What is the point of sbmm
I do not understand
so looping a killer for 2-5 gens, healing, double pipping etc but getting facecamped after means mmr goes down, so worse teammates and harder games
like can anyone explain how the hell this was ever considered something smarter than a dead squirrel could come up with?
the game literally rewards ######### survivors who hide all match and do not have skill while punishing players who actually improve
am I kissing something? This whole escape or not instead of emblems and points sound alike something someone came up with that’s narcissistic and arrogant and can’t accept it was a bad idea
please for my sanity someone explain HOW this ######### makes sense
just had a game where no one but me even touched a gen for nearly two minutes. Looting chests and hiding. Thirty minutes wasted for the killer to get a 4k. Plague had her power the entire match. Clearly new players with unleveled perks and in at 3600 hours on the game and can’t get a good match unless I avoid solo queue entirely
fully convinced at this point mmr is hidden to justify how bad this system is. So sick of people hiding all game for hatch while the players who do gens get killed
wAs It EvEn A GoOd PlAy. Watched my buddy loop the killer for FOUR gens yesterday while the two megs went down Everytime instantly. But they escaped so they get better teammates next game and my buddy has to. Have his mmr lowered after carrying those two potatoes
tl;dr can anyone give an honest justification for the current escape/kill system as opposed to the old emblem system ?
Comments
-
tl;dr can anyone give an honest justification for the current escape/kill system as opposed to the old emblem system ?
not defending the new system, but the old system was full of flaws as well:
- Anyone could reach high ranks with enough playtime, getting emblems doesn’t equal skill (neither does escaping and getting kills)
- reset every months didnt make sense for a matchmaking system and just made it more random
- it didn’t even work at all (rainbow rank matchmaking was pretty much in every match)
2 -
I do not believe the new system is calculating the right amount of MMR. I have been faced against a team that has over 7k-8k hours in the game while I still have 1k hours. On the other hand, I faced people that have not even have 100 hours. So yes while the old system has its flaws, the new system does not change that flaws and I think it makes it even worse than that.
2 -
Its simple: Even tho a player can run a killer for 2-5 gens, healing, double pipping etc but getting face camped. They wont get that every game, they will in fact perform better than the players who escaped that match you have imagined here.
Dont look at MMR in examples of one match, look at it over many matches.
5 -
When you lose in this kind of system though you get worse teammates making it harder to win every single time was my point. You loop five gens and die next game you have WORSE teammates
I had a plague match earlier where I was the only survivor in gens for THREE MINUTES and the plague had her power all game
my swf hasn’t lost to plague in probably a year.
This was stupid ######### were they hiding injured and sick in lockers for? nothing I can do when every player is feeding her the corrupt all match and not on gens
3 -
Thing is.. playtime does really not equal skill either. Using that as a sole metric can’t justify if a matchmaking system works or not. Trust me, I got plenty of hours and am still playing very poorly
2 -
The idea is that one match isn't going to swing your MMR all that much at all. Sure, in ONE match you may do a ton of valuable work and end up facecamped to death, but that's not happening every match and the few times it happens shouldn't statistically make that much of a difference to your overall MMR.
If you can play that well consistently, odds are you're escaping more matches than you die in, and that means your MMR goes up overall. This design of MMR is not unique, btw, most games that I'm aware of use some variation of "wins go up, losses go down" as the basis for their MMR.
0 -
Yes, but it you lose consistently again and again, while looping the killer for 4 gens, then you have good abilities in some places but lacking in others.
1 -
Will try to keep this short, but matchmaking is incredibly complex so apologies if this ends up being long.
There's a couple big misconceptions with MMR, and probably the most common one is how it estimates where you should be. No matchmaking system in any game- even games like chess which pioneered these ranking formulas over many years- is able to take a single match and accurately assess your abilities. There will always be too many factors at play for that to be accurate. Even if you use a different metric or a bunch of metrics, like chase time, hooks, pallets used, generators completed, healing, etc., there are still outside factors that need to be considered. Was your opponent (the Killer in this case) playing well, or were you only doing well because they were making a lot of mistakes? Likewise, did you do poorly because they're amazing at the game? Did your teammates do their part to help ensure your good plays worked, or did they let you down?
