Forums / Community / Matchmaking Feedback & Discussion

[Locked] Matchmaking Feedback Update – May 8

OP ZaedynFel

  1. 1
  2. ...
  3. 2
  4. 3
  5. 4
  6. ...
  7. 5
This post has been edited by a moderator. Please refrain from making non-constructive posts.
*Original post. Click at your own discretion.
Spoiler:
Show
ZaedynFel wrote:
Collet005 wrote:
Is it possible to add the beta aiming style (or somethig similar) to the aiming options ? After the "heavy aim fix" are you planning any other fix to aiming?
Gonagon wrote:
Now I wanted to make a suggestion to give players the option to change the aiming mechanics. Option one the current and option two the beta style aiming.
Like was mentioned, I'm not actually involved in any of that, though I have full confidence in the people who are. But I can't speak to it.

I brought up this particular issue because I already knew a little about what was going on and could clear things up a little, but I'm not involved with the folks that actively work on that stuff, so I can't say anymore than what's already been decided and planned and it ready for the public.
Ok thanks for that though. Is there anyway a message could be sent to the sandbox team then. I think this would be a really interesting feature. Because there's still a lot of people including myself whom preferred the beta aiming. But I know there's those who like this aiming better or just as much so that could really satisfy a lot of fans. Who knows maybe in an update in the future this could be made a thing. Aim mechanic customization beta style, current style etc.
Oh I have a question can you keep us up to date on when this update rolls out, like every time you hear of something good of it keep us updated on what's going on.
ZaedynFel wrote:
Party RestrictionsI still see a lot of questions about this, so I’ll review because it’s an important topic both to me as well as you.

As I’ve said in the past, our focus for party matchmaking is integrating a new model we’ve tested that does a great job at predicting whether a party will win a match. Since we can predict who will win in a party, we can make sure to match parties against teams that are just as good as them.

Not all parties are equal. Some play worse as a party, some play better. We think we can determine that and matchmake efficiently and appropriately.

For a quick review (see my past posts for more info) it works by modifying the skill that we send to the matchmaker to find your party a match. If you are good at playing in parties, and in a good party, it will find a harder match. That may mean another party, or it may mean four solo players who are much better than you individually, but equal to you when in a party. If you are one of the top players in the game, then it won’t be able to find four solo players good enough, and you will always play vs. a party if there’s one available. But if you are a Gold party, we can probably find a Platinum or Diamond group of solo players that will be the correct challenge.
Some questions here :)

  1. You said if you're a top player it won't be able to find 4 solo players, so you'll always play versus a party. What if there are no parties available? then what happens?
  2. Can your party be matched against partial teams? Say for example in 4-4, two teams of two? Or is it either a full party or all solo players?
  3. Does this mean parties could possibly play the same teams over and over and over again?? If so, this isn't a good thing in my opinion. That will get, really old, really boring, really fast! Unless you just happen to keep having super great close games, but that doesn't happen very much lol Anyway to put something in there where teams can't play the same teams back to back games?
  4. What are you basing your search parameters on when looking for another team? Is it based total wins? K/D or just there ranks? I know you said you think you can determine that, but how is all this determined?
  5. Is this system going to be implemented in all ranked and social playlists? Or just ranked?
  6. What if your party isn't a full party? Say you only have a fireteam of 3 (playing 4-4 here) Will the searching system use the old system to find players or the one your going to be implementing?
I'm​ going to be honest here. I like the idea of this, a lot..
but I seriously am having my doubts here. There are lots of times I play in a team that consists of say something like a mid Platinum player,a low Diamond Player and maybe either a other diamond player and or one or two low Onyx players. If that group I just said goes up against a mix of high Platinum, low Onyx players, and/or maybe a champion, we most likely lose every time. Yes we have the advantage of talking as we are a team, but at the end of the day, skill still wins. If a team can't be found, I don't think teams should go up against solo players who have higher skill rating. It should be players of slightly better skill, but not a lot.

I'm just concerned that this system will make it so people will not want to play with there friends who aren't as good as them in ranked as much or even soical. I hope this isn't the case though, I truly do. To me soical should just be you can get paired up against anyone of any skill. Just like in passed Halo games. I'm an mid to semi high onyx player, but a lot of my friends are high gold to low Diamond players. When I play with them now I'd say we lose as much as we win and things seem pretty balanced to me.

I know some things I put here will get answered when this new match making system goes live. Regardless, I'm looking forward to seeing how all this will work. Thanks for the continue information :)
You say you think you can achieve better MM by accounting for whether people play better or worse in a party, but will you also influence the CSR or MMR gained/lost with that calculation?

