So when is SBMM getting an update?
I know theres a matchmaking system currently but that was just a quick plaster fix, I'm sure no developer actually things Escaping shows singular player skill I mean you cant control your team to do gens or even for them to save you off hook.
Not to mention if the killer just wants you he can easily just tunnel you out and yes while the killer may lose because 3 others escape is it fair to punish that survivor who lead on the killer for the whole trial?
So this is why I'm curious when the actual SKILL based matchmaking will get added, while gens arent skillful i dont think it's fair to punish solo players because they cant control their team.
And no, dont nerf killers that's not the issue... the issue is SBMM just throwing randoms together
Comments
-
I'm sure no developer actually things Escaping shows singular player skill
you are mistaken. This is exactly what they think.
40 -
Last I remember hearing, it's constantly being tinkered with behind the scenes, so I'd assume it won't be getting big "updates" that are announced for a while? Hard to say.
As for "escaping showing skill", you're right that it doesn't show skill for one match. However, how often you escape overall does show your skill, which is why MMR systems are built around winrates.
5 -
To be actually fair and balanced, they did say they were working on a way to weigh personal performance as opposed to just the end product of escaping/killing. But considering every single time they announce a change with the intention of fixing X, they just make it even more horrible, it's probably better if they don't touch the system. After all, SBMM is only a little worse than grades was, do we really want an even worse system?
1 -
I would try come up with a sarcastic reply but it's true and that hurts my soul
1 -
They definitely need to do something.
And while i would like to agree it's still just RNG luck if you get a team that actually plays. I usually escape pretty often especially when I get an okay team but the past 2 days youd think I'm awful if you went of the escapes in those 2 days when in reality iv just had awful matchmaking and doesnt matter if I do 4 gens or get chased for ages if my team isnt also trying
I hope so, and grades wasnt too bad before they changed them so you lost less progress each month. They emblem system just needed tweaking to award chase and objective more than escapes (especially hatch) and altruism
0 -
The MMR is actually pretty accurate, when everybody in the match is about the same MMR the match is normally close. This was evidenced when they were doing tests on it earlier this year and the day they had the strictest controls on how close player MMRs could be in a match were the days with the “sweatiest” matches, while the days they turned off using MMR altogether and when they used Grades had the most inconsistent matchups.
The main reason for mismatches isn’t the MMR itself, it’s that the matchmaking algorithm for various reasons sometimes ignores ratings in the interest of getting people into matches more quickly and also that people with wildly different MMRs sometimes play together in swfs that can’t be broken apart by the algorithm. But there’s nothing wrong with using escapes and kills to decide if you MMR goes up or down, that works fine.
Finally note that the devs have never said MMR is a stop gap to a more complicated system. They did say, though, that they were looking into having MMR go up or down for survivors as a group instead of individually, that way if one person dies but their sacrifice helps everybody else escape the whole team’s MMR goes up, or similarly if everybody dies except one lucky survivor they all go down. No word on if they’re still looking into that change or not.
4 -
Yeah but the issue is it's not matching people with equal skill, and by that right I should never see players that dont do gens. I dont care if people cant lead as good chases but when you have a bunch of people walking around not doing anything then yes that's a major issue with the system.
I also dont think it's fair you could do a 20min chase and die on first hook and the game tells you that you did bad. A killer will always get you at some point and you cant save yourself either. Too much is in your teams hands for the system to punish you on their actions
2 -
I know i'm usually not one to exaggerate but it really is true that this is what they think. They said that everything that happens in a match is the middle man and that the end result is what it's important... it's more or less true.
0 -
Not at all! But this is probably one of the most misquoted things, so it's worth talking about again. Escapes are not skill, but they are a proxy to skill - that is to say that skilled plays tend to improve your chances of escape.
