''it took 2 years to create our matchmaker''

Competitive Discussion
6

I believe KarQ did an interview with Geoff Goodman or whatever his name is on his stream.

The youtube channel: Your Overwatch covered this in one of their newer videos, cba with finding the video or timestamp so feel free to check for yourself

7

Hi: it is accurate. We have to see how you use it.
For example: This is a team-game. IF you are so good and get 60 kills. You are eligible to fight another player who get 60 kills, but this players is a true smurf and you are just a selfish player who play alone and farm medals. So play as team and help your teammate to die less, you will get easier games.

You can climb with perfect mechanics and game sense. OR playing as team.

1 Like
8

If CRAY and other super computer companies can calculate Launch trajectories for an Atlas Missile, a Higgs Boson particle, play against a chess player or even predict weather patterns for NOAA. I am certain Blizzard can afford to buy a low-end unit to calculate player performance based upon much smaller perimeters that occur in this game.

1 Like
9

the only thing that i would like to see is the rank range of where people mostly sit (gold plat?) there is a more limited gap between SR for teaming. I always hate seeing a gold i’m my diamond game :frowning:

its not we are lacking of people that the Q will drastically change. the majority of people sit there anyway.

10

Actually, you underestimate the complexity of human behavior especially when it comes to interaction and its effects on your performance. The idea is to determine your contribution to a win.
Try to express that in numbers and you’ll understand how it becomes increasingly difficult with the current set up (Groups, Solo Q, diverse Hero pool etc.).
There is a reason why they wanted to avoid Solo Q :wink:
In the end, it boils down to the question of whether your MM variation delivers more accurate results than a simple random approach. I assume that removing MMR (moving average?) would probably result in more fluctuation.
In the end, only blizzards dev team knows, so saying its good, bad, accurate or inaccurate is baseless argumentation anyway.
I’m pretty sure that in most cases (MMs) a GM would be able to escape bronze with ease.

PS: Chess presents far more limitations thus allowing even brute force methods, should have used Go as an example instead.
Weather prediction (more than a couple of days) are usually quite inaccurate (depends on how you view it), on top of that its a worldwide problem, so you have whole planet working on it. So that’s a flawed comparison anyway.

1 Like
11

So essentially all efforts are pointless in a match when we are hard-stuck in silvers because of feeders? if that is the case than we simply do not even have a mode worth playing at this point. At some point the game has to allow those whom want to play a game to climb out and those that are willing to throw competitive play to be stuck or drop.

Essentially If I were to sit down with a TEAM and we write a program. I would begin examining all 29 heros. With the first change being a system which looks at their cohesion performance.

(Any hero that is good for flanking or is designed to flank does not apply.)

The rest of the heroes as such. Receive an invisible perimeter which shows up only to me as the programmer, and we will call that a cohesion perimeter. (Circle outside of their character bubble. Sorta like what lucio sees for his healing.)_ So if any character that is designed NOT to flank, is played by someone that just randomly leaves the group perimeter within a certain radius and moves OUT of that zone. It is where we start a new script.

  • Is that person getting kills?
  • just doing some small damage?
  • did they make an impact on the game, or did they feed?

That is a good start. We work with the system, tweek it and allow that to be the First factor there.

Then as the game evolves we can go back and look at other perimeters to start determining player Skill Rating and go from there.

Ultimately Team feeding is the #1 problem in Competitive play and those whom just start to throw a game. If we wanted to, we could chose NOT to penalize the other players on the team for that person’s actions. So if that IS happening, we are not bumping them DOWN. If that is too much to ask of a developer to do, especially a multi-million dollar company. Well then, I guess we will just continue to have our community go away. Which is NOT good. :frowning:

12

I would love to see all of the backseat hackers on the forums come together to make their own ideal game company. They all know how easy programming is, they should just make the perfect game already!

And it would, according to their estimates, only take a few weeks!

1 Like
13

Lol their matchmaker is actual amateur trash

14

I believe they’re talking about this:

However, as you can see, it’s really only relevant to Arcade’s matchmaker (which has been notoriously bad in the past). They aren’t trying new stuff for Comp or QP.

