Ranking system & MMR

Games Halo Infinite
21

A user’s rank should never be the validation of their argument. It can solidify that they understand the game in terms of gameplay, but not the ranking system.

It is a ranking system and not a progression system. There’s basic rules of thumb to indicate why you move up and why you move down. The variables of the real skill level of certain players is what causes a lot of fluctuations in player progression, and the loopholes that exist within the skill ranks to skew MM into a team’s favor to have easier matches to rank up make for some very frustrating experiences of getting 1 point for a win or -10+ for a single loss to the same team.

Beat teams that have a higher win % chance than your team to move up. Personal performance carries some weighting on your increase in rank. Lose to that same team and you aren’t penalized that hard unless you do really bad.

Win against a team with a lower win % chance than your team and hardly go up. Lose to the same team and you get a severe penalty at times with a small safeguard SOMETIMES if you perform exceptionally well despite the loss and have maintained that performance across many sets of games.

The variable

  • a high skill player disguised as a bronze-platinum player (account deliberately created to boost the increase of team mate rank, create soft losses, and skew MM down for easier matches)
  • a high skill player In daimond that hasn’t had enough play time to rank up to where they belong
  • unranked player (true skill is unknown amd current rating can cause for a big variable)

These accounts will have the most affects on your increase and decrease in rank.

22

My statement on his rank does indicate his knowledge and experience. He clearly thinks the system is working fine because he doesn’t experience people boosting because nobody is trying to boost in platinum.

They are not seeing the problem of boosters and people manipulating the systems “prediction” of who should win. I’ve seen it get so bad to the point where 2k players are with 3 Smurf accounts to drop the CSR so much. and the 2k player is facing 1500-1600 players and when they lose they barely losing anything. It’s rigging the system and it needs to be addressed.

23

We need to stop your social MMR from interfering with ranked MMR. It’s kinda dumb. Because in casual, I can slay all day long just chilling.

I go to rank I am facing off against lil Roybergs and lunchbox’s from 2007-2010. While my teammates pick their boogers.

They need to be seperate. Once is an actual determination of your skill vs others, others is a piss poor excuse to balance matches in social. No reason both should be used, because it makes matches jank.

1 Like
24

You’re pretty much saying what I described right? I’ve suggested my thoughts so many times already but it seems like 343 doesn’t even look at these forums. There’s a few problems…and true what people say…they’ve always existed…does not mean they cannot be remedied.

  • skill discrepancy between team mates too high. Your csr gain will be (insert reasonable % here) of normal gains, losses will remain unchanged.

This remedies the boosting affect of having a lower player on your team and normalize the loss, discouraging deliberately abusing the MM manipulating effect.

  • new accounts
    Right now you can only place into daimond, and the unranked affects players TOO much. It’s difficult to deal with this 1. Here we need to brain storm an idea that makes it feasible in our matches where both teams don’t get affected so badly by the unranked player

Under normal circumstances, the basic rule of thumb works correctly. The longer a game has been out and the more players play the more the ranking make sense in our matches. Its these odd accounts that sabotage our matches deliberately at times. The ability to just pick up a new account has never been remedied. Idk if this will ever change…

2 Likes
25

Rocket league has one of the best ranking systems IMO. It’s not performance based, its win based and when you que and you have variants in rank the lower ranked player is equalized to the highest rank in the group to prevent manipulation.

There will always be Smurfs in some aspect, but they can at least fix the manipulation side of things when it comes to that.

26

It can’t be purely win based though. The other factors have to matter. I mean, it’s still basically win based right now, but the players you play against have to mean something to move you up or down. Maybe you do agree with that and just didn’t say it in that reply. I haven’t read through every single reply here. Sorry if I assumed.

We don’t want absolute discouragement from playing with friends that are lower than we are, but there is a stark contrast to real friends playing together and a booster account. Absolutely NOBODY would ever bring a real bronze Into a high tier lobby unless their reason was to get that player to quit halo for good and have no hopes to ever get better.

1 Like
27

I didn’t mention it but yes, you should absolutely gain less if you are favored, and more if you are not favored. The problem im seeing though is even on games where im favored 56% to win I’m still gaining 1 CSR. and on similar games where the enemy is favored slightly and I lose its -10 CSR. These are not exact but I’ve dug through hundreds of games to compare on not only my games but others as well and it’s the same issue over and over.

28

Here is one example, my team had a 37% chance to win and I get 1 CSR.

tracker.gg/halo-infinite/match/7d0afcdf-1692-4e6c-81f3-c58b162b3242

29

The csr gain/loss between losing or winning in that differential is what you believe should be equivalent correct?

If you’re 6% more likely to win than the opposing team and only would gain 1csr, then by that same standard losing should consequently lower you by 1csr as well?

I can’t say I disagree there at all, and cam personally say under the same circumstances I’ve been lead to frustration. Here’s where personal weighting should be more experienced.

