They are still using your oldest data against you!

General Discussion
1

New game. New platform. Old data.

I thought that getting my skins from OW1 wouldn’t screw my account since I switched from PC to Console. Wow, was I wrong.

I started playing all the way back with OW1 beta. I’m a filthy casual and play every ranked game with my wife who, honestly, plays super inconsistently and not that great. She has other talents… Very few games per season locked us down into low ranks. I proved this with a video 3 years ago by ranking 1000 points different across three accounts in one play session. (youtu.be/jyWPuifJg1k)

I even showed a buddy how playing the same games with him and performing the same (we kept track and played the same 2 characters back and forth) advanced his rank much faster than mine because of my bad data Blizzard was holding against me.

Now in season 2, my record in support is 25 wins 13 losses. (with a similar performance first season.)

Guess what my rank is.

…
…
…
…
…
…

Bronze 1 - But look at that huge amount of games that Should Not Be Relevant.
i.imgur[dot]com/jOC1qCt.png

Damage 6:1 = Silver 1 (I think I did better first season but overbuff doesn’t have it)

Tank? 10:12 isn’t even that awful (definitely did better 1st season) In Bronze, every “healer” plays Ana, Moira, or Baptiste to solo DPS and every actual DPS flanks the whole time (badly) so there’s never follow up to a push and you’re always abandoned so… yeah good luck playing tank in deep bronze when randomly queued. Better luck winning at roulette… the Russian kind. (I won 12 in a row first season as soon as I grouped with a team that had voice)

You think higher ranks are inconsistent? Try playing against smurfing new accounts with children and newbs the matchmaker has no clue what to do with. It’s literally hopeless.

I’m old enough to be most players’ father (and unfortunately grandfather to some of these children literally stuck on walls in bronze) So my reaction time and fine motor control isn’t top tier, but my positioning, prediction, call outs, game comprehension, and overall teamplay are, at least, MUCH better than average.

Most of you who haven’t played accounts between bronze and plat don’t know what I do: The game environment is so different between ranks it requires big playstyle changes. (winning with chaotic individuals is different from coordinating with a team)

The auto-aim on PS4 does make up for my age but me and many of you are screwed by old data… PC Data at that!!!

I thought a change in game AND platform would finally free me. Nope! They screwed me as soon as I linked accounts to get my skins.

4 Likes
2

I don’t play ranked for that reason. Despite being much better at the roles than when I last played rank (I think season 7 or 8), I would still be hardcapped at that rank because I have been playing the game for so long. That being said I had gotten into diamond (this was before role queue but I played mostly reaper for dps and Zen/ana for support) and grinded my way through plat to reach diamond and then stopped because the grind was too stressful for me (esp on 2cp maps).

1 Like
3

Yeah, their system requires giant gluts of data to figure anything out because they are so philosophically committed to only ranking based on team win or loss.

I get it because team coordination really is what separates mediocre players from good ones when reaction time and technical play is similar but at low ranks they need to separate people faster. (it’s NOT what separates mediocre play from bad play)

I’m a programmer and I understand their analytics problem, but the solution really is just a few tweaks away. There’s a huge technical, game knowledge, and teamplay difference at low ranks and it’s mostly just all teamplay at the top.

They just need to to weight it utterly opposite (technical first) at the bottom from the top to stabilize the system instead of trying to use huge aggregates to stabilize.

It’s just bad analytics theory combined with philosophical commitment. They have a cooperation agenda I agree with but they are applying it too strongly and broadly for real world.

===
Just for a quick example here’s my Boy Lucio this season: (i.imgur[dot]com/bfZBt6j.png)

  • 20 wins 8 Losses (Top 36%)
  • KDA 3.9 (Top 12%)
  • Objective Time: 01:53/10 min (Top 37%)
  • Healing: 8,163/10 min (Top 34%)
  • Crit Accuracy: 9% (Top 19%)
  • Weapon Accuracy: (Bottom 19%) …nevah stop shooting!