15

Saying that they’re only testing the matchmaking changes in Arcade is not the same as saying that the changes, if applied, will only affect the Arcade.

1 Like
16

It was in a Unit Lost or Your Overwatch video I believe.

17

I’d like to see where you performed the experiment and did the data analysis to show that this is in fact the case.

1 Like
18

So then do it.
Go create a match maker.

And it’s not just ‘make it random’ because if you had a system like that, it would crash and burn.

I want you to come up a with a good system.
Go code/create it and come back to me with a good answer.

1 Like
19

It’s like saying ‘i’ll make better food!’ and then making toast each day.

Just because it’s food that doesn’t mean it’s good or better.

1 Like
20

Its the first few days of another season.

expect terrible matchmaking and tons of throwers in your placements.

21

Who was crying? The very best season was Season 1 because nobody came into it ranked.

I can tell you who cried…people who lost and though they should have won. Can tell you right now from not being the very best current player but being average then (gold) it was more than fine. It was fun and sometimes the only way you saw someone who knew how you could play the game.

The only thing the SR system has done is fake people into thinking they have access to becoming some sort of esport god. Nobody cares who you are or what your rank is, especially on console.

They should get rid of it, really. It’s just nonsense to make people feel like something is happening that matters when it really doesn’t.

22

I think the game would be much better if they had a LITTLE MORE PBSR.

Not ALL PBSR. Just… enough to be able to climb when you’re stuck there with people who fall off the map, 1v6, throw, leave, whatever.

The matchmaker, from what i’ve seen, is pretty good at creating even matches. I happen to see a lot of overtimes, a lot of close games. I feel like that was Blizzard’s main focus. It could be just me, but most of the game are close. That’s a good thing, imo.

However, i don’t think the game is that good at getting people to their actual ranks. I see a lot of good players, who try, stuck in these games with throwers and even new players who fall off the map and crap. Then the matchmaker is trying SO hard to create even matches that you’ll win about 50% of your games no matter what you do. Hard to rank up then.

Let these better players get some heavier PBSR and get them playing together at higher ranks. The game could be SO much more fun if you felt rewarded for playing well.

I get that it’s a team game, and i’m not saying you’d GAIN for a loss or have to reward completely based on someone’s performance. That’d be too hard to calculate. But… there’s enough data there to get a general idea of someone has done well enough to lose a little less SR than the actual thrower who did pretty much nothing all match.

23

Really? I give you 5 minutes to WRITE DOWN the formula of the current Matchmaker. All with SR gap limit, stack balancing, BPSR and weighted win/lose compensation.

Write it down in pseudocode. I don’t want bs amateur long paragraphs. Make it short, concise, logical pseudo code. I’m waiting.

2 Likes
24

What’s wrong with matching 6 players on each side at very similar SRs as possible so that both sides’ SRs are equal to each other. Then just let them play?

1 Like
25

Yeah I have noticed that match maker has a few major faults that nobody pays attention to. When we look at screenshots of the match maker and keep a bunch of them, we can gain a significant idea of what is going on. Including we can head on over to Overbuff to help us uncover the flaws.

Remember, these are MY findings. If you do not agree with them. People are welcome to present their evidence. I am not going to use player names here or their ID’s, due to forum rules.

During a LOSS:

  • What I noticed on matches that were losses, the enemy team had players with significantly longer account time experience.

  • The team I am on. They had the significantly less account time experience combined.

  • maybe 1 or 2 players had Previous competitive season experience compared to other team. Which had 3-4 with previous competitive experience.

My thoughts

To me it really comes down to account time experience level and previous competitive match history which indicates if the game is a loss or a win. I am starting to believe that MAYBE we should be ALLOWED to leave at the start screen and cancel the match because the system is FLAWED.

  • We get players which do NOT fill the proper roles.
  • We get players whom start flaming chat.
  • We get a system that doesn’t give us team mates whom care about gameplay quality, such as Altnerate accounts (smurffing)
  • We get a system that has stacked the odds against us.

Anyway, Competitive can either be over-hauled or it can become a game which nobody wants to play anymore. Either way :frowning: