A consensus of those unhappy with Ranked?

Competitive Discussion
1). It is incredibly tough to carry a level down from where you are. If you are a 2250 Gold player, you will dominate level 1200 bronze players, but at 1800 Silver games, you are for the most part along for the ride.

2). It's not the team with the best player that wins, it's the team with the worst player that loses.
In terms of variance, Plat is probably the most prominent rank.

Why?

a) Largest rank. Almost 30% of the playerbase is in platinum.
b) Graveyard of smurfs and deranked diamonds.
c) Probably the rank with most trolls/throwers.
09/20/2017 03:42 PMPosted by Protexus
In terms of variance, Plat is probably the most prominent rank.

Why?

a) Largest rank. Almost 30% of the playerbase is in platinum.
b) Graveyard of smurfs and deranked diamonds.
c) Probably the rank with most trolls/throwers.


I can't agree with this more. I currently have a 40% winrate according to Overbuff as a result of the pure cesspool of Cancer that is Platinum.
09/20/2017 03:13 PMPosted by Macbeth
I think the concept:

The lower ranks are terrible is 100% accurate.

You should be able to lose a game and not lose rank.
You should be able to lose a game and maybe gain a rank.

My basic example of how silly the ranking system is:

Tom ranks at 2300
Tom plays comp.
Tom gets a team who averages to 2200
Tom gets an opposing team that averages to 2350

Tom loses the game, which is what the system expects.
Why does Tom lose rank when he is predicted to lose?

Furthermore, if the SR gains are truely based on stats for the duration of the game, would a steamroll not result in LESS SR gains cause the other team couldnt compete than an equal match that lasted longer? This makes ZERO sense.

Now that I am in low masters i have much less complaints about the general matchmaking. When I was in platinum I called ranked RNG player.

You shouldn't have to grind to get an appropriate rank. The system should be able to sort it out much more quickly.

I do think there are many players that are worse than they think they are.


This^ does a good job of encompassing what I think is correct/a good idea.

Personal example; I know for a fact I'm much better now than I was season 2. That said, the general playerbase has also improved over time. However, in the last two months, I have had friends join me in comp, ensuring the quality of at least a few teammates. We have climbed and continue to do so. This season we placed in Gold/Plat, having climbed out of low gold and even silver (one was new).

During our placements, we were queued with mostly plats, a few diamonds, a few golds; had 8/10 wins. When doing qp, we get queued with much the same. However, Currently slogging our way out of the 2400 range is something of a nightmare.

In climbing, we've found that there are a few specific SR ranges that make climbing extra difficult, typically the 100 SR just below a new rank. Once ~50 past the threshold, it's fairly easy to ascend until at those ranges again.
09/20/2017 03:23 PMPosted by mrjman
09/20/2017 03:13 PMPosted by Macbeth
I think the concept:

The lower ranks are terrible is 100% accurate.

You should be able to lose a game and not lose rank.
You should be able to lose a game and maybe gain a rank.

My basic example of how silly the ranking system is:

Tom ranks at 2300
Tom plays comp.
Tom gets a team who averages to 2200
Tom gets an opposing team that averages to 2350

Tom loses the game, which is what the system expects.
Why does Tom lose rank when he is predicted to lose?

Furthermore, if the SR gains are truely based on stats for the duration of the game, would a steamroll not result in LESS SR gains cause the other team couldnt compete than an equal match that lasted longer? This makes ZERO sense.

Now that I am in low masters i have much less complaints about the general matchmaking. When I was in platinum I called ranked RNG player.

You shouldn't have to grind to get an appropriate rank. The system should be able to sort it out much more quickly.

I do think there are many players that are worse than they think they are.


Its really pretty simple matchmaking is built around probabilities. In your example Tom had a high probability to lose so he would lose less SR since it was somewhat expected. However in the reverse case Tom will gain more SR for the win because it was not expected. This is a win/win as it puts the onus on the better team to perform up to par.

The performance part of SR gains are based on stats per minute so it doesn't matter how long the game lasts and it doesn't matter if you switch characters if you were performing well while you were on a character.

Climbing should not be easy and thus will feel like a "grind" to some. But to me its more about putting in your time to get good. The normal person will gradually improve at something so if everybody is gradually improving you have to find a way to improve faster than the general population otherwise you will be stuck and it will feel like a real grind. Pretty logical if you ask me.
I was stuck in bronze for a while and broke silver end of season 4. I spent way less time in silver than I did in bronze. I blew through silver in like a week.

