities. For example, if system working as designed, the 10 game winning or losing streak should be very rare. Average you play 1000 game you will have ONE 10 game streak, winning or losing.From the feedback of the forum and players, it seems the streak rate is far more than that. If it is true beyond our feeling, there must be something wrong with it.


Basically it comes down to players only reporting loss streaks and either forgetting about win streaks or outright lying and not including them on purpose. You can't really trust information they provide. You can get all kinds of streaks on random.org coin flips though.

you need to consider what is MMR created for? Shouldn't MMR tells the true SKILL LEVEL beyond the game win/lose? if MMR works as intend, even you lucky enough get 10 win streak, internally your MMR shouldn't change much. That is to say, one should NOT get lose streak immediately after win streak with MMR system. If it does, that means the MMR system is not working as designed purpose. Actually IMHO, evaluating players performance with some kind of computer algorithm is a doomed failure. You can check my previous posts for this. One word, the team game is too complicated for single player performance value evaluations.


There's no way for the game to know for sure if the player has improved or not. Overwatch's Elo system uses a "certainty factor" which makes MMR stay in place if a player is going 50/50. But MMR will get way less certain if a player will get a big enough lucky win streak. I also wonder how do you intend to find groups without some kind of a matchmaker? Through thousands of half empty dedicated servers?

10/11/2017 03:09 PMPosted by xsun
Actually, it can be done. You can track the player drop line rate easily. Matching you with high dropping rate player even you all have similar MMR, you will find how easily to destroy one's gaming experience. I don't think Blizzard intends to do this neither. However, their MM algorithm sometimes WILL DO. Just think what is the quickest way to drag one from 100% winning rate to 50% at dedicated MMR range? The programs do not have feelings, they just want to accomplish the tasks assigned to them...


Lol so first you say the matchmaker doesn't do that and then immediately contradict yourself. Algorithms are not alive and will not do anything that wasn't specifically written to them. Therefore, for matchmaker to be rigged someone has to deliberately write the necessary pieces of code. And, dude, just how many times do developers have to say matchmaker doesn't care about win rates so you'd realize it?
10/11/2017 03:41 PMPosted by SleepyBear
The reality is that YES someone ~500 SR more than you is likely better at the game (not always , if they just don't play enough to drop).But, if you stomp 3 in a row and your are hard carrying your team. You do not face harder opponents...you face the same SR level of opponents only YOUR TEAM gets WEAKER.


Lol if nobody's skill changes as a player progresses, then how do his teammates get weaker? Besides, there's no such thing as a carrying, it's called doing your job.

10/11/2017 03:41 PMPosted by SleepyBear
Now this isn't the system out to get you. This is simply the system TESTING YOU and YOUR POTATO TEAMMATES. If YOU win then your MMR goes up and you start seeing a new pool of players. If your teammates do poorly their MMR probably still goes down even if their SR goes up (or their MMR goes nowhere). Then these people come to the forum complaining about <20 SR on a win, and make up conspiracies re: tanks and supports.Now if YOU LOSE that game (or couple games) your MMR probably goes down and you still get a new pool of people to be tested with AGAIN.


Good lord, how many times do developers have to tell their matchmaker only looks for people with close enough MMR so people would finally get it? Pools of players? Testing? It doesn't do that. And where is your evidence, by the way?
10/11/2017 03:54 PMPosted by Medal
Something rated to the statistics. It is better to learn some basic statistics 1-2-3 rather than discussions here. But I want to tell you although "The chance of HHHHHTTTTT happening is EXACTLY the same as HTHTHTHTHT", they are all fall in 5H-5T pattern. Actually it is the most possibilities if you flip coins ten times. What is the histogram chance for 5T-5H? it is about 24.6%. 4T-6H or 6T-4H? It is 20.5%. If you play 10 games, most of time you will fall in these categories, which ppl will find fair. When talking about streak... What is the 10T-0H or 0T-10H chance? Less than 0.1%. Thus it is not the next time 50% winning rate you should consider; you should think what is the winning/losing streak possibilities. For example, if system working as designed, the 10 game winning or losing streak should be very rare. Average you play 1000 game you will have ONE 10 game streak, winning or losing.From the feedback of the forum and players, it seems the streak rate is far more than that. If it is true beyond our feeling, there must be something wrong with it.