This does not mean that the game thinks you're a bad player if you have a really good match but die in the end. It also doesn't mean that you're a good player if you hide all match and sneak out the exit gates after your teammates carry you. No matchmaking system in any game is able to accurately estimate your skill from a single match. Matchmaking systems work over the course of many matches to continually refine your rating. If you're lower than you should be, you'll outclass your opponent(s) and your odds of winning will be higher, meaning you'll gradually climb over the course of several matches.
To give an example, you lose two matches and then win two. You'll end up more or less around the same point that you started. The game doesn't think you suck just because you lost a couple matches, it's looking at the bigger picture and always trying to make sure you end up in the right spot.
The reason why matchmaking systems use a single metric (like kills / escapes) to estimate skill is because adding a ton of variables into the mix has huge consequences. Say you count two hooks as a 'win' for the Killer- does that mean that the Survivor who escaped also won? If so, you have a situation where both parties can win, raising both of their ratings. This might not sound like an issue yet, but it has just created a scenario where on average you're more likely to gain rating than lose it, meaning people are on average always climbing higher. This completely breaks the entire matchmaking system.
But even so, adding in a bunch of other variables doesn't necessarily indicate skill either: If a Killer gets two hooks, that doesn't automatically mean they played super well. The matchmaker can't tell if you got those hooks through some tough chases or if you just tunneled someone off the hook and downed them before they could get to a pallet or window. Adding chase lengths into the mix is yet another variable and yet again doesn't reflect skill (you might have downed someone really quickly because of a skilled play, but they might also be in a very safe part of the map where that's not possible). Already, you can see the rabbit hole this is turning into. There's far too many factors to look at a single match and say, "This person is amazing, strap a rocket to them and send them to the moon!"
By using a single metric like kills & escapes, that ensures that one person always wins and one person loses (or they draw). In order to keep climbing, you'll need other people with a similar rating, otherwise the gains you get from each match gradually diminish until you stop climbing.
But all that said, the matchmaker's job is purely to stick you in a place where you'll win about half the time, not to judge you and make you feel bad about yourself. Whether that's because you're a legend who needs to go against other legends, or because you're crouching in a corner because you know you'll only last 5 seconds in a chase. It's only trying to find a spot where your strategy, regardless of how skilled it is, works about half the time.
To get back to the original question: Yes, we do have some improvements in the works, including team-based ratings (so dying heroically to save your team doesn't count as a loss). I look forward to sharing more info about it and the other changes when they're a little closer to release.
52 -
"If you didn't escape did you really play well?"
I'm pretty sure that's what Patrick said in the stream, or close to dont quote me
9 -
I just dont want to lose rating because my team avoided doing gens, gave up or let me die on hook for 0 reason.
System doesn't need to be perfect but I shouldn't be paired with those type of players randomly.
6 -
Thanks for clearing it up. I believe it's a common belief in the community from how it was presented in the September q&a stream last year. They did mention this but i feel like it was overshadowed by how it was said. They made it seem like players that don't escape aren't skilled. I'm glad to see that this is how it is truly viewed from the devs.
5 -
Can you please tighten the MMR range for matching high MMR players? I'm a mediocre survivor that plays solo queue to do survivor challenges and I'm getting matched with other solo queue newbs against killer main streamers with 5k+ hours that are playing for a living. The games are consistently ending fast with 5 gens up. No one gets any points and it's an abysmal gaming experience
4 -
I take some responsibility for this. MMR systems are incredibly complex and in hindsight, trying to boil it down for a 1-2 minute answer in a Q&A stream was not the right medium for it. Since we have a lot of casual players watching those streams too, we can't spend half an hour talking about how matchmaking systems work (everyone would fall asleep). Going forward, I'd much prefer to talk about the matchmaking system in blogposts and longer answers so we can give the necessary context.
25 -
I'm definitely glad to hear you guys are looking into team-based ratings so that being a Bill isn't punished.
