"tHE MMr sOFt cAp is SO eASy tO hiT"
Bullet points:
Only 5% of the player base is above 1600 MMR.
MMR splits the playerbase between sub-1600 and above 1600 in most cases, unless extreme wait times.
Even above 1600, MMR system continues to try matching similar MMR killers to similar MMR groups of survivors.
MMR gain is slowed at 1000, then slowed to a crawl at 1900.
This goes to anyone in this forum who continues giving me an aneurysm when they use the "but the soft cap is so low, MMR doesn't work!" line in an argument.
Comments
-
seems like deranking your mmr doesn't work well after reaching the soft cap.
It means the only solution is to open an alt account and lose like the first 10 games with each role to get an inital low mmr.
0 -
Yeah but does it ONLY matchmake people that are over the soft cap? From my knowledge it should matchmake people with AROUND 1600 so getting those lobbies isn't exactly hard. Either way mmr in a largely perk and RNG decided game just really doesn't make a good skill ceiling.
4 -
Thing is that argument is still very much valid, considering a good chunk of players on this forum have hours that range into the thousands. While hours doesnt necessarily mean youre good at the game, it does usually equal to a better player.
Another thing is there arent many people who frequent the forums, and as said before a good chunk of them have thousands of hours. Most people who only have about 300-500 hours dont care enough to go on forums and discuss about balance, they just want to play the game. Those people are also the VERY large majority of the playerbase.
And if that doesnt convince you, then just watch some streamers. Majority of their games (ESPECIALLY during off hours) are inconsistent. In the same game they can have some one who has a few hundred hours, and some one with a couple thousand.
So all that to say, yeah MMR still doesnt work very well at the soft cap, and thats fine for the most part considering the main purpose of MMR is to seperate brand new players from thousand hour monsters
6 -
Or you don't game your MMR in order to manipulate the matchmaking system to put you against potatoes who don't stand a chance against you?
7 -
"top 5% MMR" =/= "5% of players are top MMR" tho.
The top 5% MMR could very easily be 2500-3000. Or indeed 1600-3000. No one knows. That's kinda the fun thing with statistics BHVR releases; they are pretty useless outside of the one insular point BHVR wants to illustrate.
4 -
Hens disagrees with you. Are you claiming he doesn't understand it or his point that if you're at 1400 MMR you can be matched with people over the softcap if their wait is a minute or more?
3 -
Doesn't mean that people UNDER the cap don't get matched with those people on a regular basis.
6 -
Safe to say its probably somewhere around 100k current players total if steam alone averages 30-35k per day (which steam is not the most dbd prevalent platform) with peaks of 45k. So the "top 5%" would probably be more around 7-8k players all though that's spread around regions.
0 -
I'll face 20 hour Dwights one game and a 3500 hr 4swf the next with no wait time, tell me again how MMR is working?
9 -
Top 5% MMR means the 5% of players that are at the top of the MMR scale, not that only 5% of the players are above 1600 MMR.
5 -
Even then you would still not know what MMR these x thousand players have. (You also use "current players" - and we don't know if BHVR just counts accounts or only active accounts - and what an active account is.)
Everything MMR is just 99% speculation. But I do agree with one point OP makes; the kiddie pool aside it's meant to split the player base roughly into two; those who've reached 1600 and those who haven't. - With those sitting right around the cap probably having the most diverse set of matches as they can get matched with pretty much anyone outside the kiddie pool once the BP incentive sits at 100% - which it usually is. (What was it sometime last year. The max adjustment that is reached pretty quickly to make matchmaking fast and not have minutes long q times ... +/-500 MMR? )
0 -
This is unfortunately the point the OP has completely missed. The softcap was previously 1900 and it was moved to 1600 for a variety of concerns. The population distribution at each MMR didn't change; only the range of MMRs did and since a 2300 MMR player can meet a 1400 MMR player if enough queue time is involved the soft cap is low enough that high MMR is a myth.
