I decided to quit until the reset. Hopefully there is an mmr reset. Played too much BTB. Since that mode is full of low ranked players losing drops your mmr by a lot I’m pretty sure. It’s not easy to carry a team of that size.
1 Like
You can have a 40% win record and rank up.
Do you think those 15 games were all against even opponents? If they were you’d have won 7-8/15.
Are you playing with friends? Lots of high level Onyx stacks run into this problem.
Hard to say if the system considered the matches even or not (in terms of MMR) because there is no CSR breakdown after the games.
Yes playing with team all in mid-high Onyx. I would consider my 1650 pretty low but routinely beat teams with 1900+ and do not gain significantly.
I honestly think if I’m beating players with higher CSR, I should gain faster. The reliance on MMR seems to make the whole thing seem wacky.
I’ve seen that. There are Onyx players with a 40% win record and a 1.0 kd with 40% accuracy. Which leaves me to believe that I can’t rank up because my mmr is too low.
People tell me they lost all placement matches with a sub 1.0 kd and got placed in Diamond 1. I was placed in Platinum 1 with a 70% win record (3 quitters in each loss) and a 1.23 kd. I even got 2 overkills.
After 20 games in controller solo Graslu, the highest ranked mouse player, is a Diamond 1 with a 35% win rate and a 0.75 kd. Meanwhile I’m a Platinum 5 with a 0.94 kd and 71% win record.
I think people are over-estimating the relationship.
It makes sense to share a lot of the data between playlists - algorithm size and efficiency. For instance there is a weighting for time since last played (you would expect your rank to sag) and this would act universally across all playlists.
Remember that Halo 5 had the same set up; shared data and offsets for each playlist. And in this game people could comfortably be different ranks in different playlists; Platinum in Slayer, Diamond in Snipers, Gold in FFA, etc.
Yeah it’s all because of MMR and your opponents.
I’m rated Diamond 1 in keyboard and mouse with a 0.57 k/d and 10% win ratio. Only won because enemy team had a quitter. It’s impossible for me to play as it expects me to play like a 1700 Onyx player against some the best mouse users in the world and I struggle to pick up a weapon with keyboard.
What are you basing that on?
I can tell you from experience the game has decided that I’ve reached my “true rank” and no matter how many higher ranked players I defeat, my rank climbs at a snail’s pace.
There is a clear relationship between MMR and the rate of your progression, despite your ranked performance.
If you win the clutch games (ie. vs teams ranked above you) you will rank up.
The KD is not as a heavy weighting as a lot of people believe.
And accuracy is no indication of skill. Mine is over 60% and I am no where near Onyx. 
And keep in mind you do need a little bit of luck from match-making. Placement is only ten games - but sometimes you get lucky with maps, modes, and team-mates, at just the right time when you are served up the games you are supposed to lose.
I don’t think that’s a thing.
Consistently beat teams ranked above you and you will rank up. If the system has become confident in your ranking your MMR curve will be narrow and harder to shift (compared to earlier / placement) but you can still move up if you are improving.
Placement games do seem to be less focussed on the win. The weightings for things such as kill rates (as opposed to kd) are probably more active. And I’ve already mentioned that you need a bit of luck in terms of what games you get and who your team-mates are.
It sucks that you were dragged down by quitters.
But in the end it depends on who you beat. The three games you lost were probably the ones you needed to win for the system to give you the opportunity to rank up faster.
Congrats on the overkills… but they aren’t going to help you (unless they were pivotal in beating a side you weren’t expected to).
But also keep in mind that your curve maintains it’s volatility for a while after placement - so you still had plenty of time to justify ranking up. Your sigma value only starts to drop when you are consistently playing at the same level.
Keep in mind that as you play consistently at a level the system narrows your skill curve and makes your rank harder to shift (up or down). But no where near as hard as the lock down we had in Halo 3 
This is perfectly natural. By this stage it is assumed that actually improving your rank will take time and effort.
But if you can show the system you are beating those higher teams regularly then you can, and should, rank up.
Pretty much all ELO systems have controls on progression rates.
Chess uses a variable that drops when you get to a certain level. In essence it helps protect Grand Masters from losing large chunks of rating if they lose an upset here or there.
TrueSkill uses the width of your skill curve. As you play consistently your sigma value (standard deviation) drops faster - to tend to lock you in (the system is now confident in your rank). If you are still having “upsets” (for or against) the system lacks confidence and maintains your sigma value (volatility).
And I hadn’t thought of it before - but the rate of decay of your sigma value is probably just a metric set by 343. So if people are feeling that their ranks are slowing down a little too harshly - it’s probably a consideration for tomorrow’s reset.
I managed to rank up a bit in the last couple of weeks after pretty much being the same rank for over a month - so at a personal level I don’t feel the ranks are locking too hard.