Even if you had a human sit there and watch your match, it would still be impossible for them to stick an accurate number on you after watching just one match. The best they could do is a ballpark estimate. They don't know you, they haven't seen you play against higher caliber/lower caliber opponents before. They only have this one match to go off of.
The matchmaking system is no different. It cannot tell how amazing you are from a single match because that is an impossible task, so it doesn't even try. Instead, the system is designed to look at your performance over the course of many matches. This way, due to the law of averages, you will gradually move towards the place where you should be. You might lose two matches and win the next four. You're up overall from where you started.
And because there's no way to accurate take every factor into consideration, it does this by using a proxy condition. Yes, kills or escapes are not the only thing that makes a good player good! But if you're amazing in a chase and can last for several minutes, more often than not, that will lead to a win. If you play well, your chances of escaping will rise. Or if a Killer plays well, their chances of killing the Survivors will rise. This isn't something that's exclusive to Dead by Daylight either, just about every game will use a similar metric for matchmaking (even if it doesn't show it). You'd be hard pressed to find a game that successfully factors in dozens of things rather than using the outcome of a match.
This is also important because the matchmaker's sole job is to find you matches where you have a roughly equal chance of winning or losing. It doesn't care how you do it, whether you played super well or you used every crutch in the book. It's a cold, emotionless machine that's solely trying to put you up against opponents where your strategy- whatever that may be- works about half of the time. It is not trying to pass judgement on you as a player, not trying to make you feel bad for losing or claim you're amazing for winning. It's simply looking at the facts: If you won, rating goes up, if you lost, rating goes down. And over the course of several matches, you'll almost always end up moving in the right direction.
This is a big part of the reason why your rating is not shown. It's not meant to be something you obsess over or focus on. It's sole purpose is to find you fair matches. It will go down sometimes when you don't feel like it should, but it will also go up more often than not if you are truly below where you should be.
It's crucial for a matchmaking system to have a clear win condition. The winners rating needs to go up, and the losers' ratings need to go down. If it's a draw, they stay the same (although this isn't common). If this is not the case and both people can win simultaneously, you end up in situations where people can endlessly climb in rating rendering the entire matchmaking system useless. It would quickly become a reflection of someone's playtime more than their actual abilities.
The emblem system is actually a perfect example of this. Since emblems are independent of the other side's performance, it's possible for both the person who 'won' and the person who 'lost' to pip. The result was exactly what was expected: By the end of a month, pretty much anyone who played enough would find themselves in the higher ranks because on average they were pipping more than they were depipping. This meant: A) By the end of a month, a large number of people with a wide variety of skill levels were being matched together, making matchmaking feel random, and B) By the end of a month, the lower ranks (people who had not played much) would suffer from excessively long queue times since the majority of people had already ranked up past them. If the matchmaking system is producing long wait times for some and horribly uneven matches for others, it's failing to do its job.
In short, don't sweat it if you "lose" by the matchmaker's standard. It expects you to lose some matches, and it knows that you will win matches to offset those losses. If you think you did well, you did well. Don't let an invisible number that you can't see take that away from you.
Post edited by Peanits on15 -
so looping a killer for 2-5 gens, healing, double pipping etc but getting facecamped after means mmr goes down, so worse teammates and harder games
Skill can be conceptualized in different ways. The one you pick determines the gameplay you're encouraging. Yours appears to be (and mine as well) knowledge of the game and proficiency in its various aspects.
On the other hand, BHVR values 'winning'. It's a result-oriented, individually-focused, goal-by-any-means-necessary view. It boils down to only you escape (kill), no matter how you achieve that (within the ruleset). It's a minimalist, ultra simplifying approach. It's not illogical per se, but it carries consequences that you're experiencing.
The main point is that they probably chose it because it makes their life easier, but not necessarily ours.
So, while you can be a team player and do all the best plays to maximize your team's chances to escape, but not necessarily your chances specifically, you can very well be classified as less 'skilled' than someone who's super selfish and throws their teammates under the bus in order to ensure their own survival. They win more than you do. They attain the game goal more than you do. Simple as.