For example, say I'm MMR 1200, and party with 3 friends who are also MMR 1200 (diamond 1). Your system thinks we play "better" as a party, like MMR 1500 (Onyx), so it matches us against 4 solo players with Onyx 1500 MMRs (and presumably CSRs in that neighborhood). Let's say we win. Do we gain CSR as though we beat an EVENLY skilled team? Or do we gain CSR as though we beat a MUCH HIGHER skilled team?

If you implement the former (we gain CSR as though we beat an evenly skilled team), I think you've created a problem, because you've essentially now simultaneously given me two different CSR values. That begs the question, which CSR should be displayed as my official rank? My solo CSR (1200) or my party CSR (1500)?

Meanwhile, what if someone else plays WORSE in a party? Which CSR will be displayed for him, solo or party CSR? He would clearly want his solo CSR to be displayed, while I would want my party CSR to be displayed.
  1. You said if you're a top player it won't be able to find 4 solo players, so you'll always play versus a party. What if there are no parties available? then what happens?
Actually, I meant it the other way around. If you are a top player in a party of 4 top players, then it probably won't be able to find 4 solo top players that would make it an even match, so they would wait awhile. Maybe eventually we would allow the match if everyone involved has waited a significant amount of time and at that point just want to play.

  • Can your party be matched against partial teams? Say for example in 4-4, two teams of two? Or is it either a full party or all solo players?
Yes, anyone can match anyone actually, as long as the adjusted skill gap is still small enough to make it fair. So to4 vs. (to2,to2). Some to4s have an advantage over being in a to2, so in that case, the to2s would probably have to be individually better than those in the to4.

  • Does this mean parties could possibly play the same teams over and over and over again?? If so, this isn't a good thing in my opinion. That will get, really old, really boring, really fast! Unless you just happen to keep having super great close games, but that doesn't happen very much lol Anyway to put something in there where teams can't play the same teams back to back games?
We'll keep an eye on this. It will mostly likely only happen near the very top where they probably won't mind it because otherwise they would have to wait hours.

  • What are you basing your search parameters on when looking for another team? Is it based total wins? K/D or just there ranks? I know you said you think you can determine that, but how is all this determined?
It'll use MMR like we already do, but a state-of-the-art upgrade to the already great MMR system. It can look at whatever we need it to. Today, it already looks at who you win and lose against, and does quite well predicting match outcomes. In the future, we can add kills, deaths, quits, party state, etc.

Our prototype has yielded very good predictions. It predicts the winner, and also fairly accurately how many kills and deaths each player will have --- ahead of time!

So if we have a system that can take any two teams and tell us the probability that A will beat B, we then know ahead of time if the match is even enough. Ideally that probability would be near 50%

  • Is this system going to be implemented in all ranked and social playlists? Or just ranked?
It will be like today. We'll have the predictions available in both social and ranked, but probably not enforce them as tightly in Social. Today, Social still tries at first for a close match. We just don't enforce a max skill gap (worst possible match) like we do in Ranked, so Social gives up eventually if it has to. That means higher-skilled players like you tend to get easier opponents.

  • What if your party isn't a full party? Say you only have a fireteam of 3 (playing 4-4 here) Will the searching system use the old system to find players or the one your going to be implementing?
Same system. It doesn't need to do anything different other than modify the expected performance (skill) of your party, and find a teammate that matches that modified expectation.

I'm just concerned that this system will make it so people will not want to play with there friends who aren't as good as them in ranked as much or even soical. I hope this isn't the case though, I truly do. To me soical should just be you can get paired up against anyone of any skill. Just like in passed Halo games. I'm an mid to semi high onyx player, but a lot of my friends are high gold to low Diamond players. When I play with them now I'd say we lose as much as we win and things seem pretty balanced to me.
This is one of our biggest concerns, and why we haven't just blindly separated parties out. Our data does show what you are describing will happen if we blindly assume all to4's are created equal, so we are going to test that thoroughly first.

We've seen the same thing you describe, casual to4s actually play worse than solo players in the data, surprisingly enough.

This system has the potential to learn all that, and properly match into fair matches either way.
RzR J3ST3R wrote:
You say you think you can achieve better MM by accounting for whether people play better or worse in a party, but will you also influence the CSR or MMR gained/lost with that calculation?
The only real difference is matchmaking will be better.

Today, if a group of 1200s play like 1500s together, then MMR and CSR of both end up being 1500. The individual MMRs are inflated because it doesn't know there's a party advantage.

Likewise, today, if one of those 1200s goes and plays outside their party, they still play vs. 1500s (MMR is 1500) and they get owned until both CSR and MMR drop to 1200 where they "belong"

With the update, if a group of 1200s always plays together, performing like 1500s, and we know their party adjustment is 300, then their MMRs will be 1200, but their CSRs will still be 1500 because they will be going 50/50 vs. 1500s.