This account is stuck in low gold. My main account is pushing plat and its WAY easier winning in high gold than low. PLease explain that one
09/20/2017 03:46 PMPosted by tacobellftw
09/20/2017 03:23 PMPosted by mrjman
...

Its really pretty simple matchmaking is built around probabilities. In your example Tom had a high probability to lose so he would lose less SR since it was somewhat expected. However in the reverse case Tom will gain more SR for the win because it was not expected. This is a win/win as it puts the onus on the better team to perform up to par.

The performance part of SR gains are based on stats per minute so it doesn't matter how long the game lasts and it doesn't matter if you switch characters if you were performing well while you were on a character.

Climbing should not be easy and thus will feel like a "grind" to some. But to me its more about putting in your time to get good. The normal person will gradually improve at something so if everybody is gradually improving you have to find a way to improve faster than the general population otherwise you will be stuck and it will feel like a real grind. Pretty logical if you ask me.
I was stuck in bronze for a while and broke silver end of season 4. I spent way less time in silver than I did in bronze. I blew through silver in like a week.

This account is stuck in low gold. My main account is pushing plat and its WAY easier winning in high gold than low. PLease explain that one


There are tons of variables at play so it could be lots of different things causing short term stagnation. It could certainly be your mindset and subconscious bias of your main account being stuck causing some of the problems. It could just be that your sitting in a small valley right now and will pick back up shortly to climbing again. Low gold and High gold are very close in terms of the skill of play and are well within the margin of error of the ranking system. To me being within the same range on multiple accounts means that the SR system has done a good job of figuring out the type of skill you have.
1. Separate solo queue and group queue rankings. And match maker use these ranking to match the queues respectively.

2. Hero preference selection before queueing. Match maker actively try to form a 6 man team in similar SR that covers most of the hero pool based on those preferences. When hovering over player list, their preferences are listed rather than most played.

3. AFK check pop up BEFORE game starts when matches were just made.

4. "Avoid this player" on the same team. Not possible to avoid opposing team players.

5. Honor/karma system. Matchmaker prioritize matching good honor/karma people together.

I'd play the crap out of comp if these are in place. Right now I have 0 faith in the match making system putting me in a good match.
<span class="truncated">...</span>

Let me give an example of what I'm experiencing in my bronze play. A player by the name of GodShane was on the opposite team destroying my team. I decided to check his profile on what he has. His SR was at least 3200 (for a season high might i add) i don't remember the exact number but roughly around 2800-3200 SR. That is what bronze players are having to deal with because of the golden weapon skins and because people like to ruin the fun for others. Now i agree with the thread creator. Blizzard needs to handle this quickly. I agree because of the personal experiences ive had. and also with the other players on the enemy team being around 2000-2600 SR in skill level.


Must be an act of god that only your team has the disadvantage of facing far superior opponents. However much more likely and logical is that these players are split about even between your team and the enemies but your brain is trained to remember outliers that feel bad much better than the other.

Try not to fall into the blame universe camp. Its not good for anyone, most of all you.


funny you should mention that before our round started I had checked my teammates highs overall and current season. i am at 1660 of my overall high while everyone overall high was around 1000-1800 and for the matter that we were paired up with people who had a much larger number in the overall i fail to see how that logical in the sense because no one was that skilled on our team and we tried our damnest (and
09/20/2017 03:13 PMPosted by Macbeth
I think the concept:

The lower ranks are terrible is 100% accurate.

You should be able to lose a game and not lose rank.
You should be able to lose a game and maybe gain a rank.

My basic example of how silly the ranking system is:

Tom ranks at 2300
Tom plays comp.
Tom gets a team who averages to 2200
Tom gets an opposing team that averages to 2350

Tom loses the game, which is what the system expects.
Why does Tom lose rank when he is predicted to lose?

Furthermore, if the SR gains are truely based on stats for the duration of the game, would a steamroll not result in LESS SR gains cause the other team couldnt compete than an equal match that lasted longer? This makes ZERO sense.

Now that I am in low masters i have much less complaints about the general matchmaking. When I was in platinum I called ranked RNG player.

You shouldn't have to grind to get an appropriate rank. The system should be able to sort it out much more quickly.

I do think there are many players that are worse than they think they are.

This is a great example of what i had said earlier in my reply to this thread about someone who OVerall high was around 3200
People get stuck in diamond and lower because people bring them down when they are trying to climb back to master like me for example.... its hard to carry when youre against competent opponents and when your teammates are pesimistic and toxic and saying its gg when its honestly not, and to add further SR is not an accurate measure for ranking....
So far I've been following this tread, 99% off the reply's are intelligent and constructive feedback.

