The official Matchmaking blog post

by Cheese9Man
Reply

Original Post

Re: The official Matchmaking blog post

[ Edited ]
★★★★★ Pro

@Cheese9ManWe'll see, I guess. I thought I saw imrpovement a few weeks ago but it seems like that disappeared and now I'm back in the "10 horrible games just to get 1 good one" world for solo queuing pubs.

The post really doesn't go into how skill ratings are calculated, just how they use them. I always get the feeling the game thinks I'm a 7 or 8 when I'm really a 4 or 5. I don't want details of the formula or what my actual skill rating is, I would just like to see what things make up that number.  Level doesn't mean anything relative to skill and their post seems to confirm it, so what else do they measure to calculate a skill rating?  Do they care about how lopsided your kill feed is when you die, for example.

Using the highest rated person's skill for premades is just wrong, IMO. The lower skill players tend to pull squads down rather than the better players carrying the squad, especially when the skill gap in the squad is bigger. They've been operating on a completely false assumption when matchmaking premades so I'm glad that's gonna change at least.

Message 11 of 264 (2,868 Views)

Re: The official Matchmaking blog post

[ Edited ]
★★★★ Pro

@Cheese9ManI stopped making comments on this forum because the matchmaking data I was sharing wasn’t cared for by the majority of the forum community— but I feel inclined to comment here on this matter:

1) HUGE props to Respawn for even having a convo on matchmaking. Like seriously, I respect this so much. Lack of communication is a literal middle finger to the community, so this is appropriate and appreciated.

2) Parts of this are— unfortunately, disingenuous.

What parts? These:

“We don’t purposefully put you in harder matches to slow you down if you’re winning a lot, nor do we intentionally put you in easier matches because you’re on a losing streak.”

By designing their “dynamic matchmaking rating system” that is LITERALLY EXACTLY what they do. They 100% punish people for doing well, even in one single game. The algorithm reads your 3 kill 1st place win and the next game you are playing with people significantly above your skill level. Next disingenuous statement:

“Either way, this is the system accounting for your recent changes in skill. This process tends to be slow, so you should only feel these changes from long streaks.”

This is true for many accounts but definitely not for all. I have literally studied this across 20+ accounts and shared my findings (something I’m sure Respawn wasn’t betting on) and the results prove this to be untrue.

Accounts with a lot of data that show that a player is above average do NOT have slow mobility down the matchmaking ladder. Win a game with just 2 kills and above 50% placement and the next match is right back to the top of the MMR ladder. I’ve literally tested this almost every single day for the past 2 years so I am going to need a PhD level argument/rationale to convince me otherwise.

I guess I’m happier that they are addressing MM, but tightening of matchmaking in pubs for no reward is an ABSOLUTELY bad thing. People need to be rewarded for their proficiency, EVEN IN PUBS.

Tighter matchmaking means more people playing their doppelgänger team at any given time, which results in outcomes being more similar for people, regardless of skill.

Players who are more proficient should have better outcomes (wins, etc) in pubs. I’ll say that again: Preds (by virtue of their proficiency) deserve to beat players of my skill level in pubs. Similarly, Diamond-level players should have better outcomes than gold-level players.

Tighter matchmaking and everyone breaking even feels super forced, curated, and exhausting.

 

But the needs of the many outweigh what’s actually fair. So it is what it is. We should probably just commit fully at this point and have a system that does in-game adjustments for players performing too well or too poorly. So players on a hot streak should have their bullet damage decreased and their movement slowed slightly. We need to push for a system that results in uniform outcomes for all players. 

Message 12 of 264 (2,826 Views)

Re: The official Matchmaking blog post

Champion
@Axs5626Sxa5001 Welcome back. WhenI saw the post I immediately thought of you.

You pointed out some stuff I thought was a bit iffy too. Although I'll say they contradicted themselves a bit they did a good job. The MM does for the most part do what they say, provided you are not at the very top or very bottom.

I'm curious how the premade stuff will be affected as it seems like that will be the most affected. I have a hunch it's going to mean more premades in solo queue lobbies. One thing solo players don't realize is that they've been very protected. Yes some premades slip through but I know from thousands of matches both ways that it's a totally different animal.
Message 13 of 264 (2,789 Views)

Re: The official Matchmaking blog post

★★★ Novice

@Axs5626Sxa5001 

 

“We don’t purposefully put you in harder matches to slow you down if you’re winning a lot, nor do we intentionally put you in easier matches because you’re on a losing streak.”

By designing their “dynamic matchmaking rating system” that is LITERALLY EXACTLY what they do. They 100% punish people for doing well, even in one single game. The algorithm reads your 3 kill 1st place win and the next game you are playing with people significantly above your skill level. Next disingenuous statement:

The post isn't saying this doesn't happen, it's saying they aren't intentionally throwing all matchmaking out the door and forcing players to match with harder players because they win a couple games.  They are showing the system is designed to give the player a harder match as they get more confident and play better.  We both know that in a set play session, playing for 2-3 hours you get better as you play... the systems design intentionally accounts for your increase in skill based on the information it's being given.