If one of those 1200s goes and plays solo, instead of getting owned while they drift down to 1200, they will immediately match vs. 1200s, but watch their CSR drift down because they will be getting less for a win and losing more for a loss.

So, the overall net is the same as today, just with better matchmaking.
ZaedynFel wrote:
RzR J3ST3R wrote:
You say you think you can achieve better MM by accounting for whether people play better or worse in a party, but will you also influence the CSR or MMR gained/lost with that calculation?
With the update, if a group of 1200s always plays together, performing like 1500s, and we know their party adjustment is 300, then their MMRs will be 1200, but their CSRs will still be 1500 because they will be going 50/50 vs. 1500s.

If one of those 1200s goes and plays solo, instead of getting owned while they drift down to 1200, they will immediately match vs. 1200s, but watch their CSR drift down because they will be getting less for a win and losing more for a loss.

So, the overall net is the same as today, just with better matchmaking.
I like it. <thumbs up>

Thanks for the response!
"till it meets our quality bar"

I just hope that bar isn't the usual 343 bar...
Any chance you can expand on what a 'temporary state' means? Are we lagging? Is the controller input messing up? Are we in the Matrix?
LUKEPOWA wrote:
Any chance you can expand on what a 'temporary state' means? Are we lagging? Is the controller input messing up? Are we in the Matrix?
My explanation would be so terrible it's not worth trying. I think they'll probably do a write-up on it. If not, maybe I'll get something I can copy-paste from the engineers after the fix is in.

It's not network lag though.
ZaedynFel wrote:
LUKEPOWA wrote:
Any chance you can expand on what a 'temporary state' means? Are we lagging? Is the controller input messing up? Are we in the Matrix?
My explanation would be so terrible it's not worth trying. I think they'll probably do a write-up on it. If not, maybe I'll get something I can copy-paste from the engineers after the fix is in.

It's not network lag though.
I've wondered how accurate Wu Ip Man has been in his diagnostic of the issue. It'll be interesting to see if the engineers potentially confirm some of his suspicions.
Alright, good to know.
Thank you for taking the time to post these updates, even the ones that you're not directly involved with.
I really am hoping to see the update for the aim fix and party matching within the month. Can you give us an estimation on a potential release? And please dont tell me itll take months.
Will we ever get a voting system for something like action sack? I feel it would be beneficial for modes like it so we can target the good modes.
Gonagon wrote:
I really am hoping to see the update for the aim fix and party matching within the month. Can you give us an estimation on a potential release? And please dont tell me itll take months.
Answering questions like this is a lose/lose.

If he tells you "approximately X" and then something happens, people will be upset that it got delayed and that, "343 lied!"

If he doesn't tell you, people get mad and assume it's not coming.

My advice would be to not ask for things like this. They are working on it and we know this. They'll tell us what we need to know, when they deem we need to know it.
I know this is already an issue and it has been brought up many of times... I'm ranked platinums in the hcs playlist and I've been coming across a lot of smurfs lately. It's really turning me away from the game when I run into low level players who drop 20+ kills a game.. it's became way worse since the season reset.. atleast every match I've been running into smurfs and players who are obviously a lot better then my friends and I. Either then that I have been enjoying the playlist a lot good work.
Gonagon wrote:
I really am hoping to see the update for the aim fix and party matching within the month. Can you give us an estimation on a potential release? And please dont tell me itll take months.
Answering questions like this is a lose/lose.

If he tells you "approximately X" and then something happens, people will be upset that it got delayed and that, "343 lied!"

If he doesn't tell you, people get mad and assume it's not coming.

My advice would be to not ask for things like this. They are working on it and we know this. They'll tell us what we need to know, when they deem we need to know it.
Thats why I used the word "Estimation" since it doesnt mean an official release date.
Gonagon wrote:
Gonagon wrote:
I really am hoping to see the update for the aim fix and party matching within the month. Can you give us an estimation on a potential release? And please dont tell me itll take months.
Answering questions like this is a lose/lose.

If he tells you "approximately X" and then something happens, people will be upset that it got delayed and that, "343 lied!"

If he doesn't tell you, people get mad and assume it's not coming.

My advice would be to not ask for things like this. They are working on it and we know this. They'll tell us what we need to know, when they deem we need to it.
Thats why I used the word "Estimation" since it doesnt mean an official release date.
It's still lose/lose. He's a part of the team, so anything he gives out will be assumed to be "official" by us. He gives us an estimated date, estimated date comes and goes, people ask what happened to stuff happening on estimated date. People get mad. Just the way things work out.
  1. 1
  2. ...
  3. 2
  4. 3
  5. 4
  6. ...
  7. 5