That said, do you also have any plans to address ELO hell? Essentially, it's that the game tends to be killer-sided in low-rank matchmaking due to survivors not understanding their goals or how to run tiles (and conversely, the game being more survivor-slanted when you look at the very top.) The idea is that while moving downwards in MMR as killer increases your odds of winning your next match... past a certain point, moving downwards in MMR as survivor decreases your odds of winning by increasing the likelihood you get bad teammates, which has more of an impact on your ability to escape out the gates than also having a weak killer (because you're not designed to outrun the killer forever, and it's an impossible ask that one survivor does most the gens, gets most unhooks, and also loops the killer for days.)
It's very frustrating when I end up in a lobby full of players with less than 50 hours when I have over 3000. Currently the prestiges in lobby (which isn't the greatest idea because it encourages lobby shopping and targeting) gives me a better shot at dodging completely new teammates, but some way to segregate newbies so they're not likely to play with or against very experienced players would be appreciated. Because on the flipside, when I inadvertently tanked my killer MMR while being friendly in the anniversary event, I started getting lobbies of pure newbies and that just wasn't fair or fun for anyone.
I'd also like to know if you guys would consider tossing match outcomes if another player dies less than 60s into the match? I know something like this exists for DCs, but suiciding teammates is generally an insta-loss and doesn't help the solo queue slog at all.
2 -
To be honest, the pip ranking system is far better at determining skill than say someone who does nothing the entire match and still escapes, or someone who was hooked 2 times and still escapes vs someone who was never hooked and escapes. An escape or kill should be the final factor of the calculation to ensure points are given, but working gens, escaping a chase, not being hooked should determind how many MMR points are earned if the survivor escapes.
Also, I played over 20 games this week as survivor and lost all but 2 due to other survivors going down without a gen or 1 gen being completed. Hatch doesn't count as a win as we were told. So what's the point of MMR if you're a skilled player but lose because you get hatch or die because everyone else died early on?
Its not hard to add some variables like I mentioned, points for skilled plays, and it seems the devs are thinking in that direction but still I worry the mark will be missed. Skill is to be determined by how many gens you and the team worked, escaping chases, not being hooked, and then totalled after you escape. If you played skillfully, you should still earn points but not nearly as many as you do to escape. Like if you're the last one to be hooked and 4 gens were completed, more points earned than first hooked at no gens completed which should lower your mmr.
There is also the fact that players with more hours and higher levels have more skill. But losing because of teammates shouldn't be considered in the mmr factor. A player shouldn't be punished because they got hatch or died last because their teammates tipped the match in the killer's favor. Just some thoughts i hope the devs will take into consideration as they work on this.
3 -
For the first part, I'm afraid I don't have an answer to that just yet. Our usual MMR wizard is off at the moment so I can't bug them for an answer. That said, that's a very difficult problem to tackle since trying to discern what is and isn't your fault is extraordinarily nebulous and hard to put into code. Going back to variables, there's so many things you would have to consider to determine if you played so well that you shouldn't be there. I'll make a note and try to follow up when I can.
For the latter part, the matchmaking system already somewhat does this. Whenever a Survivor dies or DCs, the game understands that the other Survivors' chances of Survival decrease dramatically. As a result, your rating will decrease by a smaller about for each teammate that's already dead (the game does not expect you to win as the last one standing).
8 -
As far as the long chase thing that’s why the devs were looking into treating the survivors as a group. If you manage a 20 minute chase and it helps everybody else get out then your MMR would go up as a group is the idea.
0 -
The emblem system sounds better on paper, but it has the very flaw I was talking about above and is the perfect example of why introducing more variables isn't always the right answer for a matchmaking system. The 'win' condition (a pip) has no clear winner & loser. Both sides can pip in any given match, and on average people would pip more often than they would depip. This meant that over the course of the month, everyone would climb towards the red ranks as long as they played enough. It might take less skilled players a little longer to get there, but the pool of players would get so diluted that being rank 1 ended up meaning nothing.
This does not work in a matchmaking system since everyone would eventually end up in the same place, effectively matching anyone against anyone. The only thing that somewhat prevented everyone from ending up at rank 1 were rank resets, but again, this also suggests that your skill suddenly drops on the 13th of each month.
