Supports and Skill Rating

Announcements
Prev 1 8 9 10 26 Next
I am so glad I am over this "I love mercy even in comp" phase. 200 hours with pitiable SR gain every win has made me loss my eagerness to play Mercy in competitive.
I am now only playing Mercy in QP because that's where I can play without seeing a plunge in my SR.
04/19/2017 05:49 AMPosted by Kallie
04/19/2017 05:42 AMPosted by Katsu
For next season - like many others have stated, I would also like to vouch for a minimum flat rate SR gain for every victory (which works in... every other FPS game?), with a little leeway added for personal performance (so that there is some incentive to not idle away in your group half-assedly) but not entirely dependable on stats, especially now that they must measure up to historical data at some vague point in time in which the system was very different to the current and has a great deal of variables to consider.

It was a nice idea to try and award purely on personal performance but it has not been implemented effectively nor fairly across the board (particularly for support as has been evidenced), it's just not realistic. You will face far less contention from players if a tried and tested system such as the above is substituted with the overly complex one we are baffled by now (and yourself by the sound of things). Please consider, enough's enough. Support deserves to be properly addressed 4 seasons in.


I agree with what you said, except I would go even further and say this has to be addressed NOW. Rather than wait for the next season.

If I was a support main, I would have stopped playing support long ago. I feel bad for the people who do stick to support, and always getting the bum end of the deal. Supports deserve just as much credit as any other hero. I cant play a proper tank if there's not healers to back me. Same applies to the DPS.


In an ideal world where major issues were treated with some actual respect and attentiveness, it would be fixed immediately so that we can get back to supporting the remainder of the season and the forums would breathe a sigh of relief as the 'support SR too low??' threads dissipate.

I painted a slightly pessimistic outlook because of their incredibly lackluster approach and spewing this "working as intended" nonsense, so I think *at best* it will be altered for next season but certainly not the current. The whole mode needs a rework, let's face it, it's undeniably corrupt on all fronts.

Also, I genuinely thank you for your empathy on the matter as a non-support player who understands the value of our impact (seemingly rare), it's very much appreciated.
First off I would like to say how much I appreciate you guys on the dev team actually addressing community concerns. It seems like something that wouldn't elicit a lot of gratitude but the way that a lot of the other developers handle this kind of thing makes what you guys at blizzard do amazing. That being said I think that in this particular issue the reasoning you provided us sounds good but is flawed in a lot of ways. I think primarily that the MMR system disadvantages supports not because of any malicious intent on your behalf but that it is probably a product of Overwatch's uniqueness. What I mean by this is that Overwatch is the first first person shooter (at least the first really big one) since Team Fortress 2 to have healing as a and crucial role. As a result, I don't think there is a ton of precedent for dealing with healers in an MMR system in this genre. Regardless of whether or not this is true comparing heroes between games makes a lot of sense for DPS players and a little sense for tank players. I don't think it makes any sense at all for a support player. The problem is something similar to the point that the Your Overwatch YouTube channel raised in a recent video. Ive played a lot of support in the past and in a slight deviation from the video I just mentioned I think that Mercy provides the best window into why comparing between games doesn't make sense for support players. This example is relatively straightforward; a Mercy player has a high number of resurrections but is having relatively little impact because even after 5 person resurrection her team is simply all killed again and again because of excellent play on the part of the enemy team. In this situation a good Overwatch player would recognize they might need to switch healers. Maybe Ana is necessary to provide the tanks with more sustain maybe a Zen is needed to counter the enemy's tanks or maybe Lucio's ability to reposition is the critical missing link. Let's say in this scenario that the enemy tanks are the problem and our good support player switches to Zenyatta puts discords exactly where they need to be every single time and harmony orbs to their best effect. They use Transcendence at exactly the right time to counter enemy ultimate combinations and as a result ends the game with a moderate amount of eliminations and healing but a massive amount of impact on the game. It seems to me that that person who did exactly what the team needs and could have been responsible for the victory in the first place will likely be penalized for the way that they play. Because of the dynamic nature of Overwatch comparing between games is not comparing apples to apples it's more like comparing blueberries to blackberries; in that it looks similar but are fundamentally different. Overwatch is so amazing because it is a team game because it invites so many different skill sets and allows players to flourish on a variety of different criteria. The matchmaking system should encourage this by placing emphasis almost entirely if not exclusively on whether or not the team won or lost because ultimately that's what each game of Overwatch boils down to.