Also i've noticed an increase of these kinds of treads about the same topic in one way or the other. This tells me its a real problem (?).

I really hope that jeff reads this thread and gives us some feedback, maybe an assurance that the team is working on fixing some off the issues raised.

Reworking the report system and actually taking action vs toxic players is a big step in the right direction. That said i believe implementing some of the idea's and addressing some off the issues would result in less toxic players because then there would be 1 less reason to become frustrated.

Jeff bro help us out!
09/20/2017 12:24 PMPosted by Porkypine
09/20/2017 11:47 AMPosted by combofied
I'm pretty happy with ranked. Now that Dva is nerfed there is no direct meta besides mercy, you can play whatever you want and do good with it. It's pretty fun.


You're GM your perception of 1% of the games players is hideously incomplete.

Of course you're happy, you play in the games top where evolutionary pressure is soo high that the stupid inexperience based play you see down here in wood tier just won't fly.


"You aren't allowed to have an opinion because you play in the top 1%!1!1!, You don't have issues you don't know what we normal players go through, wahh wahh!". Thanks for reminding me why I don't come onto these forums very often anymore.

Newsflash: GM isn't pretty either, and there is definitely people that don't belong here such as the boosted onetricks and such, it actually doesn't take much to get to gm when you can win a game and gain 30 and then lose a game and lose 10 as a onetricks. You don't have experience at gm, so don't talk like you know what happens. I've been in every rank other than bronze so I can speak.
I have a friend who stopped playing and gave me his 1516 sr account. By the end of the afternoon it was 3200 and the next morning 3600. My main is 37xx. Seems to be working well.
09/20/2017 12:01 AMPosted by SumNakdDude
I want to get an opinion from all those unhappy with ranked.

Does everyone agree on this one point? Can we speak with one voice?

The main problem is that low ranks are super unstable. What I mean is that the lower the rank, the more wildly varying the game is and the more the skill levels are crazy unequal. High ranks seem like they may stabilize and represent something real, but low ranks, where most people are, have almost no reliability.

This leads to way more stomps and completely unpredictable team strategies because people's skills and their SR don't match nearly as well as at higher ranks.

It seems like people at the top don't understand the problem at the bottom because the system actually works better and better as you get to higher ranks. You are playing with people who actually belong where their SR says they do at high ranks, but at low ranks you have people from GM to "horribronze," all in the same games and it becomes a random mess you can't really influence.

How can you "carry" as a gold skill level player in bronze when there is a plat skill level player doing the same on the other team? What about when you belong in diamond your team is full of people who belong in silver and gold and there's one Master with a whole team that belongs in horribronze. Sorry, but the diamond is going to beat the master. Period.

TL;DR The vast majority of people who are ranked improperly because of low total game numbers are in gold, silver, and bronze. This makes the area where most people's skill level actually belong, be a chaotic mess of unfairness and a bad experience.

So can we all agree there is one single major problem that all of us (who are unhappy) are complaining about?

The problem is instability, variation, and unreliability of the area where most people play

============================================================
Now let's deal with the argument everyone happy with the system uses:

"But so-and-so climbed out of bronze, so there's no problem"

Unfortunately people keep citing bronze-GM series without considering some of the data around it. Tanking your win rate by screwing over team after team and getting down to bronze gives you a HUGE starting advantage... not only do you get placed with other people who are losing and losing, you also get an enormous rank boost advantage.

People who tank their scores are actually gaming the system and I've been considering doing it myself just to slingshot out of silver.

If you start with a brand new account, you're also starting with an advantage of not having the system believe it knows you and therefore try to hold you back to "where you belong."

Look at this data analysis of that crazy one-off series Stevo did. (why do people always cite edge cases when we're talking about a rule-of-thumb??)

https://docs.google.com/spreadsheets/d/1FLEzkX4AMFsO2tU2kNOxyqO4SCJ2kDYI8qa0QGk4V9A/edit#gid=0

There is a clear slingshot effect up to 1900 SR. The average SR gains and losses after that point stabilize to the 20-25 region afterwards, but I've personally noticed a huge shift in game play style and skill just past the 1700-1800 mark. I have to switch from Junkrat carry to Soldier right in that area. Now I'm wondering if there isn't a very specific reason for that transition point.

This shows that these examples are not applicable to most players. They are irrelevant.

Yes a GM player can get out of bronze in quick time... under the right circumstances... in the examples they show off... (not the ones they hide) sure, but what about a silver or gold player? Can they just jump right out of it or do they slog through a hellish up and down that goes nowhere and feels way too crazily varying to have any meaning? Is there a rewarding experience of reliable feedback of personal performance?

