For one thing, if the system is working as you say, this shouldn’t matter because the game is finding evenly matched teams so the variance shouldn’t be enough that it matters whether the skill of your opponents is a little higher or lower.
And second, it’s obviously not working that simply because a lower MMR player can do better in a game than a higher MMR teammate and still gain less CSR in a win and/or lose more CSR in a loss.
For the average player, yes, it is that much harder to understand… And it will be until a developer finds a way to easily communicate this information in game in an intuitive way.
Obviously subjective… I don’t think it’s a adjective that fits in this case.
And I still think the problem is the 1-1800. We’re putting undue focus on very small, largely insignificant changes.
People are literally getting upset over very little.
And that’s 343’s fault. For creating the problem and then letting it fester.
I don’t have any problem with that at all.
In fact I think it would be great to show your CSR as a value on the line 0-1800+ and have your actual MMR curve in the background.
That would be educational and informative.
You could simply reduce the CSR to being the left hand side of the the MMR curve. Using the same -3 standard deviations we use now.
You will probably need to hide the MMR for a few more games than 10 though. Wait until it’s settled to a stable width before showing it.
And you could take it right to the extreme. Show your MMR curve and then superimpose your team-mate’s into a team curve. Then show your opponent’s team curve. Show how these are used to produce the winning odds. And finally show a little animation of your curve and how it changes with the result.
It could be a thing of beauty.
Yep. Weighted to wins. And medals and/or objective scores.
You are rewarded for that. Alongside k/min there is also d/min. Although I am pretty sure it’s weighted a bit less.
Possibly. But the advice I get from better players is that it is better for your team to shoot more and not worry about your accuracy. Every little bit of damage is a potential help. Which frustrates me as I take pride in my 60%+ accuracy.
I think this is a chicken and the egg thing.
You start to adopt that style of play when you start to play better teams. When I play with my Gold friends I get to go all Master Chief on the enemy. But when I play with my sons (Onyx level) I have to go super defensive.
I think we’re forgetting something very important. Something that was basically seen as mandatory in a presentation about ranking/ELO systems in games at GDC conference. It has to be fun! The ranking system has to be fun. It’s great if you have all the data bells and whistles but if it’s not fun people won’t play, that simple.
The ranking in this game isn’t about playing other teams getting a win and moving up based solely on the quality of your opponent. Now there’s a whole database of “this is what a skilled player plays like” metrics that you need to hit or you’re getting nothing for a win. So now from the players view it’s total confusion and the player gets frustrated because it’s not about the game anymore it’s about how fast they can rack up kills. Power weapons? Nope they’re mine. Objective play? Nope, yes I’ll lose but I’ll get the CSR back next game. If I play the objective I’ll lose MMR which is more important.
It forces one type of play style over all others. This is a classic example of some data specialist who only cares about placing people on a bell curve as accurately as possible without realising it’s a video game and people play ranked for fun. The excitement of getting on a good streak and moving up a few ranks and then losing it keeps you playing. Not this BS smoke and mirrors data metrics
Ranking and match-making don’t have to be fun. They have to be quick and accurate. That way they can provide the best quality matches.
There is no point lying to a player about how good they are.
Now, by all means build some sort of XP or military rank grind around it to make it as fun as possible.
But don’t lie to me about how good I am (or aren’t).
They looked at all the metrics. Assists, Damage, Medals, etc. Some of them made the system less accurate at predicting outcomes. I know are a lot of variables that go into a team’s performance. More than just kills. But when it comes to predicting rank the only ones that showed any predictive value were kpm and dpm.
This may not fit your (or my) intuition. But they have the data. Based on millions of games.
I agree it’s confusing and poorly communicated by 343.
And farming lower ranked teams for huge K/D is not going to help you rank up.
At the most you will gain slightly less MMR.
That’s different.
It shouldn’t.
What’s different now? You get a streak and gain some CSR. You lose and -poof-, it’s gone.
