SBMM/MMR Should Measure Match Performance, Not Just Kills and Escapes
I think DBD’s current SBMM could be improved by looking at more than just the final result of a match. Escapes and kills are not always reliable indicators of skill.
For survivor, escaping does not automatically mean someone played well. A player can hide most of the match, get hatch, bring a key, or be carried by teammates and still technically “win.” At the same time, a survivor can do gens, take chase, make safe saves, protect teammates, and still die.
For killer, kills are not always the best measurement either. A killer can have low pressure all match and still get kills at the end through NOED, camping, tunneling, or survivor mistakes. Meanwhile, a killer can get 8–10 hooks, pressure the whole team, and still end with only a 1K or 2K.
A better SBMM system should still care about winning, but it should also track actual match performance.
For killer, that could include things like:
- Raw hook count
- Unique survivors hooked
- Hook-to-gen ratio
- Chase efficiency
- Pressure before endgame
- How many hooks happened before gens were completed
For survivor, that could include things like:
- Gen progress
- Useful chase time
- Safe unhooks
- Protection hits that actually matter
- Healing value
- Low dead time
- Team contribution before escaping or dying
The biggest factor, in my opinion, should be gen-to-hook ratio. That shows the actual flow of the match better than just the end screen. A killer getting 6–8 hooks before the gates are powered is very different from a killer getting 2 hooks all match and then 2 kills at the end because of NOED.
Devotion level could also be used as a secondary matchmaking factor. Not as a direct skill rating, because devotion does not always equal skill, but as an experience filter. A Devotion 45 player being matched against a Devotion 12 killer can still feel mismatched because the experience gap is massive. MMR should matter most, but devotion/playtime could help prevent extreme experience gaps where possible.
I’m not saying SBMM can ever be perfect, because DBD has too many variables for that. But it should measure pressure, contribution, chase value, and objective value, not only kills and escapes.
Better matchmaking should reward how the match was played, not just what the final scoreboard says.
Comments
@popi_mar
I like the ideas you mentioned!
DBD is probably too complex for “perfect” SBMM, but I do think measuring overall pressure and contribution would paint a more accurate picture than kills/escapes alone.The final scoreboard often does not fully reflect who actually controlled the pace of the match.
Completely agree, the current matchmaking does not work. There is an abysmal difference in skill level in every aspect between some Survivors and others within the same lobby/match.
This has been floated many times, and MMR is supposedly being reworked, but this much data will be hard to measure. Chases, for example, are cut short even when they're still ongoing by distance being created or LoS being broken. Judging "useful" chases is hard. I bring Lucky Break and Iron Will. I get hit, I turn a corner, and I disappear, then I watch the killer look for me around a tile and in lockers. I'm still wasting their time, but the game isn't registering it. I don't know how the game would fully and accurately judge a protection hit either. Was it a protection hit, or did the killer just run into someone else?
On the killer side, are you saying things like unique hooks should raise your MMR even if you 8 hook and get 0k? Wouldn't that just mean tunneling killers get to stay at lower MMR and have an easier time cheesing? If anything, I want those people sky rocketed upward. If they want to play sweaty, let them sweat at the top. If you're someone who gets 7 or 8 hooks before a kill, chances are you're doing fine otherwise, even if you lose some matches. If anything, a bigger sense of accomplishment in-game is needed for these situations.
Though I agree kills/escapes need to not be the only factors, I think it's also very hard to judge anything else. That's why the system is as it currently is. I do think hours/devotion should be accounted for. A 1k player shouldn't land a 10k opponent.
You already don't gain MMR from hatch escapes. A player who gets hatch consistently will have a pretty stagnate MMR growth rate.
This is probably my most unpopular opinion when it comes to this game, but I'm not entirely sure the DBD community wants what it's asking for when they're talking about this.
Since the MMR is invisible and thus not a way to gauge your personal skill level or show off to others, what you're really asking for is for having harder survivor opponents as killer because you had an 8-hook, 0-kill game. Or a harder killer opponent as survivor because you just got tunneled for 5 gens.
The community have been begging for the measure of skill to be changed to something that actually measures skill ever since this insanely terrible “kills/escapes = skill” system was implemented. BHVR have said for a long time now that they are “planning” changes but like so many other things that don’t directly impact their revenue it will probably just be forgotten about. I have given up on BHVR prioritizing anything unless 1) it brings in more revenue or 2) it prevents loss of revenue. From a business perspective I totally get it, but it sucks that these changes that would make the game a much more enjoyable experience for most players are probably never going to happen.
@cogsturning Yeah , I agree with you on that. Some things like chase quality or protection hits would be hard to measure consistently for sure.
My point is more that kills and escapes are also imperfect. A survivor can play well and die, or play badly and still escape. A killer can create pressure, get 7–8 hooks, and still end with low kills.
So I’m not saying every action should directly raise MMR. I’m saying broader match data could be used as a secondary layer, even if it’s not perfect. It would still give a better picture than only looking at the final result.