Basically it comes down to players only reporting loss streaks and either forgetting about win streaks or outright lying and not including them on purpose. You can't really trust information they provide. You can get all kinds of streaks on random.org coin flips though.

you need to consider what is MMR created for? Shouldn't MMR tells the true SKILL LEVEL beyond the game win/lose? if MMR works as intend, even you lucky enough get 10 win streak, internally your MMR shouldn't change much. That is to say, one should NOT get lose streak immediately after win streak with MMR system. If it does, that means the MMR system is not working as designed purpose. Actually IMHO, evaluating players performance with some kind of computer algorithm is a doomed failure. You can check my previous posts for this. One word, the team game is too complicated for single player performance value evaluations.


There's no way for the game to know for sure if the player has improved or not. Overwatch's Elo system uses a "certainty factor" which makes MMR stay in place if a player is going 50/50. But MMR will get way less certain if a player will get a big enough lucky win streak. I also wonder how do you intend to find groups without some kind of a matchmaker? Through thousands of half empty dedicated servers?

10/11/2017 03:09 PMPosted by xsun
Actually, it can be done. You can track the player drop line rate easily. Matching you with high dropping rate player even you all have similar MMR, you will find how easily to destroy one's gaming experience. I don't think Blizzard intends to do this neither. However, their MM algorithm sometimes WILL DO. Just think what is the quickest way to drag one from 100% winning rate to 50% at dedicated MMR range? The programs do not have feelings, they just want to accomplish the tasks assigned to them...


Lol so first you say the matchmaker doesn't do that and then immediately contradict yourself. Algorithms are not alive and will not do anything that wasn't specifically written to them. Therefore, for matchmaker to be rigged someone has to deliberately write the necessary pieces of code. And, dude, just how many times do developers have to say matchmaker doesn't care about win rates so you'd realize it?
I may not make myself clear enough? I didnt say anything about MM system itself, How it works, it is a mystery to us. I never say it will or it won't. No one can say except the dev team.
I just want to address that OP said MM system cannot do something dirty, which is not true. If the calculation result leads, it can do. However, just as say it again, the odd behavior of MM system is not purposely designed, or intended. It may be the side effect of the whole MM system.
You get fever, you go to doctor and take the pill, and you find you back pain after you recovery. I think it is not that hard to understand.
Again, all the matchmaker does is put 12 players together in a match and divide them by teams. It can't possibly predict the result of the match and if the team has lost, that's its fault entirely. And that was the point of the initial post. The matchmaker is not responsible for matches' outcomes, is all there is to it.

Loss and win streaks come naturally in any sort of a random distribution. More than that, I'd find it strange for a player to go like win-loss-win-loss-so on with no streaks in-between. They happen, and they're part of the experience, and it's something that can't be fixed. Which was the point of your post, I believe.
10/11/2017 04:01 PMPosted by Medal
10/11/2017 03:41 PMPosted by SleepyBear
The reality is that YES someone ~500 SR more than you is likely better at the game (not always , if they just don't play enough to drop).But, if you stomp 3 in a row and your are hard carrying your team. You do not face harder opponents...you face the same SR level of opponents only YOUR TEAM gets WEAKER.


Lol if nobody's skill changes as a player progresses, then how do his teammates get weaker? Besides, there's no such thing as a carrying, it's called doing your job.

10/11/2017 03:41 PMPosted by SleepyBear
Now this isn't the system out to get you. This is simply the system TESTING YOU and YOUR POTATO TEAMMATES. If YOU win then your MMR goes up and you start seeing a new pool of players. If your teammates do poorly their MMR probably still goes down even if their SR goes up (or their MMR goes nowhere). Then these people come to the forum complaining about <20 SR on a win, and make up conspiracies re: tanks and supports.Now if YOU LOSE that game (or couple games) your MMR probably goes down and you still get a new pool of people to be tested with AGAIN.


Good lord, how many times do developers have to tell their matchmaker only looks for people with close enough MMR so people would finally get it? Pools of players? Testing? It doesn't do that. And where is your evidence, by the way?


Its the same as QP. The same players get shuffled around to make new teams. Fairer teams. If people insta q right after.

With millions of players why would we see the same people for a few games in a row. Only with some enemies on your team and now some of your team on the other team.?
I already made a giant thread going over probability and how only after 10 games for 5 million (or whatever players) we expect so many to have a giant loss streak...

and that's only after 10 games. That's not after hundreds, thousands of games.

But, they want Blizzard to defy probability.
@Medal, thanks for actually understanding what I'm trying to say. And actually trying to help others understand as well.