It may better reflect your performance in a given match, but it simply will not work for matchmaking since any system that allows both parties to win leads to endless climbing.
About the hatch: Taking the hatch does not count as a loss. Hatch escapes are considered a 'null' - meaning the game treats them as if they never happened. (Note: Bearing in mind that each match is treated as 4 separate 1v1 matches at the moment. Escaping through the hatch as a Survivor nullifies the whole match, but the Killer can still gain rating from the other three Survivors they killed.) Nobody gains nor loses rating by escaping through the hatch.
To the point of more hours/levels indicating more skill, I'm afraid this simply isn't true. Yes, having more experience playing and more perks to play with can improve your odds of winning, but everyone has a different skill ceiling. Some people, try as they might, will play the game for years and peak in the average range (and that's okay!). Some people play more competitively than others, and that's great too. No matter which game we're talking about, there's a reason why everyone who's been playing a game for a while can't suddenly make the jump and start playing in tournaments. Everyone's different, and there are absolutely people who have been playing for years in the average range.
7 -
I hope in the future that SWFs can be matched more fairly. They get so many extra game advantages when they use voice communication, but it feels like they are getting matched up as if there were all solo survivors.
Maybe SWFs should have a higher MMR soft cap, and their MMR can be adjusted higher for purposes of matchmaking.
0 -
Thank you for the answer! I didn't know that it took dead teammates into account at all. And that by itself partially addresses the ELO hell issue by punishing you less as long as you last longer than any derpy teammates. While I agree that it's impossible to determine what is and isn't someone's 'fault' outside of a case-by-case basis and there's not many variables you could meaningfully look at, the idea that your MMR takes less of a hit if you last longer might already help even out the balance when a strong player ends up in a weak lobby. Certainly it does if you end up in a lobby where the other players give up. I'm glad to hear that.
1 -
First part is super interesting to know more about! Please, let us know or make a blogpost about that, if you can.
1 -
I like how all these explanations sound in theory. But somehow they don't translate into reality. I'm supposed to "win" half of the time. - So how do 20+ deaths in a row happen? How come my escape rate sits, consistently, around 30%? How come I keep encountering opponents who very clearly have a better understanding of the game than I do (not even taking perks and add-ons into account)? Something just keeps not adding up when I compare how it is supposed to work to how it actually ends up working in my experience.
1 -
There's a couple things that may be contributing to that. For instance, the game build up confidence in your rating the longer you play. When you start out, it radically adjusts your rating each time you win and lose, but over time, it adjusts your rating less. This is because the game's trying to get a feel for where you should be, then trying to fine tune it the longer you play (since you're typically not going to get rusty overnight). However, big balance changes can cause your escape rate to drop, and it will take some time for the system to adjust.
Outside of that, there's some other conditions that can cause your rating to decrease slower than normal. For example, if you are the last Survivor to die, your rating will be adjusted less than if you were the first. Or if the Killer's rating is significantly higher than yours, the game expects you to have a lower chance of Survival so you'll lose less rating if you die.
8 -
I imagine team MMR will be something like giving draws to the survivor if two or more survivors escape exit gate with the killer losing mmr equal to two survivors escaping. The two survivor that escape go up in MMR. 3/4 kills is win for killer, the final survivor escaping through exit gate or hatch during end game collapse will be given a draw instead of a win/loss to prevent the whole sole survivor+wake up farming MMR.
that is how I would do it. as Peanut said, escapes are proxy to skill, they are not skill itself. The skill itself is doing actions needed to win. In patrick's analogy for hockey, hockey uses goals to determine victor, but in order to score goals, you need puck carrying skills, passing skills, goal shooting skills in order to score goals and beat your opponent. In DBD, you need generator skills, chasing skills and altruism/teamwork skills while the killer requires other skills that vary from killer to killer to get kills.
1 -
Lowkey the system we had before the current one was great, I mean sure the ques might've took longer, but you actually played against and with people similar to your level.