3 -
Video is still good, it is just this one short sentence (like not even 2 seconds in the whole video) where they were wrong. OP is just cherrypicking stuff from it and ignoring everything else.
3 -
Yeah, the video is fine. It's just the premise of the OP, that MMR effectively works properly and pairs only players of a similar skill level to other skill levels, that is faulty. That's the premise I'm criticizing. 1600 doesn't involve only 5% of the player base; it involves a large enough amount of the playerbase, especially considering it can go down to 1400 MMR, that 'high MMR' doesn't effectively exist. That's why Dowsey now gets a match in a few minutes instead of waiting hours.
2 -
Yeah, the queue time is the real red flag here when it comes to the accuracy of MMR; if it was as rigidly accurate as the "TOP MMR/TOP LEVEL/TOP PERCENTAGE" crowd wants to be believe, their queue times would be nearly eternal. I get it, they put thousands of hours in and want to believe their dedication has set them apart, but it just isn't so.
You just can't have it be that narrow and keep the queue times from alienating people.
There is no subsection of the player base who aren't subject to the birdshot level inaccuracy of the MMR system. Watch just about any streamer and you can see it, if you're willing to. And sadly, it kind of has to be that way to keep the game playable.
3 -
Exactly, and that's why Patrick, during one of the dev streams, also talked about how, if matches were to be completely fair, people would be waiting for hours for matches so the system prioritizes speed over even skill levels. Dowsey, as my example again, is definitely in the top MMR but, after the skill cap was reduced, switched from hour long waits to a few minutes. Lowering the MMR cap did not magically create a bunch of more skilled players; it just increased the range of players he can match with.
Otherwise, who's going to wait an hour for a match? Nobody.
2 -
And this is why being at or near the soft cap sucks so much. Once you get close it's mostly up to matchmaking which side wins. There comes a point where you realize you didn't really win because of your skill, you won because the matchmaker flipped a coin and got heads this time. (Not saying skill isn't relevant, but matchmaking creates so many games that are just totally hopeless for either side, and that's depressing.)
1 -
I like to pick on Tru3 because he's such a donkey, but if he (and others) were in the tiny pool they claim to be, they might get 2-4 matches in a 6-8 hour stream.
I recall when BHVR was testing their different SBMM configs, and you could tell when they made it too accurate, because if you weren't dead in the middle of the MMR curve, the queue times got painful.
1 -
The time people play will be a factor I’m sure.
Tru3 plays on UK/EU servers and streams during the day when most people are at school or work, and gets very sweaty games until late afternoon when people are getting home and putting the game on, and there is a noticeable difference in skill of players he is matched with.
0 -
"the 1600 and above bracket consists of 5% of the playerbase"
Seems pretty self-explanatory to me.
0 -
But he still gets potatoes sprinkled in. Not a ton, but they are there.
Which is the point. Him going on about how they're all demigods doesn't change it.
0 -
....that is something the dude who made the video got wrong.
"finally according to data given to us by Behavior it seems that the 1600 MMR bracket consists of something close to five percent of the player base"
This is what he is saying while showing the slide "Killrate for every killer (Top 5% MMR)"
Same as before "top 5% MMR" =/= "5% of players have MMR above 1600"
"top 5% MMR" means the players with the highest MMR rating. What MMR these players actually have, that we don't know.
Heck, we technically don't even know what the phrase "top 5% MMR" means. It's not the MMR score itself, that much we can safely assume I'd say, given that "top 5% of 3000MMR" would be the bracket 2850-3000. Much more likely it's "accounts with the highest MMR rating that in total account for 5% of the player base". Going backwards from 3000 we don't know how far down we need to go until we get to the 5% mark.
4 -
It amazes me how many comments I see across social media, where players hit red rank and assume it means they're in high MMR
2 -
Back when I used to watch Tru3, the time of day he plays was something his viewers would point out to him constantly. I remember he had a period where he did change the time of day he played and his games were so much more chill. But then he went back to his regular time and things got sweaty again. I think he wasn't pulling in as many viewers at that time.