10/11/2017 05:23 AMPosted by Jessicka
50% win rate is a myth. I'm currently on 45.5% after 150 matches, and have never ended a season with better than 48% (normally less). Maybe it tries but I pick heroes that are too prone to people throwing matches I should have won, maybe I'm just unlucky. My win deficit last season, should 50% have been real would have put me in Platinum, but it didn't.

The manner in which I'll climb steadily, then hit a wall of 12 or more consecutive losses does cause me to ask questions of the system, as this is a consistent pattern over more than a year. But I don't think it has anything to do with forcing 50%, because, well, surely it wouldn't happen if I hadn't past the 50% mark, right?

Not too sure what you are suggesting here, so are you saying instead of forced "50%", the system is still somehow against you and putting you at sub 50% win rate?

You hit a wall and consistently get losing streaks at a certain point?

I don't know what to tell you.... despite all I've said I'm sure you understand that the matchmaker ONLY just make matches? It does not have any influence over wins and losses.

Perhaps it's better to just take a break when you have 2 losses in a row, and start playing again when you are refreshed? When people are tilted they don't make the right decisions and they won't perform as well as usual. Perhaps it's something to do with that.

In reply to xsun, because quotations in this forum is weird....
while you basement and suggestions are good and well known, I think there is something I want to address with your conclusion:

Something related to the statistics. It is better to learn some basic statistics 1-2-3 rather than discussions here. But I want to tell you although "The chance of HHHHHTTTTT happening is EXACTLY the same as HTHTHTHTHT", they are all fall in 5H-5T pattern. Actually it is the most possibilities if you flip coins ten times. What is the histogram chance for 5T-5H? it is about 24.6%. 4T-6H or 6T-4H? It is 20.5%. If you play 10 games, most of time you will fall in these categories, which ppl will find fair. When talking about streak... What is the 10T-0H or 0T-10H chance? Less than 0.1%.
Thus it is not the next time 50% winning rate you should consider; you should think what is the winning/losing streak possibilities. For example, if system working as designed, the 10 game winning or losing streak should be very rare. Average you play 1000 game you will have ONE 10 game streak, winning or losing.
From the feedback of the forum and players, it seems the streak rate is far more than that. If it is true beyond our feeling, there must be something wrong with it.

I think you completely missed the point here.... As Medal pointed out. It is NOT ABOUT the patterns at all. It doesn't matter what % each pattern occurs. It is the fact that DESPITE the previous patterns of the coin flips, the next coin flip is independent of that pattern.

The ONLY correlation I can think of that's even remotely close to what you are referring to might be pachinko machines or slot machines in casino that would guarantee no jackpots are hit for a certain period of time.

But guess what? That is also a myth. It is illegal to actually do that. You legally cannot rig something that is based on chance to happen to favour or unfavour someone or something.


you need to consider what is MMR created for? Shouldn't MMR tells the true SKILL LEVEL beyond the game win/lose? if MMR works as intend, even you lucky enough get 10 win streak, internally your MMR shouldn't change much. That is to say, one should NOT get lose streak immediately after win streak with MMR system. If it does, that means the MMR system is not working as designed purpose.
Actually IMHO, evaluating players performance with some kind of computer algorithm is a doomed failure. You can check my previous posts for this. One word, the team game is too complicated for single player performance value evaluations.

You cannot make assumptions on something we cannot see. All we know about MMR is that it is based on your performance and it is used to find matches for us.
Whether MMR + SR is a good representation of someone's true skill is another discussion completely. If you've read the last section I actually am against performance based SR.


Actually, it can be done. You can track the player drop line rate easily. Matching you with high dropping rate player even you all have similar MMR, you will find how easily to destroy one's gaming experience.
I don't think Blizzard intends to do this neither. However, their MM algorithm sometimes WILL DO. Just think what is the quickest way to drag one from 100% winning rate to 50% at dedicated MMR range? The programs do not have feelings, they just want to accomplish the tasks assigned to them...

You need to be very careful here about making accusations. You are assuming they actually track our match history. Where is the proof?

Say that even if there is a "drop rate" field in our MMR. It is actually a completely useless field to decide if a player is good or bad.
Why? Because..... as we discussed before....basic statistics..... patterns CANNOT predict the outcome of next match.... players with high "drop rate" does not mean they will DEFINITELY lose the next match.

10/11/2017 03:41 PMPosted by SleepyBear

That is the only part you are wrong about. I don't know if you have never truly climbed a significant amount (range) or are just full of ideal wishful thinking.