I just wish for an unranked arena playlist. With SBMM, I think it would be fun and less stressful and I would not need to worry about being a good player like when playing in ranked.
Or like what you said, there being no need to show the CSR change… I really hate it. #feedback
Your discussion of CSR gains/losses as a percent of total CSR is probably not a good way to look at it in my opinion. Players don’t really exist on the full spectrum of CSR, their day one CSR versus their optimum career CSR is generally going to be a pretty narrow band. I would venture to say it’s relatively rare for players to change CSR by more than a full class (e.g., Plat 1 → Diamond 1) over the course of their career in the game.
For example, I will live most of my career in between Diamond 1 and 6. If I spend some time away from the game, I may drop to Plat 6. If I work hard, I will occasionally spend time in low Onyx. But that band is my meaningful CSR range, so a change of 15 points is a much larger percent of that band versus everything down to Bronze, which will never be relevant for me. Make sense?
As an aside there is a weighting to your MMR if you spend time away from the game. A kind of form decay.
It does make sense.
I guess I’m using the “whole” band to emphasise my point - that the scale of the CSR is arbitrary and that by using 1-1800+ they have created a “false economy” over the value of each point.
And yes, as you describe people will oscillate around their band (or actual skill performance). Most will stay in an even tighter band than you describe. And on a day to day basis they will oscillate in an even tighter band.
So why are people getting so upset about it?
A few points for each win. A step back with the loss. As far as your MMR is concerned nothing has changed. Yet we have players fuming with rage over it. Squads trying to manipulate the game for a few more points. And God knows how toxic people are going to be if/when Champion status comes back.
I don’t know. Maybe 343 think this is a good thing?
Out of interest. You describe drifting between D1 and D6. Would you feel any different if you kept doing so, but without an MMR score each match? Or if you had a number on a different scale; like drifting between 33 and 40 on the old 1-50 spread.
Search my gamertag, check season summer 2018 and after (Fall season 2021 with win rate of 27% i gain a rank) of H5.
My MMR for no reason was adamant about the the fact i was a diamond (I asked Menke my MMR). Next season with a lower win rate I was considered champion.
When i said even with win rate of 80% your MMR can go down i didn’t explain.
From season summer 2018 to summer 2019 i always played with the same people,
A fellow champion player with a little k-min difference was around 100 point above me (1900), he will gain aroun 5 point against diamond team.
I was a at 1800 and gain little less.
A onyx player with a good difference of k-min sit at 1600. Most of time he will gain nothing or losing nothing
A onyx player with huge difference of k-min drop MMR at every game (every week he started at onyx/diamond six and end up at platinum six losing 300 hundred poitns). Gain zero even against top player (Champs 1 to 100) and losing max point (30) even against top palyer (1 to 100). Fun fact every weekend he played alone and farming poor platinum/diamond/ony players he will gain back his rank.
You will notice k-min will net you almost 800 points.
I kown all this becaus for almost a year i argued with Menke about this stuff.
MMR work only with huge population and if you play alone.
Also, drift isn’t the word I would use personally. For me, drift is the “insignificant” rank movement that just represents measurement error. Movements between D1 and D6 are fairly significant and I would expect them to correlate with time periods where I am practicing more, playing better, playing with my team only, etc.
I meant without using the same scale as the MMR, 0 to 1800+. Sometimes I call the CSR number the MMR number because that it where it tries to be. My bad.
Use a smaller one instead.
I guess I was using it more for the 10 to 15 points up and down.
D1 to D6 is definitely more a slide.
I think Menke once mentioned that players can vary day to day by plus or minus 150 pts. Just in fluctuating form.
Yeah, day to day in any skilled activity you’re going to see variance in performance. Skill is a noisy, inconsistent thing to try to measure, as we all know.
A smaller scale would have pros and cons. It’s probably more accurate in that it only shows you movement when it’s more confident you have actually made a meaningful move. But, players want instant feedback, even if it makes them mad. Pros and cons.