Is it fair to ask players to play 300 horrible garbage games to finally reach a place where the game is actually representing your real skill?

Our problem is slogging through. Slowness and crazy variation. We're not complaining that it's impossible. That is NOT the argument, so please stop "proving" that straw-man is incorrect. We don't care because we already agree with that.

TL;DR Exceptional people can do exceptional things. Some people put in more time than others and deserve more reward for it. Fine, great, sure. Now let's address the rest of us.

============================================================
What's an actual solution they could implement to make those unhappy... happy?

(what would solve the huge variation)

1) Something like the TrueSkill Through Time (TTT) mentioned in this paper:

http://www.ets.org/Media/Research/pdf/RM-15-03.pdf

By storing game outcomes and points awarded the system can go back and re-process games based on later outcomes to get more realistic scores with fewer games.

If you were playing in bronze with someone who ended up in GM, then whatever score they gave you was inaccurate because you weren't playing a bronze player.

This system fixes that!

2) Allow a player controlled manual reset of all historical game data on an account at least once per season. If I played only ten games per season in all previous seasons, the system is sure I now belong deep in bronze and will not allow me to climb quickly because it's using biased data.

Clearing your record and being able to start over is CRUCIAL for fairness.


Make a minimum number of games per season to be eligible for the season reward. Forcing people to play more games and get ranked properly.

Also, stop using season highs. But final scores.
This isnt a real problem the only issue is people dont know how to lose. If your in silver/gold or lower thats where you deserve to be, i got up to plat and noticed the difference in playstyle compared to silver and i can assure you that if your in silver your that bad and nothing you say or do will change it unless you actually get good.

Learn to lose because theres no participation trophies here.
09/20/2017 07:32 PMPosted by Alias
I have a friend who stopped playing and gave me his 1516 sr account. By the end of the afternoon it was 3200 and the next morning 3600. My main is 37xx. Seems to be working well.


So.... 1700 SR gain in a single afternoon huh? Are you sure you want to stick with that claim?

Lets see, if you won every single game at a generous average of 25 points per game that would be a minimum of 68 games. Even with stevo's crazy unheard-of-one-time-only 51 game win streak in the mix it took him 97 games to get from 1500 to 3200.

stevo has about 86.5 hours of videos and if we assume there is no editing, he played 297 games in that time for an average of about .29 hrs (17 minutes) average per game.

97 x 17 minute average = 1649 minutes (27.5 hours)

So you're saying you accomplished in a few hours what took stevo about 27.5 hours including one of the biggest win streaks ever recorded....

Maybe you're not bold-faced lying.

Perhaps the problem is that we are regularly getting feedback from people like you who think like people with a gambling problem. They can't remember the real past and always tell everyone a ridiculous story of their wins because that's all they remember.
09/20/2017 08:02 PMPosted by Ruvix
This isnt a real problem the only issue is people dont know how to lose. If your in silver/gold or lower thats where you deserve to be, i got up to plat and noticed the difference in playstyle compared to silver and i can assure you that if your in silver your that bad and nothing you say or do will change it unless you actually get good.

Learn to lose because theres no participation trophies here.
Here is the thing. If I am able to have a higher win rate in a higher bracket shouldnt that make it frustrating to lose in a lower bracket?

If I truly belong in low gold shouldn't I be dead weight pushing plat on another account? Look everyone has bad games where their teammates have to carry them. But for the most part in those higher SR games I am able to carry my weight just fine and quite often card. Which tells me that I am not jdeadweight in those higher gold SR games.

I am going to do an experiment the next 2 weeks. I am going to try to play say 30 matches in both accounts which is about 4 games a day. I am very curious to see what the SR and win rate is relative to each other. But as long as I can remember playing this game the account with the higher SR is hands down easier for me to win in which let's just be honest with ourselves makes no logical sense.

In warcraft arenas if you told someone that it was easier to win at 1600 rating (low plat in OW terms) than it is to win at 1200 rating they would laugh at your face.
Hi, thanks for the response. I'm a programmer too btw. :-)

09/20/2017 08:57 AMPosted by SumNakdDude
you'd suddenly have horrible huge SR losses and almost nothing gained, or the opposite after each calculation.
Some threads in this forum actually sound like that, but I agree it's improbable they do TTT already to any large extend.