Rn it's the same system until you hit a threshold. Then it's back to dbd rank system days.
Atleast that's how it feels and what I think about it.
1 -
Thank you for explaining patiently and clearly ! I feel like I understand better how the system works.
4 -
This honestly wouldn't be a bad post to pin
Way to many people misunderstood how the system works
5 -
@Peanits any ETA about MMR changes? or improvements for solos, because, honestly, i play solo and every match i cant play nicely because the mmr killer level agaisnt me and my random teammates is complete desbalanced, and is hardly ever my matches as survivor are enjoable.
0 -
I think MMR does not work AT ALL and there must be something seriously wrong in the way you collect metrics if you truly believe the average chances of survival are 50%. In particular I wonder if you can explain:
- How is it possible that random players like me can get a 36-win streak on killer (recorded, all 4Ks, and I can reproduce it).
- Why killrate is used as a metric for game balance since the MMR's exact job is to bring it to 50%.
0 -
What killer were you using?
0 -
SWFs can be just as garbage as solo... just thinking of my 2 friends, lol
0 -
Nurse (link to the thread I made back then)
But shortly after I also got 10+ streaks with Trickster (easy) and Bubba (more difficult)
0 -
Yeah but what about those games where you're doing good chases and pushing gens but your team just isnt...
Dying isnt what bothers me, it's the fact I'm put into pointless games I had 0 chance in because my team for some reason refuse to do gens and the game punishes me.
If I play bad sure, but I shouldn't be punished for stuff completely out of my control that's all
0 -
Lowering your MMR isn't a "punishment", it's just setting your rating so you go against slightly easier opponents next time. If anything getting a lower MMR when your team loses is a consolation prize.
4 -
It does work, I have 10 minute queues and face majority 4 MAN SWF
Edit: My fresh, nurse account that has 12 losses (like 8 ddosed disconnected) and a streak of 1126 games, followed by a current standing 761 winstreak I am currently playing.
0 -
I have 10 min queue too, and I face sweaty SWFs occasionally but then inexplicably I get a game like this:
0 -
Yeah well my games are constantly like this no matter what, your at a MMR softcap where you are probably in a very large pool still.
0 -
Well still you're kind of agree'ing that the MMR does not do its job then if you get 10+ winstreaks. While for you it can be explained because you're a Nurse tryhard and likely at the top of MMR bracket, but me I'm a baby Nurse with 30 hours and yet still my killrate right now is 99.8 % (lul maybe even higher than you noob)
0 -
If BHVR was struggling to make a good matchmaking system from the ground up and we were just suffering through the process, this would be all well and good.
But that's not the case.
The Emblem system and having players matchmaking within their own rank color was such a far more enjoyable experience than what we have now. At least we knew what our rank was back then, matches were much more consistent, and survivor team composition for SoloQs was actually pretty good.
If they kept improving the Emblem system we'd be in such a better spot today, in my opinion.
They took a good system and replaced it with a worse system and still have yet to get it to work even close to on par with the old Emblem system. SoloQ is in the worst place it's ever been because this SBMM system does an extremely poor job at team composition. As a SoloQ going into a game today I have absolutely no idea if I will be paired with survivors who can loop and get stuff done or scared little baby Neas urban evading in corners of the map. There is no consistency as far as I can tell.
2 -
It is a punishment, its saying I did bad and will put me with bad players more often.
I dont wanna verse killers way below my skill, and it's not even the killers that are making me lose it's the survivors.
I want a challenge, I want to verse good killers with good builds and mindgames. I dont wanna verse easy killers... it would bore me and if I get paired with even worse teams I'll definitely pass. I'd hate to grow low mmr where no one does anything
1 -
On the other hand maybe your solid plays and chases help everybody else escape and your MMR goes up in that revised system compared to now when it goes down. In that case you're getting "rewarded" for everybody else getting out, right?