Should you lose by a landslide, I believe the consequence should be more severe. Should you win by a landslide, this is where difficulty lies because you beat a team you were expected to have a higher win chance against, and by winning showed they aren’t where they belong…should this show you aren’t where you belong by the same standard? Or that team losing so much csr becomes the difference of you witnessing you beat a team of 1600s that played at the 1500s level, and that the system offsets your gain by it suddenly reacting as if you beat them at a lower rating than they were during the match. (The consequences are interchangeable, vice versa)

In other words, the ranking they are after the match is what the system uses to give you your gain, and not the rank they were during your match…which could possibly by upwards of a 15csr difference multiplied across each player resulting in a different average mmr.

Its really hard to understand at times man. I can sit here and try and tell you I completely understand it…because at a glance…it makes sense…but under closer scrutiny and these types of matches…we’re up in arms trying to decide what we should really expect vs the reality we’re served.

It all goes back towards it being a ranking system and not a progression system where the systems designed to not give such sudden increases or decreases without warrant…buuuut…its not perfect and this is where the frustration you all face, myself included is felt.

Fortunately we love halo and continue to play, even if at times we find ourselves “spinning our wheels” feeling discouraged to continue playing when we go up 1 and down 10 feeling it should be closer in gain loss to the comparable teams.

1 Like
30

and heres one where we are favored 56 % to win and I gain 14

tracker.gg/halo-infinite/match/61e12554-2fd4-4fc0-9fa6-7964ba4e1bcc

The system is all over the place I honestly have no clue how to rank up even when I’m performing well every game.

my last 25 games I have a 76% Win rate I went 19-6

I started at 1518 ended at 1555. In what world does that make sense lol.

31

It won’t load for me not sure why. What rank were those players during the match and what rank are they today?

Compare this to yourself as well and maybe your team mates is possible. It’s understandable if that’d way too much digging.

32

Thanks, I’ve read that before. It is very interesting.

See, I care!

I had read that section as sharing base data between playlists - the general shape of your skill curve (mainly width and thus volatility of rank).

And then the offset for each playlist moved the curves on different parts of the x-axis… Allowing them to be functionally independent when they needed to be.

Anecdotally my ranks for Slayer (my core jam) is vastly different to when I’ve ventured into snipers or swat. In Slayer I’ll get put against Platinum and Diamonds (and way too many Onyx). In swat I’m playing Silver’s and Gold.

Functionally two very different MMRs.

The only bit I couldn’t rationalise is the placement in each playlist. This is where I envisioned that the system opened up your skill curve again (increased the standard deviation back to initial value). I guess they could have more than one offset… one for the mean for each curve and one for the width.

My bad for not elaborating…

ELO type systems are open ended by nature. And if you are drip fed long runs of wins… Or you truly are unbeatable… you keep going up and up.

There is a bit of natural braking in that the gap to your opponents gets larger and your gains diminish… but chances are the house has bolted by then.

It’s less of a problem in structured tournament type sports where you aren’t feeding your ELO score easy wins 50+ times a day.

You could put ceilings on the ELO. Some have regular culls. Some change the scoring system - like Halo 5 introducing set gains for Onyx and adding a weighting to losses. This was my train of thought when I said (this) MMR want open ended.

And I thought the gist of what Hanke was saying is that the MMR and CSR can drift apart… but as soon as you get a result in the other direction the system used this opportunity to reconcile.

Maybe, for the handful of pro players they can get their MMR into 5 figures… But that shouldn’t impinge on the 99.998% of us left behind.

I do wish 343 would set up a Hanke type liason.

33

so me and my teammates all started around the same rank and we’ve mainly only played with each other in open for the last 200 ish games. We’ve all had our bad games and underperformed etc. and everyone has had their dominating games over the course of these games.

On average I would say that we all performed fairly similar accross the board in terms of stats and averages. One of them is current 100 CSR higher. There have been games where I was MVP and he was 2nd or 3rd I would gain 1-3 CSR while he is gaining double digits. Multiple times, consistently.

That’s why I’m skeptic of performance being the biggest factor because for my experience, it just doesn’t add up. I went over my last 50 games and these are the stats.

Wins - 34
Losses - 16
Win Rate - 68%

-won game placement-
1st place: 18
2nd place: 7
3rd place: 8
4th place: 1

-loss game placement-
4th place: 1
5th place: 8
6th place: 4
7th place: 2
8th place: 1

Starting CSR: 1553
Ending CSR: 1555

If I’m winning 68% of my matches even if I’m bottom of the scoreboard EVERY GAME I should be ranking up. the fact that I’m 2 CSR higher shows there is something fundamentally wrong with the system.

1 Like
34

This is where its tough because th context of the matches matter. That’s a sizeable amount of wins where your placement is good, and even the losses. Someone would literally have to go through all the matches to look up the context of team make up, what players are on up/down swings, total matches time of each player etc.

I get the frustration man trust me. We want it to be simple, at a glance make sense. We can’t rank up indefinitely just by winning as mentioned before…but there are variables that perhaps can be seen as to extreme and result in rank yoyo, or even soft cap on ranking up.

Ideally the system exists to get us to our respectable ranking tier and have decently fair matches where matches are not too lopsided. It isn’t a system meant to literally grind in terms of say a platinum level skill player just believing with enough play time they can get into onyx. The match context matters. Their win % becomes scrutinized to depict if the rank movement has made sense or not made sense.