@Zakon05 I get what you mean , but I don’t think people are asking for “harder matches” just for the sake of it.
The issue is that kills/escapes alone don’t always reflect what actually happened in the match. If a killer gets 8 hooks and 0 kills, that doesn’t automatically mean they played badly. And if a survivor gets tunneled for 5 gens, that also doesn’t mean they lacked skill just because they died.
I’m not saying MMR should reward every little action or make every match harder. I’m saying the system should have a better way to separate a bad performance from a strong performance that ended in a bad result.
Since MMR is invisible, it’s not about showing off. It’s about matchmaking being more accurate. Right now, the final result can be misleading.
@Your_Dad_Playing_DbD Exactly. That’s basically my point too…
The current system treats kills/escapes like they automatically equal skill, but the match itself can tell a very different story. A survivor can carry pressure, take chase, do gens, make saves, and still die. A killer can get strong pressure and multiple hooks but still end with low kills.
I don’t expect BHVR to measure everything perfectly, but the system should at least be better than “final result = skill.” Match performance should matter more than just the end screen. Since MMR affects the quality of matches, it should be based on more than the simplest outcome.
I've been saying for a while the MMR is a joke and doesn't exist and honestly it doesn't. Remember behavior put out the stats and said only 18% of the players are in "high mmr". Until they drop the mmr ratings which they probably never will because they don't want their game completive, we will never know what defines high mmr and what's realistic is the fact that their prioritizing que times more than anything. Cause again, paraphrasing something they said on stream, "We could give you all accurate MMR but you'd be waiting hours for a match.
mmr could be better and it atleast needs a softcap because I shouldn't be loading into my matches as a 3-4k hour player playing with 200 hour players. It's just not fair to them or me.
Pardon me for a horrendously stupid question your about to hear. What is MMR? i have heard its like a way of skill based matchmaking. I dont understand it that much
The way I've always understood MMR systems is that they're not supposed to measure a match, they're supposed to measure many matches. To elaborate, my understanding of how and why an MMR system works the way it does goes like this:
If a player plays one match and wins it, that could be up to a variety of factors, any number of things could've contributed to the win outside of that player making smart decisions and executing on them properly.
If a player plays ten matches and wins six, though, that starts to indicate better that they're playing well. A variety of factors still applies, but something has to be going right for their approach to the game.
If a player players a hundred matches and wins sixty, that's even firmer evidence, at that point they must be doing something right- it has to be up to their decisions at least as much as it's up to random flukes.
So, that'd explain why DBD's doesn't currently measure match performance and imo suggests it wouldn't really matter much if it did, since one match doesn't move the needle on its own. You'd just be kinda adding in more avenues for false positives and false negatives, really.
It also seems, at least to this layperson, that it's better future-proofing against balance problems, that way you can just safely address whatever problems exist in the game without worrying about how it'd affect the matchmaker.
I'm sure letting some of those variables act like modifiers to increase or decrease the amount of MMR gain/loss after the win/loss could make a degree of sense, though. Still, what I'd want to see in that regard is for us to have a better system to measure your performance after a match that isn't the MMR system, like overhauling the emblem system. That'd be the best of both worlds, people get rewarded for good play on a match-by-match basis but it doesn't mess with the matchmaker.
It's a hidden point system used for matchmaking. You gain points when you kill/escape through gates. You lose points when you get no or low kills/die. The amount of points you gain or lose depends on the amount your openness have. Defeating a higher level opponent will boost you more.
It should mean you get matched with people similar to your skill level, but even if the system was a true test of skill, it seems to just match people almost randomly at times.
If I recall, pips used to increase your rank in the old system, and those are based on doing a little of everything. I don't know if I'd adapt that exactly as it has its issues, but that's probably a better start because you have to at least do a variety of things regularly. The tricky part, as I see it, are killers who hard cheese. If you 4man slug with no hooks, you won't boost, and as far as I'm concerned, those people need to sky rocket to the top of the sweat tower.
There's an unpopular opinion thread going on, but probably my most unpopular opinion is that BHVR is right on how they reward MMR (well, survivors should be treated as a team, but I don't think that would make a huge difference) and the hockey comparison actually makes sense.
Games (whether video or sports) usually have some type of score objective / win condition. Other stats are interesting, but the only one that is truly relevant is the win/loss.
Usually the argument's against are something like 'so I could loop a killer for 8 minutes, but be left on hook and lose?' Sure. The thing is, a single trial is not meant to be the overall arbiter of a player's skill. Again, for many games a single data point is not really reflective of overall performance. That trial will ultimately be irrelevant when the data set is hundreds or thousands of games.
There are things they could address about MMR. Backfilling being one of the biggest. Probably consider increasing the time it takes to find a good MMR match. Consider how MMR is adjusted when players take a long break or play a new killer. As I said, treat the survivors as a team.
Unfortunately, given some of the other design decisions of the game, its always going to have more problems than other games, but the issue is not just boiling it down to win/loss.