Anecdotally I will say that as someone who likes to accommodate the team and as a result ends up playing a variety of roles from Mercy to Reinhardt to Trader I can't say it's not frustrating to watch my friend who plays almost exclusively DPS characters slowly climb away from me in the MMR. On several occasions he has gained as much as 10 or 20 rating more than me and yet freely admitted that my play had a larger impact on the game than his. Conversely there have been times where I have played Reinhardt and he Ana and I gained more rating despite the fact that my friend had a much larger impact than I did. Maybe this disparity could be solved by looking at how the matchmaking system rates healing. Unfortunately this still leaves the problem of people who switch between characters in the same rol, let alone someone who switches between roles. The key issue then becomes is it possible to measure individual impact accurately? Is the MMR capable of looking at a game and telling the difference between a sleep dart that wins the game and one that gets no value? Can I tell the difference between a solo resurrection that makes all the difference and five person that makes none at all? If the answers to those questions are no then I don't think the system is really able to rank individuals in the first place and that we are, in fact, in a sort of awkward in-between place where people are sometimes punished and sometimes rewarded for good team play and sometimes punished and sometimes rewarded for bad team play. It seems to me to make the most sense to have a purely win and loss based system that scales with the likelihood of a win or loss. This has worked very well in a lot of other first person shooters and seems to be the only way that is both fair to the largest number of people and practical.

I'm not sure that anyone will necessarily read this massive block of text but I've tried to be as rational and well-thought-out in my criticisms as I could be and if anybody braves this monolith and can see flaws in my reasoning I would be happy to hear them. Perhaps there is some part of the MMR system that I am unaware of. With the information available though I think that this is a very comprehensive argument against the current MMR system.
Raw numbers can't distinguish high-impact actions from regular "spam", measuring individual skill in every match is a pure Utopia. This system suffers with every balance change, average statistics become extremely inaccurate. Smurfs and trolls make it even worse.

My opinion: disable all that "hidden MMR voodoo magic" for mid-season games, keep it only for placements. Just flat +-25 SR gains, no streaks. Smurfs still will be placed close to where they should be. Much less whines for "unfair matchmaking".

P. S. "On fire meter" should be frankly renamed to "Killing spree meter"
You cant compare stats for SR gain, its not gonna work.
Tank meta ana vs Dive meta Ana.
Genjis who calls "i'll blade to get out the support ults and we win with nano visor after".
Saving sleeps for crucial targets/moments (nano blade, nano visor) counts the same as sleeping winstons?
Hit a 6 man grav & pulsebomb coming? I'll just overkill with nano to get nano assists.
The list / situation goes on. It's not gonna work like this :(
These are my last 10 competitive games.
Start 2918 -> Finish 2915
It's fair?

