http://dbd.game/killswitch
What percentage of players are at the MMR cap?
One quick question based on the test results that I'm curious about, what percentage of players are at the MMR cap? For reference the devs talked in the article and previously about how the MMR has a built in cap where it treats everybody who is at or above that ceiling as if they were that same rating (e.g. everybody over 1900 is treated as a 1900 for matchmaking). They then go on to say they're lowering that ceiling a bit along with reducing the acceptable range bracket for matches. So for instance, where now it might be everybody at 1900+ can get matched with players down to 1400+, in the new system the cap might be 1700+ being matched down to 1450+, something like that.
Which all raises the interesting question, what percentage of players are actually at that ceiling? I've always just kind of assumed that the ceiling is set high enough that no more than, say, 10% of players are impacted by it, but that's just speculation on my part, I could certainly be wrong. 🤷♂️ So I'm kind of curious what percent of players fall into that "top MMR bracket".
Comments
-
If we're only counting forum users, 100%
21 -
yes. Killers here are all at top mmr as killer and as surv also
2 -
That's a very interesting question.
I'd also be concerned with gains/losses for killers - as it feels like even if you get 2k a game, your MMR still tends to 'creep' upwards.
1 -
While I can't give an exact number (I don't have that handy), I can say that it's very, very few. The distribution is a bell curve. Most people are in the middle and tapers off the higher or lower you go.
But to address your concerns, the reduced range more than compensates for the reduced cap. Players from the middle who couldn't be matched with with the best of the best before still won't be matched with them now.
16 -
I'd actually go further with this question since it's a good one.
Can we get a break down of the percentile distribution for low, middle and high mmr (including what the numerical brackets are). This would genuinely go a long way to helping people visualize just how little of the playerbase is significantly affected with changes to the system.
Secondary to that, would the devs be willing to show us how many players are above the "soft cap" for reference? I don't think there is any harm in seeing collective numbers even if you don't want to show individual players their rating.
1 -
Thanks, Peanits! I figured it was low, your “very very few” sounds like it’s closer to 1% than 10%.
Someone else mentioned a concern about long term MmR drifting upwards, but I suspect that probably doesn’t happen much (since a very low percentage of people are at the cap) plus I’m guessing Behavior periodically monitors how many players are in various brackets in case they need to adjust the ceiling or initial ratings.
1 -
Muddy little Hag fingers crossed I guess, because I'm optimistically cautious, emphasis on the 'cautious'.
I'd be so interested in what the actual gain/loss math looks like, and what MMR I am. I can estimate until the cows come home, but at the end of the day it's a frustrating known/unknown.
1 -
I'd definitely be interested in seeing the distribution (or our actually MMR range, even though that's a pipe dream). It seems like one of the major sources of frustration stems from the fact that players don't know where they are in terms of MMR. I know every high MMR player is here, which is pretty neat.
0 -
Due to the way an MMR system calculates the rating change, that's actually not too much of a concern. The amount your rating goes up when you win is dependent on your opponents' rating. To give a massively exaggerated example, if you're at 3,000 and you beat someone who's at 1,500, the system goes, "Well yeah, of course you beat them." and barely changes your score- if it moves at all. Meanwhile, losing to someone who's way below you is going to lower your score way more than someone who's relatively close to you or better than you.
In theory, you can endlessly climb by consistently beating people of a similar skill level. But when you get so ridiculously high that there's nobody on your level to match with, losses mean more than wins, so there's only so far you can break away from everyone else. The range of ratings can grow, but only if there are enough people to support it. (Nerdy aside: The way the system balances things like this all on its own is very interesting to me, and I have spent a lot of time staring at charts and bugging the people who work on it with all my questions.)
9 -
But how can that be? Every forum member assures me they only play high MMR. /s
Post edited by thrawn3054 on3 -
If it's a bell-curve, you are most likely right. 68% of a bell-curve is 1 standard deviation. 95% of a bell-curve is 2 standard deviations. 99% of a bell-curve is 3 standard deviations.
So the top percentage will either be 16%, 2.5% or 0.5%. My guess is 2.5% is considered the top players.
0 -
This sounds like the stuff my wife looks at for a living - but from a neuroscientific perspective rather than a gaming one. Definitely interesting.
Are there any plans to let us see our own MMRs, or at least give us some rough numbers regarding how much a kill/escape is actually worth? I don't see how this could be taken advantage of and it would be, at least, well - yeah. Interesting.
Shrug. I've always maintained that I play at an intermediate MMR, probably pushing into high-intermediate on my very best killers. I'm almost certainly a fairly low MMR survivor.
1 -
Could you share more stats on the kill rates and the variety of perks, killers, items, etc?
That would be interesting to see, please 🙏🙏🙏🙏🙏🙏🙏🙏🙏🙏🙏
2 -
Also along those lines I'm assuming that when the system calculates how much your MMR goes up or down after a match it's using your actual MMR and not the capped MMR value that's used specifically for matchmaking. So for instance if the cap is 2000 points, and a 3000 MMR killer beats a 2000 MMR survivor, they don't actually increase too much since the killer is significantly higher rated. Whereas if it were using the capped values then both players would be treated as 2000 MMR and the 3000 point killer would go up as if they were equally rated opponents.
So while the cap is used to help get very high rated players into matches more quickly, I'm assuming it's not used for adjusting the MMRs up and down.
0 -
Probably 69%
0 -
Just FYI Mandy mentioned in another thread earlier that there are currently no plans so far to show MMRs or MMR brackets. I wouldn't be surprised though if they replace the Grades with the character icons, that seemed to be a popular suggestion and I think it got some positive response from the devs at one point.
1 -
Awesome! Forum bingo is the best. 😄 Thanks for the replies!