09/20/2017 08:57 AMPosted by SumNakdDude
2) Which is more destabilizing? Players placed improperly (too high or low) with the system certain it's right and placing additional people based on this false information... or entities the system knows are currently unknown and therefore used as a less reliable guide when placing other players who interact?
I see, you expect the improper placements to be corrected faster by the resets. I think it is much more important, that their rank converges to the correct value (which it doesn't in the current system).

So on the point of resets I think we can agree that they will not fix anything by themselves in the current system, would not destabilize a good system too much and resets may improve the speed of convergence in a good system.

09/20/2017 08:57 AMPosted by SumNakdDude
Unfortunately I think the per-group idea is infeasible. Every ad hoc group that forms is a new entity to store and try to score. I'm not sure I understand how you can argue that a data reset would mess things up and then immediately propose constant new entities being put into the system.
Yes ad-hoc groups are at a disadvantage. Maybe the "stay as group" button would need to be removed as nobody wants 10 placement matches for a random group that never meets again. The solo SRs and more permanent groups SRs stay untouched by these new entities though and should be a good anchor.

09/20/2017 08:57 AMPosted by SumNakdDude
I see your response to Tankz and it makes even less sense to me. You say to calculate it like it is now. You've reached a mutual exclusivity. You cannot have groups be treated like new individuals and then ask that they be calculated from individual scores like they already are. That's asking for opposite things to exist simultaneously in the same place.
This is only for the initialization of course. It should be better than starting the new group as as a complete unknown. In the current system the solo SR is modified based on group performance as it's the only score. In my system there is no influence except for the initial SR value of a new group being calculated from the solo SRs. I don't think that's a contradiction. Whether it will be good enough to have no placement matches for new groups at all is not really predictable now.

09/20/2017 08:57 AMPosted by SumNakdDude
What Jake Lyon asks for in that vid is to have two totally different scores. How you play in parties and how you play solo. That's something different from what you have proposed. What he proposed is workable but I don't think your suggestion actually is possible from how I understand it.

But you do see the basic problem I describe? Different groups containing the same player can have extremely different strength and thus a rating that keeps all groups together (including solo or not) will depend a lot on the players group choice, which it really shouldn't if it was meaningful in any way.
Simplest example are 3 players of the same skill. 2 are one tricks of hero A, the third one tricks hero B, they are bad at all other heroes. Noone plays solo, only duo groups. Now the ranks of the hero A players depends mostly on how often they play with each other. They will rank higher if they play more often with the hero B player. This should not happen as they have the same skill.
I have not seen any suggestion that would solve this problem except mine, where each of these groups will have an independent score. Would love to hear your thoughts :)
09/20/2017 08:14 PMPosted by SumNakdDude
09/20/2017 07:32 PMPosted by Alias
I have a friend who stopped playing and gave me his 1516 sr account. By the end of the afternoon it was 3200 and the next morning 3600. My main is 37xx. Seems to be working well.


So.... 1700 SR gain in a single afternoon huh? Are you sure you want to stick with that claim?

Lets see, if you won every single game at a generous average of 25 points per game that would be a minimum of 68 games. Even with stevo's crazy unheard-of-one-time-only 51 game win streak in the mix it took him 97 games to get from 1500 to 3200.

stevo has about 86.5 hours of videos and if we assume there is no editing, he played 297 games in that time for an average of about .29 hrs (17 minutes) average per game.

97 x 17 minute average = 1649 minutes (27.5 hours)

So you're saying you accomplished in a few hours what took stevo about 27.5 hours including one of the biggest win streaks ever recorded....

Maybe you're not bold-faced lying.

Perhaps the problem is that we are regularly getting feedback from people like you who think like people with a gambling problem. They can't remember the real past and always tell everyone a ridiculous story of their wins because that's all they remember.


Stevo gained about 15 per win and lost 30 per lose because he played Symmetra, which resulted in it taking an abnormal amount of time to rank up.

This guy could have played any other hero or an abormal Sr gain hero in ranked up in much much shorter time. I played in plat and was gaining like 33 per win as doomfist and 29 per win as mcree each game, and even more once streaks started (at least I think that's what it was).
My main problem is how much of a gamble queuing into each game is.
You can queue into a game where you have 5 dps players, or 4 mercy mains, or 3 winston 1 tricks.

And that is just classes. The worst part, by far, is the throwers, trolls and people screaming abuse down the mic, which is honestly a very common occurrence.
It may be my rank (plat) bu the amount of this kind of behavious I encounter in comp is ridiculous.

I feel like more often than not, games are decided by which team is unfortunate enough to get the troll, thrower or rager on their team.

Join the Conversation

Return to Forum