I do wish there was an easy means of going through all the matches man. None of us probably have the time to do that with the way the back and forth pages load on the tracker.

1 Like
35

The system is great. Not saying there can’t be improvements… or that this implementation may have some teething problems.

But I do like TrueSkill.

And while I’m only a lowly Platinum… It doesn’t stop me from being baffled that people think a 60% win rate should guarantee them increased rank.

Or that 343 thought it was a good idea to put your MMR out there in the first place. It was never going to work as a target to grind.

They would have been far better off dividing Onyx into percentiles (eg. Onyx 12%) and buffering drops to give you a sense of stability in your climb (so if you drop rank you have to stay there a while before your CSR actually tells everyone).

In a later post you go on to detail your frustrations with teams manipulating their MMR with Smurfs etc.

I agree that is more of a you problem than a me problem.

I don’t know how 343 fixes it. I hope they do. But that’s really a separate problem from the ranking system.

1 Like
36

Not disagreeing with you if this is the case, I just wonder what your thoughts are on the point of playing ranked beyond a certain point. Most people are looking at it going “I want to get to that next tier, or I want to make it to Onyx.” If the system is set up to make a determination of who you are and to keep you there – what is the point of playing?

37

Performance isn’t just kills and deaths, I’ve had a look at you and your friend who is rated higher profiles. While we don’t know for sure what factors into the performance aspect, some things that are known to have an affect on it are quitting games, playing more games, k/d, win/loss and it’s safe to say damage and accuracy are almost definitely considered. I’d say damage is potentially the most useful indicator of player impact in a game than anything else. Deaths is also a key one for me personally. Players who stay alive more are usually the better/more useful teammates provided they are still performing their key functions.

Your friend Joelly
Games 430
1.14 k/d
56.7% win loss
5,017 average damage
Matches quit - 2
47.2% accuracy

You
Games 467
1.14 k/d
58.2% win loss
4,781 average damage
Matches quit - 13
47.9% accuracy

Me
Games - 73
1.23 k/d
60.3% win loss
5,243 average damage
Matches quit - 0
51.9% accuracy

While there are no doubt other factors to consider, I think it’s safe to say that your friend does have stats that may be considered somewhat better than your own, and I have stats that outperform his. It is also not entirely clear as other things to consider such as k/d is more important in slayer and less important in objective so that is also something that can’t be considered here without further review of games. Your score is also factored in. Which makes sense given our respective rank progressions.

I think the main thing is to not overthink it. You shouldn’t be trying to play in a way the system will reward you, but instead aim to just elevate your game however that may be. If you really care more about CSR gains then just play solo as it’s easier to gain momentum and prove your performance if you can adapt around any teammates.

1 Like
38

Dw bro I appreciate your input and I’m in Onyx. You don’t have to be a high rank to understand how the system works at all.

I’m not the top rank in a bunch of games but that doesn’t stop me from researching and learning how the system works.

1 Like
39

We only get better by becoming better at the game. The aim for a player can be as simple as getting into onyx. Which like me, is what I’d like to do. I get weird random matches where I beat lobbies with 1600s in them for a nice csr payout, then get smashed by a team with a fake gold or a hard match against daimonds that played like onyx 1800s. It’s rough. Meanwhile a dude on My friend list is coasting in onyx on the upswing…i check his matches out and he hardly faces onyx players.

What gives? Why are my matches so hard sometimes and why do some of my friends avoid inviting me because it makes their matches more difficult? It’s frustrating to say the least.

How we beat the better teams and our improvements in our game play matter more than the rank itself. It would be really nice to get what we feel we deserve after a Match…but oddly enough at times we get more for a ridiculous win where we did less and the match felt easier than the win where you clutched a very intense match you almost lost.

The point in playing beyond where you reach your ceiling is for the sheer enjoyment of the game itself. It may sound dumb…but it is not. The rankings exist for healthy MM. We ourselves can treat it as a vanity emblem. An achievement. It does feel nice for a player to say

  • onyx 1800 here. Here’s what I gotta say about this topic

Makes that player seem more credit worthy. Though…sometimes that’s the wrong message we should carry within our community where everyone’s opinion should matter, and rankings really should only indicate a players understanding of the game (although your point of the other user not experiencing the MM manipulating is very valid and rings true)

343 has done a CRAPtacular job with player incentive. Absolutely no good static commendations…and I feel the frustration of playing beyond the point where you are when you feel locked to that point. Spin your wheels and derank after 10 wins and 1 loss. Who enjoys that? Even after reading everything to make sense of it will not take away the taste of that bitter loss erasing all those wins in the span of a few minutes of 1 single match. There goes the night.

1 Like
40

I appreciate everything you said, while still acknowledging the incredible frustrations of trying to figure out why one is stuck at a certain rank.

This sums up where I’m at. I’ve hit the Diamond 5 ceiling. Got incredibly close to Diamond 6, and then a few bad matches and I’m back down to mid Diamond 5. An entire evening wiped out. If there are forthcoming fixes that would help take the sting off nights like this, I’m all for it. Until then!

1 Like