The reality is that YES someone ~500 SR more than you is likely better at the game (not always , if they just don't play enough to drop).

But, if you stomp 3 in a row and your are hard carrying your team. You do not face harder opponents...you face the same SR level of opponents only YOUR TEAM gets WEAKER.

Now this isn't the system out to get you. This is simply the system TESTING YOU and YOUR POTATO TEAMMATES. If YOU win then your MMR goes up and you start seeing a new pool of players. If your teammates do poorly their MMR probably still goes down even if their SR goes up (or their MMR goes nowhere). Then these people come to the forum complaining about <20 SR on a win, and make up conspiracies re: tanks and supports.

Now if YOU LOSE that game (or couple games) your MMR probably goes down and you still get a new pool of people to be tested with AGAIN.

This system works well testing skill. It is roughly accurate in a way that is quite cool considering the complexity of the game.

But it is this testing phase that pisses people off. People don't like losing. And they hate being made an underdog because they think it is forced losses. They think of each game as super important and don't think in batches of games changing their MMR and SR just roughly reflects their MMR. Which is how some with > 50 % win rate drop and others with < 50% winrate climb.

My point is that if you consistently TRY YOUR BEST. You will get to a roughly accurate SR. You will WIN and you will LOSE. But eventually you will get to a point that you don't really get long streaks and that is YOUR SR. Until you IMPROVE some aspect of your play.

I was a zarya one trick in season 2. Dropped from 3000 to 2500 in season 2 after her nerf. Decided to learn more heroes, picked up pharah, winston and Ana.
I climbed from 2500 range to 3200 rangein season 3. And stayed in 3000s for about 3 season now. Maybe Diamond is my skill cap. So I'm done with competitive mode on this account for now. Just placing every season for the rewards.
I made an alt acount last season too, practiced and played more heroes but only flex playstyle. Climbed from 2700 range to 2900 range.
And I can tell you 500 SR+ apart you WILL notice a lot of differences for sure....
200 ish maybe not but still noticeable. I agree that the variance in skill level between 2 players even of exact same SR can be huge, but you are missing one important thing in your counter.

How does the matchmaker sort "potatoes" to JUST you when EVERYONE is using the same MMR and matchmaker system?
Enemy team is JUST as likely to get potatoe team mates as you.
The match maker only make matches, it does NOT test you or anything like that.
It may APPEAR to when you get bad team mates by bad luck.
But for purpose of this discussion, please understand MMR and matchmaker only make matches. It does not setup specific players to lose.

10/11/2017 04:59 PMPosted by Aloris
I already made a giant thread going over probability and how only after 10 games for 5 million (or whatever players) we expect so many to have a giant loss streak...

and that's only after 10 games. That's not after hundreds, thousands of games.

But, they want Blizzard to defy probability.

I actally want them to make a "forced 50% win rate" mode for April fools.
And let people try what it is actually like playing in that mode.
Whenever you are over 50% win rate, your next match will be 6 full human enemy teams, and just have 5 giant potatoes sitting in your spawn on your team.
You have to fight that 1 v 6 and try to climb.
Maybe a leaderboard of somekind for people who rated highest in this mode.
Not too sure what you are suggesting here, so are you saying instead of forced "50%", the system is still somehow against you and putting you at sub 50% win rate?

You hit a wall and consistently get losing streaks at a certain point?

I don't know what to tell you.... despite all I've said I'm sure you understand that the matchmaker ONLY just make matches? It does not have any influence over wins and losses.

Perhaps it's better to just take a break when you have 2 losses in a row, and start playing again when you are refreshed? When people are tilted they don't make the right decisions and they won't perform as well as usual. Perhaps it's something to do with that.

I'm saying the 50% is a myth, because I have had no such experience with the game. Is that really hard to comprehend?

And yeah, I make steady progress - usually gaining 3-400SR, then for whatever reason hit a wall and lose the next 12-15 matches and lose it all. This has been entirely consistent behaviour for over 12 months. It doesn't even surprise me any more when it happens because it's such a predictable cycle.

I've tried taking breaks, I stopped playing competitive as much, if at all on a weekend, but this just shifted the streaks to a week night. I tried changing heroes up, adjusting my hours. But still, the cycle continues. It doesn't even tilt me anymore, it's just something I have to play through and hope it's less bad and my progress after is better - but it never really seems to be.

Join the Conversation

Return to Forum