2 -
Here's a further illustration for the OP showing why the statement 1600 MMR is the top 5% is not accurate. The link I'm leaving below shows season 5 of WoW arenas which did use the elo system, just as DbD, chess and the other MMR systems use.
As you'll see, titles were awarded by what percentage of MMR the person was in according to their region. Why region? Because the point at which you hit that depends on the competition and scale of players. MMR rating is specific to the groups and subgroups in whichever sysyem is measuring it. As such, nobody can state a certain MMR means x percentage of people in the game overall as what percentage is x MMR depends on the participants in it unless BHVR uses a global system for all servers. Does BHVR do this? Who knows? Somebody in BHVR who's not talking knows and that's it.
We can look at population distribution and note that, in the season I showed from WoW, 1600 meant approximately 1/3 of the arena players. In chess organizations, we can see a 1600 rating means 20% of the players in one group, 25% in another, and maybe 1% in a highly select group. So, can we draw a conclusion from that to DbD? Not a chance; we don't have enough data. We could infer that, since 1600 is greater than 5% of the population in every other elo system with a large playerbase, it's unlikely to be 5% of the playerbase in DbD and that seems reasonable but I have no idea what the 5% cutoff is and neither does anyone else outside of BHVR.
Since DbD can draw a 1400 MMR player against a 2200 MMR player if there aren't enough players to match against the size of the population is expanded even further. I would guess that about 40% to 50% of the population would be at 1400 MMR+ especially as Killer and that's why matches can be so inconsistent and you can get normal survivors one match and Seal Team Six the next or survivors will face a newer Killer one game and then a death machine that cuts through them like an anime cartoon the next.
That's just how it is. High MMR is a myth and that's why players like Otz, Dowsey, Hens, etc will get so many easy matches in a row. The alternative, matching to MMR level accurately, would leave players at I would guess the top 20% waiting 15 minutes to an hour for one match and they would quit playing. If BHVR releases the data for what the top 5% of MMR is we'll know what it is for the time range of the data but unless they do so nobody has any concrete idea.
0 -
Not to post this to get people to call them out, but this got me thinking;
When Otz and the rest did Hardcore Survivor last summer on new accounts, how likely is it they actually reached the soft cap of MMR?
0 -
Scott Jund explains this very well on this video.
https://www.youtube.com/watch?v=GHEkN-oDLrk
I do agree with him, top MMR is a meme, Ive been paired against certain Killer streamers who have around a 90-95% (checked their streams, they rake in win after win after win) win rate and I know Im not THAT good to be paired against them yet Ive encountered them several times, sometimes even after having a big losing streak, they are probably ~2200 or even higher and there is no way Im that high as a Solo queue with a 25% non Hatch/EGC escape rate (as those dont increase your MMR).
0 -
They played 101 games of survivor with only 26 deaths between the four of them, once you're at 1k MMR you can only get up to 20 MMR per game if you have at least 10 minute long games
In the most conservative case of every game lasting only 5 minutes and there always being a hatch escape and deaths always being the max MMR loss that would mean an escape distribution of:
- 26 deaths (-20 MMR/death * 26 deaths = -720 MMR)
- 101 hatch escapes
- 277 gate escapes (10 MMR/escape * 277 escapes = 2770 MMR)
Assuming that no one person died more than the others the MMR would be:
1000MMR (Base) + 2770/4 (Gates) - 720/4 (Deaths) = 1512.5 MMR
In the absolute worst case scenario they get within ~10 games of the softcap, more realistically the MMR gain per match is probably closer to ~15-20 MMR since they extended matches to try to get more emblems and pretty much all of them would be above the softcap in that case
2 -
Is very easy to know if yourself is on high mmr, dont overthink nothing
0 -
Thanks a lot for this breakdown friend! Really interesting.
0