What this does is to lower the skill-ceiling and compress players of different abilities in a smaller pool. The god-tier looper who makes also sound macro-level decision might very well be labeled as less skilled than the immersed gens-before-friends jockey who rushes the objective and gets out with Sole Survivor in their back pocket as plan B. The cracked killer who knows the ins and outs of their power, has amazing game sense, can pull off incredible skill shots, knows all the mind games in the book and goes for chases and hooks might very well be considered as 'skilled' as the guy who runs the strongest add-ons, bloodlusts their first down and facecamp with Deadlock to force an easy 3v1 that they can cap off with a final NOED kill if need be. They kill just the same amount, hence they win the same, hence they're equally 'skilled'.
That's what the devs have chosen for their game, which is within their purview, obviously. It certainly is really disappointing, as far as I'm concerned.
4 -
Thanks for this well-written explanation! I certainly feel better now.
1 -
This is literally why camping/ tunneling as killer and dcing in hook to get on to the next match and hiding for hatch happens though. This idea of winning=skill encourages the same behavior (no pun intended) that everyone is always angry about that being said many good points were mad, such as the long queue times and how people could pip with emblems and the other side win so match making was thrown off
anyway thankyou to the community manager was giving such a detailed answer to my admittedly over the top energy question lol
2 -
This content has been removed.
-
This might be asking for more information under the hood, but I think there are still some misconceptions about how the system works. Specifically, the 'there are problems if both sides can be considered winning simultaneously' part. (I don't disagree, I just think I misunderstand something here).
The system seems to assume there are only two parties. In a 'team of survivors' vs killer scenario, there is only one winning side (or a draw), so assigning a winner is just one team or the other.
But the way I understand it, survivor matchmaking is based only on your individual escape, not as a team. The game isn't considering win/loss in 4v1, rather 1v1v1v1v1.
Take, for example, a match where 3 survivors are sacrificed and the last survivor escapes via the exit gate. In this example, 3 survivors 'lost', the killer 'won', but we also have the survivor who escaped 'won' as well.
Does this not fall into the troublesome category you described? Both the killer and a survivor won the match, despite being opposing teams.
Or is likely I'm misunderstanding something here. Any clarification is appreciated.
1 -
Happy to clarify!
In this scenario, it's treated as if the Killer 'won' 3 separate 1-on-1 matches and 'lost' a fourth. Overall, they would gain rating.
We actually have team based ratings as well, but this only affects things on the Survivor side. If you were to die and the other three Survivors escaped, you would lose less rating (it is assumed that your actions helped them escape). The same goes if you escape, you will gain more rating for each Survivor that also escapes. On the flipside, if more Survivors die, you will lose more rating if you also die and gain less if you happen to escape.
The key is that these each offset each other. There's an equal positive effect for every possible negative effect, so it ends up being more or less neutral in the end. The only difference is that it can feel less punishing when you get chased all match, die, but your teammates escape.
3 -
Ok that makes more sense. I was under the impression that the survivors results were independent of each other. Having the 'match result' factor in was the piece I was missing.
Thanks for the clarification!
0 -
By the end of a month, pretty much anyone who played enough would find themselves in the higher ranks because on average they were pipping more than they were depipping.
Isn't this an issue with the safety pip system though? The problem with the old system was that you really couldn't lose pips. Thus even if you weren't very skilled you would slowly climb. What this did was make the early reset of going back to green ranks and quickly getting to red ranks result in fairly solid games but as time went on in the month the quality of games quicky fell. Isn't removing the safety pips and making it harder and harder to pip the best solution to matchmaking in this game? Focusing on the smaller things like chase and doing gens instead of letting your team die while you hide in the corner waiting for hatch. I feel that working on the grade system would be overall better for the games health than a kill/escape system.
2 -
The safety pips certainly wouldn't have helped, but the bigger issue is that it's not possible to 'balance' that system in a way that also makes it roughly even for matchmaking. A Survivor can do things that earns them emblem points without the Killer losing any and vice versa, meaning the end result of the match could still be considered a win for both. Even if safety pips were removed and the system was perfectly rebalanced so that both sides cannot win simultaneously, the bar for pipping would have to be frustratingly high in order for it to even remotely work.
But even if that were the case, you would still have situations like the OP described. You may have been chased all match and done really well, but because you didn't fix generators or unhook anyone, the emblem system might spit out an unsatisfying result.
And in the process you'd lose one of the bigger advantages to the current system: Your result wouldn't be affected by your opponent's rank. Whereas the current system can look at a match and say, "This one might be tough for you, so we won't lower your rating by much if you do end up losing", the emblem system is completely independent. It solely focuses on your performance and your performance alone, it doesn't factor in how good your opponent may be.