------------------------------
Principio -> 2918 (18/04/2017)
------------------------------
1º 2859 vs 2841
Victoria: +23 (2941)
Carta: 32%
Eliminaciones: 2
Daño: 1066
Sanacion: 16.883
Muertes: 6
Asistencias OFE: 12
Asistencias DEF: 25
Resucitados: 9
Tiempo: 15:07
------------------------------
2º 2938 vs 2915
Victoria: +23 (2964)
Carta: 28%
Eliminaciones: 3
Daño: 232
Sanacion: 16.902
Muertes: 10
Asistencias OFE: 12
Asistencias DEF: 26
Resucitados: 7
Tiempo: 16:05
------------------------------
3º 2942 vs 2948
Derrota: -26 (2938)
Carta: No
Eliminaciones: 2
Daño: 585
Sanacion: 7.881
Muertes: 11
Asistencias OFE: 7
Asistencias DEF: 16
Resucitados: 7
Tiempo: 12:40
------------------------------
4º 2918 vs 2949
Victoria: +28 (2966)
Carta: No
Eliminaciones: 4
Daño: 1262
Sanacion: 18.652
Muertes: 8
Asistencias OFE: 8
Asistencias DEF: 22
Resucitados: 19
Tiempo: 16:58
------------------------------
5º 2951 vs 2938
Victoria: +21 (2987)
Carta: 25%
Eliminaciones: 2
Daño: 420
Sanacion: 10.695
Muertes: 6
Asistencias OFE: 16
Asistencias DEF: 18
Resucitados: 6
Tiempo: 14:08
------------------------------
6º 2936 vs 2939
Derrota: -24 (2963)
Carta: 25%
Eliminaciones: 2
Daño: 664
Sanacion: 13.305
Muertes: 9
Asistencias OFE: 6
Asistencias DEF: 15
Resucitados: 9
Tiempo: 13:49
------------------------------
7º 2977 vs 2905
Victoria: +22 (2985)
Carta: 21%
Eliminaciones: 9
Daño: 1946
Sanacion: 15.845
Muertes: 13
Asistencias OFE: 9
Asistencias DEF: 30
Resucitados: 11
Tiempo: 21:19
------------------------------
8º 2978 vs 3020
Derrota: -19 (2966)
Carta: 30%
Eliminaciones: 0
Daño: 162
Sanacion: 12.764
Muertes: 5
Asistencias OFE: 6
Asistencias DEF: 13
Resucitados: 7
Tiempo: 10:32
------------------------------
9º 2964 vs 2955
Derrota: -21 (2945)
Carta: ???
Eliminaciones: ???
Daño: ???
Sanacion: ???
Muertes: ???
Asistencias OFE: ???
Asistencias DEF: ???
Resucitados: ???
Tiempo: ???
------------------------------
10º 2950 vs 2928
Derrota: -30 (2915)
Carta: No
Eliminaciones: 0
Daño: 200
Sanacion: 7.383
Muertes: 7
Asistencias OFE: 3
Asistencias DEF: 5
Resucitados: 2
Tiempo: 9:06
------------------------------
Final -> 2915 (18/04/2017)
------------------------------

Today I will play 10 games and I will present them here.
Everything indicates that Mercy does not work well points

04/19/2017 05:20 AMPosted by Necrox
Thanks Overwatch for this bad changes.
I've lost -331 points
I will not play anymore healer Mercy.
https://gyazo.com/550819c9a29bb45e9dd01ac0b216e71a
This is a terrible way to give SR. Why make it so complicated? Team wins, they get the same SR, maybe depending on the skill of the other team. Blizzard says this game is a "team" game. Also, I hope you aren't going off pure "healing" because when I'm on a good team where we are all doing our jobs, I heal less.

Just give everyone on the team the same SR and you won't have to deal with posts. There are a ton of reasons on this thread as why it's silly to do this.
I said in the start and I will say it again, Blizzard in my mind is trying to be too clever with their system and it's hurting it. badly..
The recent Your Overwatch video talks about many of my fears I have had since the beginning of comp in ow.

I would love to hear answers from Blizzard on these ideas.
Individual performance against players of a similar skill level in different matches seems like a poor way to dictate what SR someone gains/loses. For example if i play Tracer into a Winston/Mei/McCree comp vs another Tracer player (of similar SR) who plays Tracer in a match with no Winston/Mei/McCree...both of us will have a vastly different match.
04/19/2017 05:20 AMPosted by Necrox
Thanks Overwatch for this bad changes.
I've lost -331 points
I will not play anymore healer Mercy.
https://gyazo.com/550819c9a29bb45e9dd01ac0b216e71a

Here's mine!
http://pasteboard.co/6bJNvlGNK.png
I started maining Mercy at the beginning of April.

HEROES NEVER DIEEEEEE!!!!!!
So if you are going to continue spewing "working as intended" nonsense, I have a feeling myself and a lot of support main will just not bother until you actually fix it.
To all of you who say you need to play badly to manipulate the stats in order to win more points.. good luck with that! In the long run that will lose you more games and mostly likely plunge you even lower than you are now.

Also.. as if you're good enough to "inflate" your stats at your current season high SR. Let's be honest, you're struggling to even do anything useful during a match, much less try and manipulate anything.