 

This being said, I personally have no problem with this design, the issue at hand is one they have seemed to identify, the level or "bucket" in which you change could drastically effect who you are placed with... For instance lets say you finish a session in bucket 3, you've lost 3 games so your skill rating goes down.... next day you come back win your first game because you are feeling more confident.... now you've moved from bucket 3 to 4 because you were right on the edge last night, inherently putting you with way higher skill players, causing you to get slaughtered and question your entire reason for playing, but this then lowers you back into bucket 3.  The systems design flaw is obviously going to be located at the edge of "buckets" because players can move between these based on winning and losing a couple games, where as players in the middle will see slower moments by inherently being further from the edge of the bucket.

 

All being said one, among many, major flaws in this system is definitely not enough buckets, the change from bucket 3 to 4 should show a skill change from gold 2 to 3, not gold 2 to diamond 1, in which the system does now. I agree totally that I can win a game or two and go from players below to right at my skill to preds it feels like, in fact it happens so often friends and I make fun of it.... 

 

Over all I am really impressed someone even took on trying to explain a complicated topic to an average crowd.  Most will never see the post and out of those who do, statistically 20% will actually understand it what is being talked about.  

 

Message 14 of 264 (2,609 Views)

Re: The official Matchmaking blog post

★★★ Novice

@Axs5626Sxa5001 wrote:

“We don’t purposefully put you in harder matches to slow you down if you’re winning a lot, nor do we intentionally put you in easier matches because you’re on a losing streak.”

By designing their “dynamic matchmaking rating system” that is LITERALLY EXACTLY what they do. They 100% punish people for doing well, even in one single game. The algorithm reads your 3 kill 1st place win and the next game you are playing with people significantly above your skill level.




The post isn't saying this doesn't happen, it's saying they aren't intentionally throwing all matchmaking out the door and forcing players to match with harder players because they win a couple games.  They are showing the system is designed to give the player a harder match as they get more confident and play better.  We both know that in a set play session, playing for 2-3 hours you get better as you play... the systems design intentionally accounts for your increase in skill based on the information it's being given
This being said, I personally have no problem with this design, the issue at hand is one they have seemed to identify, the level or "bucket" in which you change could drastically effect who you are placed with... For instance lets say you finish a session in bucket 3, you've lost 3 games so your skill rating goes down.... next day you come back win your first game because you are feeling more confident.... now you've moved from bucket 3 to 4 because you were right on the edge last night, inherently putting you with way higher skill players, causing you to get slaughtered and question your entire reason for playing, but this then lowers you back into bucket 3.  The systems design flaw is obviously going to be located at the edge of "buckets" because players can move between these based on winning and losing a couple games, where as players in the middle will see slower moments by inherently being further from the edge of the buckets 
All said though, among many, major flaws in this system is definitely not enough buckets, the change from bucket 3 to 4 should show a skill change from gold 2 to 3, not gold 2 to diamond 1, in which the system does now. I agree totally that I can win a game or two and go from players below to right at my skill to preds it feels like, in fact it happens so often friends and I make fun of it.... 
Overall I am really impressed someone even took on trying to explain a complicated topic to an average crowd.  Most will never see the post and out of those who do, statistically 20% will actually understand it what is being talked about.  

 

Message 15 of 264 (2,735 Views)

Re: The official Matchmaking blog post

Champion
@Rumpel1028 What level of player are you? Axsnumbers is a masters level player. At that level the MM doesn't really fluctuate very much. It takes a lot longer to get put into an easier bucket.

For me, as an average player, I see the fluctuations and agree with you.

I also agree the blog did a great job explaining a very complex subject in simple terms.
Message 16 of 264 (2,726 Views)

Re: The official Matchmaking blog post

★ Expert

All I know, is that the current system doesn’t work, although the break down is appreciated, I want a fair end result, that’s all that matters to me as a player/consumer.

 

I’ve shared numerous screenshots of my SBMM on here, top 100 pred opponents, 3 man pred premades, or the complete opposite being matched with silver/gold players while being P1.

 

Just do what you have to do, and give me a fair gaming experience.

Message 17 of 264 (2,719 Views)

Re: The official Matchmaking blog post

[ Edited ]
★★ Expert

Awful mm today. I'm being put on bot teams 8/10, and sus enemies a couple of times too. I'm spending more time afc than playing. Might as well jump on Star Trek then.

 

As I said...

_20230118_173443.JPG

Message 18 of 264 (2,704 Views)

Re: The official Matchmaking blog post

Champion
@Gabrielmojo26 I'm scared that ranked players are not going to get that. To me ranked should be the easiest MM. Same rank with maybe allowances for a couple small rungs up or down for premades. To me the article seemed indicate this is not the plan.

To me the article seemed to be improvements to pubs. More brackets, keeping people closer to their proper bracket so you don't get a skill 8 with a skill 2 like we all know happens now, which they admitted.
Message 19 of 264 (2,689 Views)

Re: The official Matchmaking blog post

★★★ Novice

@hayhor Yes, but what seems strait forward is not always that.  I am a system designer personally, so I know for a fact that saying something is easy isn't always the case.  It's really easy to say "well just do this" than it is to actually implement it that solution.  There are many many variables to take into account that they probably aren't even mentioning in the post, and probably even more we've never thought about.  Just reading this short post made me, someone who already designs systems, realize there is more to MM than I even realized.

Message 20 of 264 (2,676 Views)