Well, how about we look at my old PC play on Junkrat: (i.imgur[.]com/xfaNuu5.png)

  • 106 wins to 98 Losses (plenty of data)
  • Eliminations: 20.51/10 min (Top 11%)
  • Objective Kills: 7.79/10 min (Top 13%)
  • Solo Kills: 6.10/10 min (Top 3%)
  • Final Blows: 14.71/10 min (Top 6%)

Does that seem bronze to you?

Yes… TOP 3 PERCENT. Artificially inflated by being stuck down with newbs by a broken system. (I’m not that good. The MMR system is just that bad at ranking)

2 Likes
4

Then just stop playing if you don’t like the game.

5

I have a friend in a similar predicament.
He is just hardstuck bronze despite the fact that he deserves at least Gold. The matchmaker, due to the fact that he was bronze for a while on release, says otherwise.

This is why an MMR reset was kinda needed.

6

One who fears loss is already defeated.

1 Like
7

Impact is the key. These stats which seems good doesn’t measure your impact on the game. All elims are counted even if just 1 bullet touches the enemy and healing has little significance to rank. I can imagine many in bronze to have bloated stats because the competition is all over the place.

Lucio for example, do you beat to save your team from losing a team fight, do you beat to save them from a team wipe ult, do you speed boost to secure kills, do you boop people out of position to reduce their advantage. Most of Lucio’s kit is geared towards team play and if he is played as an individual you are leaving most of his impact out of your games.

8

I don’t know what your point is. You don’t know how the ranked system works?

That sounds about on point. You’ve haven’t even played 40 games. What the system does is put you much lower ranked than where you “should be,” so that you can grind up.

I think win/loss is actually one of the things they look the LEAST at. They mostly look at how well you perform. I actually won about 33% of my games last series, and I ranked up still. I went 7-15.

I don’t know where you are getting your figures from…overbuff? If so, those figures are not even remotely close to correct. No one has access to those figures except for blizzard.

And even if they were correct, getting 20 elims/minute in bronze doesn’t make you on par with someone that is getting 20 elims/minute in GM.

It’s funny because that’s what they do? Or at least what they did in OW1. They had something called PBSR (personal based skill rating), which would adjust your SR gains/losses depending on how well you performed individually. That was in effect for everyone under diamond. But in diamond and above, PBSR was no longer used. I don’t know if they have something similar for OW2.

I am often amazed by how frqeuently I see people saying “their mm system sucks they should change it to this,” and their suggestion is actually what is already happening. People don’t understand how it works so they just assume? I guess. It’s kind of funny though.

9

I totally get that and I 100% agree. As for support, I always play selfless and do exactly the things you’re talking about.

The problem is that they weight the win/loss + history about 95% and personal stats about 5%. I understand that and agree with it for high rank!!!

The issue is that low rank teamplay doesn’t sway the game nearly as much as personal skills because everyone (except smurfs) is so bad at all of it. There is a huge difference in skill level and game knowledge swaying the game too much to rely on win/loss and team effects.

At high rank there’s only so good you can be a technicals like aim and generally FPS stuff so teamplay timing and positioning is everything.

At the bottom everyone is garbage at those things AND at the technicals so your main contribution is actually just your technical FPS gameplay since playing coordinated (for DPS/Tank) when everyone else is playing individual DOESN’T WORK.

You’ve got to make stupid high-risk-high-reward plays at low rank to sway the game instead of self sacrificing “good” plays. I’ve played low ranks and plat…

Playing right for plat is playing wrong for bronze. Especially Tanking. (Well, Okay, not so different for support, actually)

The issue is that you’re not understanding analytical systems that must use historical aggregates. (IE They have to use previous rank to decide future rank)

I don’t and can’t grind so therefore I’ve always matched below my level. That’s bad design.

No. Sorry but that’s completely false. It’s very nearly 100% win/loss to capture your team-play contribution instead of individual play that can look good on paper but be counter-productive just like Goldark mentioned.