In the end, this also has it own pitfalls. It would also feel very frustrating to finish a match after killing three Survivors only for the game to say, "Yeeeeah, but you didn't do it fast enough so I'm giving the win to the Survivors." - no system is perfect, and a very strict emblem system would likely be just as frustrating if not more.
0 -
This was I think one of the best explanations for the MMR system and I personally thank you for that. I am also working in the field of Data Analysis and creating machine learning algorithms and I know that there are several input factors, loss weights, and variables that affect the overall model. I am glad there is a good explanation written here.
2 -
What I don’t understand - at all - is why the pip/scoring system isn’t a much better measure of skill than kills and escapes. Based on my experience, my score as killer or survivor is good when I play well and bad when I don’t play well, regardless of whether I got kills or escaped the match. Yes, I realize that farming could skew the skill score, but that’s already true with the current system which can be manipulated when killers farm or just go AFK. The emblem system rewards players for doing all the things in the game that correlate to skill, while the current MMR system only uses one flawed metric of kills and escapes.
At the end of the day though, it doesn’t really matter anyway. Regardless of your MMR score, the matchmaking system is so biased in favor of short queue times that we still frequently see brand new players getting matched against semi-pros.
0 -
A Survivor can do things that earns them emblem points without the Killer losing any and vice versa, meaning the end result of the match could still be considered a win for both.
Wouldn't that mean you both move up in rank and face harder opponents. Your team would be of higher skill along with the killer/ the survivors you face are now harder to down and do gens faster.
You may have been chased all match and done really well, but because you didn't fix generators or unhook anyone, the emblem system might spit out an unsatisfying result.
Wouldn't the emblem system actually have a better chance of removing this though? The system already gives credit for gen progress while in chase. Might not be perfect but better than what we currently have imo.
the emblem system is completely independent. It solely focuses on your performance and your performance alone, it doesn't factor in how good your opponent may be.
But isn't that ideal? Like if I move up in rank my opponent is better thus harder to do well in chases. I find the main issue with the current system is in soloq. I find that I'm completely at the mercy of how well my team can do. If they don't do gens, I lose. If they can't run the killer then the killer will never chase me and remove those players quicky, so I lose. I feel that the old system I could work towards getting my pips and it was less of an issue with getting weak teams. The new system feels very frustrating when playing soloq because I feel it's more of how bad/good my team is rather than my personal skill. If I'm better than the killer the killer can ignore me and chase the weaker links and then finish me off last bypassing my skill as a survivor.
2 -
Hey Trollinmon, not sure if you're trollin' but Peanits posts address most of these:
To your first question:
"It's crucial for a matchmaking system to have a clear win condition. The winners rating needs to go up, and the losers' ratings need to go down. If it's a draw, they stay the same (although this isn't common). If this is not the case and both people can win simultaneously, you end up in situations where people can endlessly climb in rating rendering the entire matchmaking system useless. It would quickly become a reflection of someone's playtime more than their actual abilities." Peanits
To your second question:
"And because there's no way to accurate take every factor into consideration, it does this by using a proxy condition. Yes, kills or escapes are not the only thing that makes a good player good! But if you're amazing in a chase and can last for several minutes, more often than not, that will lead to a win. If you play well, your chances of escaping will rise. Or if a Killer plays well, their chances of killing the Survivors will rise. This isn't something that's exclusive to Dead by Daylight either, just about every game will use a similar metric for matchmaking (even if it doesn't show it). You'd be hard pressed to find a game that successfully factors in dozens of things rather than using the outcome of a match." Peanits
To the third point:
Specifically, "the emblem system is completely independent. It solely focuses on your performance and your performance alone, it doesn't factor in how good your opponent may be." and you asked "But isn't that ideal?".
I'll try to answer this one. When trying to match players in a PvP games, you match based on defeating opponents in matches and moving up. Not by demonstrating particular & finite skills that are involved in the match/game.
- FPS Games - MMR based on wins, not headshots
- MOBAS - I believe while they may take KDA into consideration, there is a weighted effect or multiplier based on whether you win or lose
- Racing (do these have mmr) - If they have MMR probably based on who won/lost, not how many times you crash.