1 -
Question: if you are, for example, 1000 points above the soft cap, and right at the same time you find 4 players that are also 1000 points above the soft cap, does that mean that you're much likely to match with them?
1 -
The matchmaker would treat all of them as if they were at the soft cap. Though for what it's worth, climbing that far above the cap is next to impossible for one person to do, let alone four, and the odds off all of those people running into each other are astronomically small. Realistically most people who manage to get above the cap stay relatively close to it.
4 -
Why don't you just show us our personal MMR so I can actually believe anything the devs say regarding MMR
0 -
Thanks for the answer Peanits ❤❤❤ 😃😃🤩, I've been wondering this question for a long time!
0 -
In all seriousness, the reason for this is probably because, to the forum members, their matches FEEL incredibly hard. That doesn't mean they're facing high MMR survivors. That doesn't mean they're high MMR. If a matchmaking system is working correctly, your matches will probably feel hard because you're facing people of your own skill level. When you get better, your opponents also get better.
Most people would probably be surprised by their MMR and, if their MMR was reset to say the lowest or the highest, would be surprised just how much variance there is.
This is probably one of the weak points in the MMR system and not showing the ranking. I think the downsides of showing the ranking are greater than the upsides, but it does suck that you can't see yourself progress so your matches just still feel hard without realizing you got better. I've seen this a few times though where I played a few different killers, then started using one killer more than the others. When I go back to the killers I didn't use very frequently, the survivors that I found difficult are a lot more predictable.
0 -
You could end every "I'm high MMR trust me bro" comment by simply showing us our MMR. No one is afraid to know the truth.
3 -
Is this all implemented into the game now or is it some part of a future update.
On the road so don't have time to read everything.
0 -
Is there a possibility of showing players how much their MMR increases/decreases after each match? I think if players see the amount (like -1, +10, etc) maybe it will make MMR not seem as bad. Right now it's assumed all losses hurt your MMR because it's unknown but I think if people were destroyed then see they only dropped a couple points (if any) in MMR maybe the losses wouldn't feel so bad/frustrating. Kind of like with old rank system if you safety pipped when you lost it didn't feel as bad as when you got a negative pip.
0 -
It's lonely at the top bros.
1 -
Stupid question from someone who hasn't studied how a MMR system works:
1 What's the point of a soft cap?
2 Why not making it so that if right at your time, when you're above the soft cap by X and queue, if you find also 4 other players above the soft cap by X you will almost surely face them?
3 Why isn't it called a hard cap then since it's almost like it?
0 -
It's been discussed a lot, but no, it's not something that's planned at the moment. Showing someone's rating or their rating change would push the game in a more competitive direction. While there's nothing wrong with playing competitively, it's not for everyone. Getting hit with a big fat -10 after your match isn't a great feeling, and it would drive casual players to play more competitively to avoid it. Ultimately, that's counterproductive: They just want to play casually, but playing more competitively is only going to make their matches harder.
In pretty much any other game, you don't see your rating in a quickplay mode, and that's because it puts unnecessary focus on that number and stress on maintaining it when it's really just there to give you a roughly even match. In a hypothetical scenario where the queues were split between casual and competitive modes, this would be a much better fit.
That said, if we were comparing a big "You lose" screen to "You lose, but only a little bit", then yes, that would absolutely be better.
2 -
I'm opposed to showing MMR values overall but I think you have a point. When I did gladiatorial PvP for WoW and we (duos) got matched against teams with a much higher MMR we normally got turned into paste but we saw our MMR not drop at all.
On the other hand, when we did win, I felt kind of sorry for our opponents because their MMR dropped by 60 to 70 points or some other ridiculous number while ours jumped by the same.
The only downside is that it's pretty disheartening to see a large drop in MMR and everyone will eventually have an off day and pull defeat from the jaws of a sure victory. That's the only concern I have.
0 -
I'll try my best to make it short:
- To make sure that the best players out there get matches. Originally we didn't have one, but the absolute highest rated people could end up waiting literal hours for a match.
- This one, I'd have to ask someone who's more familiar with the system, but the odds of this happening are so low that even if this did happen, it would really only make a difference in such a miniscule number of cases. I can't stress enough how rare that would be.
- A hard cap would be setting an absolute limit on your own rating. While it sounds similar on paper, it ends up being very different in practice since the rating difference determines how much your rating changes. Say you're 500 points above the cap and beat someone around the cap: Their rating wouldn't go down by much since the game expects them to lose. But if you hard capped, you'd have the same rating as them, so their rating would drop way more. (TL;DR: Hard cap would assume they are the same skill, soft cap lets them continue to climb but still find matches)
4 -
Thank you for the answer Peanits 😃❤
0 -
It's crazy that we're pressed into a system that we are given absolutely no information on.
1 -
In another post you mentioned needing incentive for rank v unranked - if seeing +/- at end would encourage competitive play and with many wanting to see their MMR and +/- wouldn't that be a good incentive for a ranked mode? It wouldn't be something that would force people to play ranked like receiving a reward would do but would be something that'd set those that want to play competitively apart from those playing casually.
0 -
I didn't go into it here for the sake of not derailing the thread too much, but there's a lot more to it than just showing ratings and calling it a day. We would have to be confident that the game could properly support two queues. This goes beyond just player count since the distribution of players based on preferred mode, rating, region, time, etc. all needs to be factored in. I don't doubt that we could create matches for two queues, but the thing you have to keep in mind is that splitting players between two queues is going to decrease match quality. If you sign up for competitive and there's not a lot of people playing around you, it's going to have to pull people in from further away, which is going to have the opposite effect of making the match less competitive.
We've discussed this on multiple occasions, but there's a whole lot more to it than there seems. At least for now, it's not in the plans.
3