And I agreed. The point was that I’m consistently shooting fish in a barrel because the system is borked.

A tiny contribution to the adjustment that was minor at first and so small as to be irrelevant later on. I know through not just what Blizzard has said but through extensive testing.

You’re just not thinking it through.

If you believe they make you grind and I’m not putting in enough games…
If you see that my incredibly high performance is just an artifact of shooting fish in a barrel…

Then you agree with me. You probably just don’t understand historical weighting and Bayesian priors and that’s what is tricking you into thinking there is strong personal performance influence.

If you watched my old video, the greatest single contributor is where they think you “should” be based on your old data. (because they need and use tons of aggregate data to, theoretically, separate out your tiny effect from the team win/loss)

You were significantly below your historical rank, right? That was the reason you ranked up. The grind, not personal performance.

That’s my whole gripe and if you think about it I think you’ll agree.

10

Um. Why would I not understand that? Do you think that I thought they just reset everyone every season?

If you don’t play the game a lot, how do you expect the system to properly rank you? What exactly is a magical “good design?” If they can’t use your data from playing the game?

Um. I’ve had many seasons with negative win loss ratios and gained rank. So…you’re wrong. By the way, just saying, “it’s false” doesn’t make it false.

Or it’s the fact that despite losing most games, I’ve literally witnessed an increase in rank over multiple seasons, purely due to outplaying my opponents and teammates.

I was below my historical rank, yes. Just like every single person in OW2. That is what they do, they rank you lower. I have told you this.

Since you seem to be having a very hard time understanding their system, I suggest you read their latest developer post about matchmaking. It (kind of) explains how it works. At least, it might help you to get a better idea of what is going on.

Overwatch 2 developer blog: Matchmaker and competitive deep dive, part 1 - General Discussion - Overwatch Forums (blizzard.com)

11

I suggest you don’t overestimate your comprehension of a subject or underestimate others comprehension of it.

I read it and understand it from a perspective I don’t think you can conceive of. As a developer myself I know the design constraints and theories around ranking. (and I know what they gloss over and hide and why they hide it) I know the actual requirements of the mathematics and programming and how there are fundamental design issues with trying to suss out small effects from larger ones.

Google this: “Elo vs. Regression to the Mean: A Theoretical Comparison” for just one perspective I think will expand your horizons

You’re being tricked by their terminology like “your skill” when they mean the data they have about your performance that comes from mostly win/loss that is heavily adjusted and weighted by past aggregate ranks. (a feedback loop of doom)

I know they have to use some past data. The issue I have is about weighting and application of specific design metrics and theory. Something the average person knows literally nothing about and shouldn’t have to know. (If you haven’t heard of “TrueSkill” for instance, then you haven’t been doing any academic research and should adjust your self-image about your authority to speak about this topic accordingly)

There’s nothing shameful about not being a nerd about academic topics… so please don’t get offended! :nerd_face: You don’t have to know everything in the universe.

I’ve outlined a system however, in my post here: How to Make Lower Ranks more Fun

Just read between the lines, friend. Everything they say is “skill” is actually just your win/loss rate. Here is a quote from that article and pay attention to what they say:

When a player wins more than 50% of their matches, we start to increase their MMR faster. This levels out once we see the player’s MMR value has them joining fair matches, which corresponds to a 50% win-rate. With this change, we’ve seen that new and returning players reach a 50% win-rate with much fewer matches than before.

Wins. Win-rate. IE it’s not actual personal metrics. They don’t use personal performance hardly at all. When they talk about your personal performance they MEAN your win/loss rate.

They fundamentally believe your skill and your win rate are the same thing so they use the terms that way.

They have an ELO philosophy for a team game and that is a scientifically untenable theory for individual ranking design in a random team game. (please read the article I told you to google) They are fooled by the fact that stable teams whose members do not change CAN use ELO to treat the team as a single group.

I’m not trying to be mean here but you just do not yet understand the underlying theory and design elements I’m addressing.