The thing those all have in common are that the overall MMR is comparing the opposing people and their results (winner(s) and loser(s)). If all people wanted to do was prove how good they are in isolation, not considering their opponent, shouldn't they be proving themselves in a single player game?
Hope that clears any confusion.
0 -
what is the "win condition" for killers and survivors? just curious.
i think a lot of players would stop complaining if they knew those two goals.
although many would probably argue against it no matter what it is.
i'm assuming the win condition for a survivor is to... well... survive?
and the win condition for a killer is... or at least it should be to kill the survivors if we're trying to avoid the whole... both sides winning thing
if that's the case then one could say that each killer is playing 4 matches at a time while the survivors are only playing 1.
if a survivor dies that's a win for the killer. so in a single match the killer could get 4 wins. 4 losses. or any combination of the 4.
but if that's the case then do bleed outs count? mori's? special executions? or is it hooks only?
0 -
I find the main issue with the current system is in soloq. I find that I'm completely at the mercy of how well my team can do. If they don't do gens, I lose. If they can't run the killer then the killer will never chase me and remove those players quicky, so I lose.
This is part of team-based games. you lose and win as a team. your teammate failure become your own failure.
0 -
My entire point is that matchmaking doesn't work and you are completely at the mercy of your team. I had a game a few days ago vs a Nurse. I ran her for 3 minutes at the start before she gave up. She went to my team and 2 blink'd 2 players. She removed them from the game and the meg hid. I ran her for literally 10 minutes before she gave up and on me entirely and found the meg. She ended the game with a 4k.
So now I ask this one simple question. Show me a single PVP game that is does this. Imagine if League was so utterly broken that a new player is allowed to beat a diamond player. The entire issue here is that if my team is bad I lose. My mmr goes down and the problem is still there. Am I not better than this Nurse when I ran her for nearly 15 mins without going down? Well this system says no. It says that me running the Nurse for 15 mins and not going down means I'm trash because I didn't escape. I didn't escape because of my team and now I get the honor of getting even worse players. The old system allowed me to get a pip and move up in rank to get better games.
2 -
What a counter intuitive way to design a match making system in a game like DbD.
No one is expecting an MMR system to accurately grade your skill level after a single game. That isn't reasonable and almost everyone knows that. That isn't what most players are complaining about. By far the most common complaint about the current MMR system is that there's so much RNG involved during a game, (map, tile generation, player perks, spawns, etc) that often times a "win" or a "lose" can be entirely due to something outside of player control. It especially doesn't help that there are hidden mechanics when it comes to match making that have absolutely no place in a system like aims to make matches fair, such as deliberately sending players who win a lot to a map that they often lose on.
The problem isn't that the game cannot measure every bit of RNG to correctly access whether a player did well. That's not reasonable or even possible. The problem is that DbD is simply not the type of game that any type of MMR could work because there's too many things that influence a match. And as flawed as the emblem system was back when it was used to match players together it did a better job of measuring how well a player did during a game than the current system. And you could actually see it.
There's no good reason why players aren't able to see their MMR ranking. Not being able to see it stops players from being able to see if they improved over a the course of several matches. And makes it impossible to give feedback on if the system even works. Like it or not, players need to see what their MMR ranking is. Not just so they can see how good/bad they're doing, but to ensure that their opponents are at roughly the same level. I think I speak for almost everyone when I say that match making has never felt more random. There's so many games I get as killer where the survivors should obviously have never been matched up with me and get slaughtered within minutes. And others where it feels more fair. But it's all at the whim of an invisible number that doesn't account for player skill, that doesn't account for game RNG, and is by far the most arbitrary and unintuitive way to match players together that I have ever seen in a multiplayer game. And I can't even see it!
I'm going to even begin to explain how having some survivors "win" a match by whilst their team mates "lose" when survivors are supposed to be working together as a team is laughably unfair. This system needs to change. If not replaced entirely. Having it only worsens the experience of most players.
0 -
Except in other team base games bronze players don't get free wins on plat+. If I don't play in an unfun way I lose regardless of my skill. Killer finds me, sees it will be hard to down and then leaves for the easier chases. If they can't hold their own then I lose. My games are basically the killer gets 1-2 hooks or my teams just last 20 seconds in chase. To have a system so bad that you are basically required to matchmake outside the game/bring entire builds around the bad MM is honestly unacceptable.