Pretty simple. If you beat higher ranked players, you should rank up lol.
All this mumbo jumbo about skill curves and TrueSkill MMR bla bla bla 
At the end of the day, it’s not fun to stop ranking up even as you win the vast majority of the games.
So whatever the solution is or isn’t, something needs to be done to address this because it’s simply not fun.
You do.
But if you are the lower ranked team at point A and you beat a higher ranked team at point B… you have to have some sort of control over how much A goes up and B goes down.
You can’t make A = B on the basis of just one game.
If the system is relatively confident in your rank… the change should be cautious.
I actually tried to read a mathematical paper on TrueSkill last night.
It was indeed mumbo jumbo.
No, it’s not fun. At least not in the sense of immediate gratification.
And once you hit your rank ceiling any improvement will be slow and marginal - so it could never provide this sort of fun.
343 need an alternative. The often asked for XP rank to grind!
1 Like
“You do”
Haha, but you don’t! I agree with Aristo on this one. I established with Ken that out of a 25 game sample I won 52% of games, including the slim majority of 50/50s (important wins, even in chess which I also play you expect Elo gains beating an even-ranked opponent unless you’re like Magnus and are at the way far end of the distribution) and lost only one game where we had a high expected win percentage (ie “bad loss”) and I lost a not-so-cautious 100+ CSR. Ken explained it based on my individual performance which wasn’t even that bad (.98KD) but if so, that just suggests that the W/Ls really don’t matter.
You both seem to be saying different things? You’re both clearly smart guys but it seems to me you’re on different pages about the important of wins versus individual performance.
I see you typing there, Ken. What, I can’t just complain about my deranking in peace, eh???
2 Likes
But you’re not playing players equal to your CSR, you’re playing players equal to your MMR which is clearly 100+ points lower than your CSR was at this point for whatever reason. If you were able to play at a level 100 points higher you would likely win every game or almost every game by a fair margin. Because your team would be inappropriately rated Vs the enemy team. This is how players spike up their MMR.
I’m pretty sure the expected win rate for a to4 is 57% so 52% is also an underperformance given the circumstances.
Like I said, you’re playing lower rated players because the system believes you are lower rated than you current CSR. If that wasn’t true you’d be stomping in every game. But it appears to be true as your k/d was under 1 and you won half your games. That’s not telling the system your MMR is inflated. It’s telling it your MMR is probably really close to accurate. Your winning 50/50 games and losing 50/50 games against players that are rated this lower MMR, not the CSR you once had. If they were rated higher MMR you would be climbing as you’re performing adequately against those players.
Complain all you want fella, I hope to bring you some clarity on why it works like it does. You don’t have to enjoy or like it and you can keep complaining until they change it to Trueskill3. Hoping it makes it less frustrating, but the more you know can definitely make it more frustrating so sorry if that is the case.
1 Like
You’ve jumped from 52 to 57 on TO4 percentage… I’d be curious about a citation, is that in TS2? Even if so, no idea how those hold four years later.
But anyway, that begs the question–I was stable at 1530-1550 for say, three weeks. How exactly, if my MMR is actually 1480, did I jump to 1580? I should have been gaining 1 or 0 per win, but I gained enough consistently to jump 30. Then, soon after, I sunk like a stone despite winning roughly everything I was supposed to.
As I said before, we’re at an impasse because you’re claiming something unprovable. I can justify any CSR change in any situation by just saying “everyone’s MMR must have been like X”. You went +15 in a clutch win versus people way above you and went down? Just throw the “MMR must have been way off CSR” in and call it a day. Get what I mean?
And man, if MMR drift is 100+ points away from CSR, then the system just sucks. That one there’s just no way around. That’s a system that’s just not performing well.
1 Like
Yeah but your MMR is adrift and you couldn’t prove it should be higher.
Other than by winning all the games I was supposed to according to Halo tracker, except one? We’re back you vs. Darwi–Darwi says wins will take care of themselves, you say my individual performance was enough to tank the CSR regardless of winning the games you’re supposed to win according to predicted W/L probabilities.
Maybe rewind. How do you think I arrived at situation where I was Onyx 1582 but my MMR was about 100 lower? What do you think caused that level of drift? I certainly didn’t win 50 straight and the game grudgingly gave me one Onyx point each time.
You won games that had a 50/50 match but your MMR at this point is massively askew from your CSR for whatever reason.
You are 1600 you beat 1600. Your CSR is 1700. This proves you might 1605, not 1700.
Who knows man? Rank decay from not playing, bad social games, worse performances than expected continuously, other players compared to you have gone higher up and proven more which pushes you down. I dunno
You know, I had about enough of this Ken. You think you’re so smart. Let’s settle this the old fashioned way.
1v1 Derelict, rockets only, no radar, split screen. Bring your best second. Oh, and no anniversary BS, we’re playing OG on disc.
1 Like