MMR vs Ranks

Shroompy

So Ive been thinking recently about what has been said in previous dev streams. More specifically, how theyve said that MMR has been making more balanced matches than ranks ever have. But how do they measure that?

There is no reason why veterans should be matched with new players and its not like its a rare occurrence. There is no reason why a system thinks that is a fair match. The old rank system was even better at preventing that.

Which is what I dont understand, why waste so much time and ressources on a flawed system when you have a REALLY strong foundation with the Emblem system? It was far from perfect but it MUCH closer than MMR ever will be. The difficulty of ranking up was the problem, it was too easy and every one started to clutter up to the higher ranks.

I know BHVR is hesitant on getting rid of MMR because of how much time and ressources were dumped into it. But absolutely no one is benefiting or enjoying this system.

Pulsar

 But how do they measure that?

If the entire game averages out to be closer to a 50% Kill Rate, it's more balanced according to them.

GeneralV

It is as I always say: just because you can change something, doesn't mean you should.

The Emblem system was not perfect, even I am forced to admit that and you know how much I value things from the past, but replacing it with a clearly flawed system was definitely a terrible idea.

Everyone who was around those atrocious test sessions called it. All BHVR had to do was listen.

lagosta

The emblem system only served to time gate new players. Veterans with thousands of hours would be matched with and against clueless survivors/killers all the time and could consistently go 50 or more matches without losing. It was only good for people who played for a long time before and reaching red ranks as a new player meant you were getting destroyed.

MMR needs different criteria, but to hell with the ranking system. Keep that thing under the ground.

dugman

The emblem system was terrible, the current system is definitely better.

As far as measuring the results, I'm assuming they look at the statistical ranges of kills per match for matches which have all five players in roughly the same MMR bracket. Their goal is presumably an average of around 2 kills per match and median results between 1-3 kills per match. (So most matches at least one person escapes and one person dies.) And anecdotally, at least in my games, that seems to probably be the case.

Mismatches though are caused by other factors unrelated to whether or not the MMR itself is accurate.

• If someone leaves a lobby, the game has to find a replacement as soon as possible to prevent the four players who stayed in it from having to wait an even longer time to play. So it will shove in the closest match at that exact moment rather than wait for a better one
• Players with high outlier MMRs might not have enough opponents in their range to go against, so after waiting a while they'll eventually just get put into whatever lobby is available. Likewise for new players. (Which ironically is probably the cause of the thousands hour streamers going against brand new people with very, very low ratings. There aren't enough people in either of their brackets so they get put into whatever slot is up at that moment, sometimes against each other.)
• Some people play in swfs with their friends who can be very different MMRs. While the system does make reasonable matches when all the players are around the same MMR, I don't think it necessarily is as good at estimating what MMR a killer should be when some survivors are much higher than others in the same group. It's probably just doing something simple like taking an average, when to be really accurate it would need to do something more sophisticated like have a neural net trained to input survivor MMRs and output an optimal killer MMR.
• 
  

Shroompy

@lagosta said:

The emblem system only served to time gate new players. Veterans with thousands of hours would be matched with and against clueless survivors/killers all the time and could consistently go 50 or more matches without losing. It was only good for people who played for a long time before and reaching red ranks as a new player meant you were getting destroyed.

MMR needs different criteria, but to hell with the ranking system. Keep that thing under the ground.

"Reaching red ranks as a new player"

That shouldnt of been happening in the first place, ranks had a MUCH better foundation than MMR and really all that needed to be done was make it more difficult and you would have a much better system than MMR ever will be

Shroompy

@Pulsar said:

 But how do they measure that?

If the entire game averages out to be closer to a 50% Kill Rate, it's more balanced according to them.

Which literally just means newer players are getting destroyed lowering the average, and the vets destroying these new players are just increasing the average

Pulsar

@Shroompy said:

https://forum.deadbydaylight.com/en/discussion/comment/2800652#Comment_2800652

Which literally just means newer players are getting destroyed lowering the average, and the vets destroying these new players are just increasing the average

no no trust me bro it works

Twins needed a nerf, they were too good

/s

bm33

I miss the emblem system, it at the very least encouraged better gameplay and matches were more fun. It's great their stats say mmr is working but it has removed the fun in the game. The play styles people hate - killers camping and survivors hiding - happens way more now because there is no emblem system penalizing them and keeping them mostly in low ranks, now since only escapes/kills matter these playstyles can be found in all ranks.

fulltonon

They both do not, period.

lagosta

@Shroompy said:

https://forum.deadbydaylight.com/en/discussion/comment/2800671#Comment_2800671

"Reaching red ranks as a new player"

That shouldnt of been happening in the first place, ranks had a MUCH better foundation than MMR and really all that needed to be done was make it more difficult and you would have a much better system than MMR ever will be

A much better foundation? Really?

Anyone who had an average pip greater than 0 would rank up if played enough games and eventually be in red ranks. And you must have forgotten how survivors competed against each other for unhooks to the point of hookbombing. Or how everyone in the match depiped if the killer's skill was way above or way below the survivors. Not to mention how cryptic and unwelcoming the whole Emblem system was for the players, people didn't even know what constituted a win.

That system only served to help people who made a living out of this game to stomp casual players.

Shroompy

@lagosta said:

https://forum.deadbydaylight.com/en/discussion/comment/2800720#Comment_2800720

A much better foundation? Really?

Anyone who had an average pip greater than 0 would rank up if played enough games and eventually be in red ranks. And you must have forgotten how survivors competed against each other for unhooks to the point of hookbombing. Or how everyone in the match depiped if the killer's skill was way above or way below the survivors. Not to mention how cryptic and unwelcoming the whole Emblem system was for the players, people didn't even know what constituted a win.

That system only served to help people who made a living out of this game to stomp casual players.

Yes, it did. A foundation is the idea of it and its potential. Increase the difficulty of actually ranking up and you already have a better MM system, compare that to the system that only takes into account kills and escapes.