Let's not pretend there's this massive amount of playstyles and magical moments that can't be measured by stats. Everyone does the saaaame things on the same heroes. Some just do it slightly better and are thus slightly higher rated.
and why can't we have normal ELO system like in other ranked games

be it dota or starcraft or even league or w/e

point difference between teams = more points if win or lose more if lose
if about same average give it +20 or 25 and lose just as much (scarcer at master/grandmaster)
Sorry heals, we pair you with bad players who are never ever grouped so forget getting good heal stats or those nice rezzes when rein keep shooting himself to the other side of the map while road is off somewhere in the trees.

Look how bad your stats are now, you clearly are a bad healer. Even if you win this disaster you prob should get like 3 SR gain cause you did nothing and that final push healing on soldier who nobody noticed but managed to kill them all because of it was meaningless.
I would really try to invest into machine learning and see how that improves the SR gain system. You can show the computer a bunch of really good plays (hand selected by actual people) and eventually, given a large enough pool of plays the computer will be able to decide what a good play is, and therefore be able to give more SR to actually good players. For example, imagine if a Genji kills the opposing teams supports and dies promptly to a Winston. Numerically this would result in 2 kills. However, we all know how important it is to get rid of the supports, so in reality, this is actually a very solid play. However, if a Solider 76 comes along and wipes the other 4 members of the opposing team with an ult (those team members being squishy targets such as Tracer, Sombra, Soldier, and Symmetra) this will result in 4 kills. Numerically the Solider did better, but we all know that the Soldier was able to do this because the Genji did the hard work. Please look into this Blizzard, as I think that it can really make the SR win/loss system a lot more fair.
Why can't Blizzard just make everyone on the winning team get the same amount of SR since its a team game? Maybe even get a SR boost to people who get votes at the end of the game.
I am a Lucio main, and I was happy when he got buffed. I found out you gain more SR based off your stats compared to other players. However at my SR (Low Gold) most Lucio players just leave the song on healing the entire game, so they would get more SR than me. I actually know when to ult and speed boost, but that will not raise my SR
Why do I get punished as a support if my teammate is so reckless with Rein that he charges in every 30 sec not hitting any pin but dying with 2k shield even if I hit the nade and all the shoots after it. The dps and tanks are more stat based (dmg done/dmg block) but support are basically screwed if the teammates dying to widow head shots or going in 1vs 6 because there is not enough heal that can save that tank or dps.
So the system definitely needs a rework and the current system should be used to calibrate new accounts till a certain point.

Also this current system doesn't know a lost fight like 3vs6 and I am not going to go in and charge the enemy ultimates by dying and staggering my re-spawn for that extra 150-200 dmg done.
Here's a Lucio Main:
First thing i noticed: i only become "On-Fire" when capturing the objective on Capture the Point.

Second Thing: Max SR Gains even when i'm as much on Fire as POSSIBLE!! (i say that for a reason!!!) +23
Third Thing: Min. SR Lose: -26 even tho, i do really great.

In average also when i do decent. I gain about +18 - +19 each game.
so basicly i need 2 Wins to Balance 1 Lose. a bit... unfair i guess?
...
The change to the handling of assists also affected how often a player might earn “On Fire” status during a match, but the amount a player is “On Fire” does not directly affect SR adjustments due to player performance. There is some correlation between the two because both systems are trying to measure “How well are you playing?”, but there are significant differences. The determination of being “On Fire” examines not just your own performance, but your performance relative to your teammates. The calculation of your SR adjustment after a match doesn’t look at your teammates, but instead compares you to the performance of other similarly skilled players with that hero across an enormous pool of competitive matches. So, we compare your Genji play to the play of other Genjis, Ana vs. Anas, etc. Since we’re comparing “apples to apples”, we shouldn’t see any kind of support specific bias in SR adjustments due to player performance.
...


Wait, so if a player's performance is being judged against "an enormous pool of competitive matches", then what does this mean for heroes who have received substantial balance changes? If a hero receives a hard nerf (like Ana's recent change), then are Ana players still being compared to data pools which include the pre-nerf Ana? That would make all post-nerf Ana's perform at a lower level, and dock the SR gain or loss accordingly.

If this is the case, then any nerfed character should be dropped from my personal player pool, in favor of any character who's received a direct or indirect buff. Not making this swap makes gaining SR an uphill battle.

Join the Conversation

Return to Forum