0 -
Unfortunately, I think you are wrong. I believe it is quite common knowledge in the community that with the MMR is Kills = win for killer, escapes (ie. "surviving") = win for survivor. There was a whole fiasco (not the Lupe kind) regarding the whole thing. This was explained during a livestream where the devs discussed MMR system and addressed community questions. Peanits also explained that above. And indeed, MMR wise, one match of DbD is measured as 4x 1v1s (although it is more nuanced than that with regard to survivor; please see Peanits explanation).
Although, to answer your questions for "what counts" (these are to the best of my knowledge):
kill by Mori = count as killer W
kill by Bleedout = count as killer W
kill by Special Execution = count as Killer W
kill by hook = count as Killer W
Hatch Escape = NEUTRAL (neither killer nor survivor are considered winner or loser) ; again remember this is based on 1v1, so if hatch is involved there are 3 other 1v1 where either Survivor won or killer won.
DC = I am uncertain of, but I imagine it considers the DCer the loser.
Hope that helps!
0 -
it does. thank you. i must have missed that stream.
0 -
I am going to have to shout out my dude Devil_Hit11 for being more concise than my wordy ass.
"This is part of team-based games. You lose and win as a team. Your teammate's failures become your own failures."
I believe all, or atleast most, PvP games you win and lose as a team.
As for the "a new player could never beat a diamond player in league". That is not true. What is more likely true is "a new LoL account cannot be matched against a diamond LoL account". As a new Lol PLAYER could jump on their friend's Diamond Account, get matched with other diamond players, then win by either being hard carried, having their team capitalize on opponents' mistakes, or maybe the opponents aren't trying (diamond players can play casually if they choose).
I think it is unrealistic (and incorrect) to make the assumption that "rank up = better matches". More data points should equal better matches regardless of whether you win and rank up or lose and derank.
0 -
Of course! Happy to help!
0 -
Test whether this is also deleted here (of course NO insults! just my arguments that I provided) if you ask yourself "why aren't there any people here who have a different statments except "bhvr you are the best, you are so right "Arguments or people who have a different Arguments" quite simply it will be deleted and or the people will be banned, so you also hold back the statements you don't like clean :)) I assume that's also a reason to delete it great "Free country, freedom of speech" ITS ALWAYS! unless it doesn't suit you, then it's ok to delete these statements :D
0 -
What I still don't understand is how the system factors in DC's. Considering how often those happen, and people leaving the match early via 1st hook, it counts it as a "kill" for a killer, but the other 3 survivors are usually left in an unbalanced game. Regardless of how well they might be doing. Which also has them depiping a lot... So how is that balanced for the survivors?
0 -
Wait you think a new player on a diamond account can win games in diamond???????????
Buddy if you play a diamond account as a noob you can sit in fountain with the rest of your team since they already opened. Oh boy I sure can win the game vs a Jhin that has 8 kills at 12 mins and is 2 tapping me.
I think it is unrealistic (and incorrect) to make the assumption that "rank up = better matches"
Well that's fine that you think that but that was already the case back in the day. R1 for soloq was the best place to be. Push on reset and avoid near reset to get the best matches.
0 -
its because of matchmaking buffer. if your really good player like 1700+ MMR, you can get teammate that are 1600 MMR or lower. this can result in imbalance matches because you could be facing a killer that is 1800 MMR, your skill-level is of appropriate level for the killer but your teammate are 1500-1600 MMR player due to the mmr buffer which are vastly inferior to 1800 MMR killer player. they lowered buffer significantly because killer's complained that their matches were too hard/matchmaking was too strict. I don't really like the buffer being so low but it is whatever.
0 -
Oh I de-ranked at soft cap because of the amount of hand holding I saw every game with killers. My games for soloq are the worst they have ever been, and that's going back to release. I had the same day as the Nurse game a 3v1 vs huntress. She chased me for 5 gens tried face camping and then we 3 out her with her getting 1 hook. I will get other soloq players around my skill where we destroy the killer and my next game can have players clueless on how to play. My games at the soft cap for killer are also interestingly void of many soloq players. They even stated that soloq is around half the players and yet at the soft cap is mostly duo+
Also might I add that hexy and angrypug are both legacy players that stream. Hexy has been saying you need to be on a watchlist if you main soloq and pug quit the game as a soloq main.
1