0 -
@Peanits is there any chance we can get some clarification on the MMR softcap and how many people are in that range? There seems to be a rough belief that up to 25% of the player base are at or exceed the MMR softcap. Would we be able to get statistics about the overall number of people at each MMR range?
Some of the raw numbers were part of the MMR tests from a couple of months ago, but the numbers didn't really have any context because nobody really knows the distribution of the numbers.
0 -
I think thinking about MMR increases and decreases in terms of ‘wins’ and ‘losses’ might be muddying things a bit. I conceptualize it as something like a range of skill-fullness in a given action. Things like:
Seconds(n) spent 10 meters away from the killer without being visible in their viewport and not concluding in a chase state
Times generator progress was interrupted
Chase escapes where the chase took longer than (n)seconds, or included (n)meters of travel
By calculating various categories of skill sets, matchmaking could then make some preferential matching (just preferential so queue times aren’t affected) like survivor players that demonstrate good looping skills with killer players that are roughly as good at catching survivors in a loop. Or someone who focuses completing gens like crazy with someone who typically applies good gen pressure. The trick would be figuring out the key behaviors and the categories they feed into (and having the logging in place to determine it).
1 -
Awesome explanation!
0 -
If this is how it works, would it be an idea to losen the strictness of MMR adjustments at least a bit after each midchapter/chapter? Because everyone needs to adjust and especially people that stop playing for a few months might be confronted with a big bunch of changes.
Like new killers that you don't play against from the beginning but when most of them already got a feeling on how to use the power while you don't even know how the power works nor which addons are available. Speaking of addons, also when perks are reworked and you get info, explosions, blocked gens and status effects of perks that you barely saw before last patch
0 -
MMR is a flawed system in DBD because of so many variables and the fact the game seems to balance the majority of killers (clumping) to match with a top 4 man SWF. The majority of killers playing are at a predicted MMR skill (due to inherent game flaws) around the top percent of survivors playing. This puts a tremendous strain where there is not enough survivors that are like robots to use mathematic calculations on the fly to determine the absolute inch necessary to time arriving to pallets perfectly. So MMR never works because there is not enough survivors playing nor are they enough at various MMR levels. I have no idea why BHVR is so stubborn to keep this unfun system in place. MMR is more suitable for symmetrical games where both sides always have a fair chance at winning and it is always due to skill and not "luck" or what the opponent does or brings (especially the killer).
Survivors are at the mercy of the killer in order to rank up or down. The system needs to be abandoned in favor of the old ranking system so survivors feel accomplishment to playing while dialing back very bad and unfun killer buffs and buffing survivors and survivor perks. The game is getting less and less fun the longer MMR stays in place.
0 -
I like how different people see different things from this. Personally, I see being able to lead the killer on a long case as less of a skill thing than being able to do gens, unhook/protect/heal allies, sabotage hooks, get saves etc. Which is why when I play killer I tend to gives that player hatch as opposed to the one who ran around the map or clicked their flashlight qt me and begged to be chased. Who actually contributed to the teams success more? I get that some would say that if a survivor looped the killer for 4 gens, they were the hero of the match as they enabled the other 3 to secure the win. But without the other 3, it would have ultimately been a loss. Plus it's on the killer for not recognising and breaking the chase. Personally, if you click at me, you'll be ignored the entire match and get to leave with your 9k bp.
2 -
Yeah I always played solo and met friends doing so. I felt I was playing with people near my skill. But I would still mostly play by myself.. I really enjoy the peace. Now all I get are potatoes. So the only time I play is with a swf. So I’ve almost completely stopped playing. People don’t really talk about ranks.. they bring up the grades. Idk the emblem rank system should be brought back and improved. But I think most really enjoy the swf I just want my solo experience back but I don’t think they ever will go back. Sure you can argue that ranks didn’t completely show your skill but I was almost always a rank one and my teammates were pretty great and the killers too. I was challenged. Now it’s not the killers as much as it is the survivors screwing me over.
1 -
something, something, 5 gen chase not skillful, something, something, hockey -Idk BHVR devs
1
