I really would like to see a fix to the ridiculous SR differences I see across roles. Nothing more ridiculous than a 2500 support player but 1500 on dps? Come on you know some roles are getting thrown into the gutters for the funzies.
I can see maybe a 4-500 spread in lower ranks but you shouldnt be high bronze as a dpser and a plat support player. You are clearly throwing on dps. You see it all the time. Just get rid of role que the game was honestly better with open que as the competitive mode.
Sr across the ladder right now is laughably deflated
Yes this. Roles should be matched. Especially when you see that your team has plat+gold dps vs. Dia+plat dps => it is gg go next. Tank role is a bit more forgiving and I do not mind running into +300 sr tank but in dps role that is pure murder.
I disagree, tank diff wins most games. I will gladly take a plat+gold dps if my tanks are dia+plat vs plat+gold tanks. They will get run over so the dps will not be able to do anything. The worst is when you get a diamond mercy one trick though, unless they are getting boosted by someone.
Why couldnt this be done for two weeks or something in the new season… Everyone agreed! On both sides!!
Keep the MMR/SR (if there is no way around it), but atleast USE it at the role level. And even if there are stacks, I am sure theres “other stacks” that match for the other team.
But it is my thoughts, team average SR is contributing to the roflstomps.
This line about MMR being an “extension” of MMR is crap. These are distinct systems which form a double standard. SR is rank. MMR is performance data used to effect handicapping in ranked competitive matches. More info:
Just wanted to link the proof of what I was feeling.
Estimated win percentage based on team average MMR. But we all know the REAL win percentage starts at the MMR of your tanks. DPS or Support after that.
Role based MMR average is my feeling where the game should at least try. Do a blind A / B test - SOMETHING. Experimental card - plz lets try it.
Just to be clear, your game is decided to go or no-go if that win percentage is lower or higher than a threshold. Therefore team balance is based on Average MMR of the Team.
Combinatorics of winning will beat stats always. You want fair orderings (ranks by collected win/loss) before you care about distances (skill diff between ranks). When lining people up against the wall to choose teams, I don’t care how much taller or shorter you are, I care who is taller, period. And I really shouldn’t be load-balancing the teams - I should be randomly shuffling people within some bracket.
So for fair laddering, you don’t cherry-pick teams around hidden “stats”, you ship within some narrow SR band (order-preserving), expanding that outwards as supply/demand is mismatched (e.g. lowpop/sparsity).
Against adversarial populations, statistical estimation using most ensemble learners/classifiers, need to be retrained with short epochs, are constantly foiled and spoofed on, or deteriorate to random anyway. They end up complicated, unexplainable, unethical, and/or corrupted.
So with the thought of OW2, I think role average MMR can apply to OW2 too right? Maybe then can apply it to OW1 for now (since we still got a year to go) just to see if it works.
untrue. I almost ONLY played support for 2 years of this games life, then I started playing tank. I only recently have been playing DPS. I am getting better, I was a plat Support. and was a plat tank. but it gave me silver DPS, and I dropped to my current bronze, while I learned the game from the DPS perspective. I know where to stand for my ranks, I know when my DPS underperforms, when my support under performs and when my tank underperforms now. and when I say that, I mean both me… and my teammates.
Cool. Couple of questions, do you sometimes get steamrolled? Or steamroll the opponents?
If yes do think one particular role group was not equal in skill to the enemy role group?
Fact: Right now the predicted chances of a win is based on the teams mmr, I simply want to make it a little more complicated by comparing role group average mmr per group and then compare win percentage per group.
Ideally even weighted average where tanks have the most influence. If the win percent is lower than 40 or higher than 60 in any group, drop the players and make a new lobby.
This can still affect OW2, and thus leveraged if done in OW1.
so far on this account, this season games have been relatively easy. I’ve said it before, I think the fact that I played a lot more QP than Comp before I started focusing only on comp… that I think I was being punished for my MMR being way different than my SR.
My DPS steamrolls are much less often on both sides. My other roles are … more different… and now I see what the importance of a tank not feeding is. if the first dead is my tank, we likely lost the whole fight. Whereas A DPS or Support can be winnable. But watch tanks go without their team is madness… BUT… I will say this… teams have to also follow the tank, which I’ll find it’s just me and a Rein when he pushes trying for a full team push. Pushing w isn’t always easy because of that reason in low ranks,.
I thank you for contributing, but do you feel the roles from both team should be evenly matched? Say if supports on opposing team are silver / gold, then the supports on your team shouldnt be bronze?
Imagine now in OW2, Gold tank goes against Bronze tank, you know how toxic that’s going to make everyone? And doesnt it kinda defeat their own system? Why make MMR if your not going to use it on a more granular level? Thats saying hey we have MMR but we dont wont to use it TOOOOO much.
If SR is a fit-for-purpose metric to pay out, score, and rank people by, it is a fit-for-purpose metric to matchmake them by. That is how you get rank integrity and fairness in one’s label, classification, and reward. The way you get fair games is simply shipping people randomly within some small SR band (relaxing that band during lowpop hours to satisfy the time constraints). That is how you get a ladder meaningful and not fake.
Anything else is rigging the contest odds. The second an MM takes your (nefariously hidden) performance assessment, analyzes it, and adapts your next match around that data and any prior history - it’s rigging. Period. Word for word by definition straight outta the dict.
There is no free lunch. If you rig matches (even towards 50/50 “fairness”) you break the ranks and delete ladder integrity. If you have a “fair” ladder (i.e. fair in that 0 rigging is present), you absolutely get stomps and streaks and matches feel terrible a lot of the time. But that’s what real competition is. Only Disney gaming would spoonfeed you fake competition through rigging behind the scenes and call the whole sham an esport.
I think SR could be a good metric for even matches… I think that it isnt because the measurement of a gold tank and support doesnt match the measurement of a gold DPS.
I also note that the game also changes difficulty as your rank number goes up. I just flipped over the number… so now Im going diamond border +1 rather than diamond border +99. The difficulty of games has always been tied to that.