That’s not an accusation or a dig. It’s reasonable. You shouldn’t have to be an academic expert on a topic to just play and understand your entertainment. There is no possible shade on you here.

1 Like
12

I’m not doing either one of those things. It is very clear that you don’t understand how it works.

Oh good. Then it should be very easy for you to understand how it works. Yet, you seem to struggle.

That’s not correct. I’m sorry.

At least, based on personal experience, and how my rank/SR was adjusted in OW1. I don’t really know how it works in OW2, since they basically hide EVERYTHING.

I’m glad you are not ashamed of your ignorance.

Is there some part of your brain that is glossing over my statements? I have LOST MORE GAMES THAN I HAVE WON, AND HIT SEASON HIGHS IN RANKS. READ THOSE WORDS PLEASE. Do you just have some kind of selective reading disorder?

13

so op complains that because of limited sample size, the mmr is rated a certain way and now hed have to grind to prove the system where he actually would be.

thats just the limits of the system. if you want the system to rank you correctly, play 50-100 games solo. playing with other ppl screws with mm and mmr, because as you said, the game with your wife are very low rated bc shes not playing that well.

you brought this upon yourself. id advice you to either not care bc you enjoy playing with your wife or of you care you maybe shouldnt play the competetive mode with sb of very different skill level so blizzard can mm and rank you properly.

14

It’s obvious you are getting emotional and reactionary at this point so let’s not continue.

But I will clarify for you. Prior data. (your previous rank) plays an enormous role in weighting the direction your rank trends.

When you are at or near your expected rank, the win/loss will start to play a greater role and when you are outside of that range your weighted history will have a greater role.

The system is designed to use the largest possible set of data.

15

Pretty sure it all boils down to blizz being a scum company. You play more, showing you’re more commited, meaning you’re likely more willing to play more to grind rank.
Would bet anything they got engagement based matchmaking going on

16

Could you just explain why you think this? Do you actually have some evidence to support this theory? Or are you just making things up. Because I have never experienced anything like that, in thousands of hours of gameplay.

17

I’m just really confused why you are completely ignoring my statements. You insist it only looks at wins/losses, despite me telling you that it doesn’t, from actual personal experience. It seems you are just making things up.

18

Limited sample size per season when they arbitrarily place you lower each season and expect you to grind up.

…and when you don’t grind to give them more data they add a stronger and stronger weight each season (by using the aggregate win/loss and final rank) to place and keep you lower ranked.

Yes, that’s a good point and I wish I had the time but I don’t. I know I’m not the only one with a similar problem and it screws up their whole system for people to have this problem.

It’s where a lot of instability is coming from.

Again, I don’t have a great argument here.
I like the game, but playing in a really unstable environment is also a huge detractor. I’d love the game so much more if they’d just alter their theory just slightly, and that would help a ton of other people at the same time.

So I “complain” because I just want enough people to know there is a solution that eventually someone at Blizzard reads it and thinks about it.

That’s actually my goal. Trying to get their attention so someone can tell them to stop thinking in ELO terms. …more specifically that they can promote team play without enforcing it at every level.

That’s the issue. they have a philosophy of selflessness over selfishness and I agree with them. It really does change the game. It just won’t work as a ranking system for low ranks.

19

thats a whole seperate issue they introduced with ow2. they are however not actually ranking you lower, only the visual representation. you are still getting matched according to your mmr. which basically means yes you gotta grind like 30 or so games per sason to climb up to the rank that matches your mmr of last season. its basically an auto climb. but yeah its stupid and i hope they get rid of it moving forward bc feedback from the players has been negative across the board.

i think all things considered, you probably should accept you are not really the target audience for ranked mode.
since both qp and comp feature mmr based matchmaking, it might be the better path to just stick to qp and arcade. comp is aimed at ppl who play a lot and got the time to do so.

1 Like
20

Dude, it’s a video game. Chill. Your rank means absolutely nothing to everyone but you. If you aren’t having fun, stop playing. No reason to make yourself angry over something like a video game.