Isn't Ranked Just Swell?

I love when by ALL TRACKED METRICS I am top scoring on my team. Every. Game. UTTERLY CARRYING (the best ONE GUY can carry 3 invalids) dumpus, dingus, and moron who won’t just turn off their mics already as their callouts are ineffective and they’re 3 seconds behind on reacting to any and all action that goes on and yet because the game thinks I belong in the same boat as them, they lose 8 OUT OF 10 games for me and I’m apparently ONLY Plat 3? Yeah ok, suuuuuure. Next you’re going to tell me the Paramount+ show isn’t a soulless corporate cash grab and 343 is a competent game studio.

Comp isn’t worth it unless you stack the team with your buddies and FORCE it to give you any proper rank, otherwise it feels totally unfair to the average single queue-er

Half the things that should be weighted highly like playing the objective, literally the entire point of the game, means absolutely nothing to the massively vague “win or lose” So you’re saying I can sit around with my feet up and as long as my team wins I’ll be carried to a high rank but if I absolutely SWEAT MYSELF TO DEATH, nah g your team was bad, something you can’t control, ha ur bad <3 343

Literally no reason to even play if I’m not going to be judged on my PERSONAL competency.

8 Likes

I’ve been saying this exact point since launch. Most people in the Onyx level where carried there by teammates in the diamond/plat who actually play objective.

You can literally have two and half minutes of hill time and a positive KD and if you lose the game you’ll lose two-three wins worth of rank. Makes no sense.

3 Likes

It’s nice that 343 explained MMR and CSR, but without a visual explanation in the game interface or at least via the online site, we’re left guessing and frustrated. A visual representation of CSR and MMR and how the math works out would, at least, EXPLAIN how, when, and why any of us are suddenly placed in a match and overpowered and de-ranked or excel at a match or win, and then also lose rank.

What seems frustrating is a constant adjustment to get players at 50% win-loss. It’d be nice to play the game within a larger bucket of players and win or lose within that bucket.

3 Likes

I’m a little lost here.

HaloTracker has you at a 52% W/L. And a K/D of just over 1. This kind of suggests that ranking and match-making are working pretty well for you.

Glancing through your last half a dozen losses or so. Yep. You did well. But I don’t see a lot of evidence of you carrying and/or your team-mates bombing.

It’s a team game. And in essence your rank is a reflection of your ability to function within a team. Of course it’s going to favour groups of player’s who function as a team. If you want an individual skill rank you are going to have to wait for Ranked FFA.

But on the same token there are plenty of solo people out there who climb to very high ranks (just not me, sadly).

What do you mean by massively “vague”.

Ranking systems work on the principle of win or lose. That’s how they work. And you can’t get any less of a vague metric. It’s black and white. And the reason they come back to the win loss is because of how vague all the other metrics are.

And there isn’t any clear evidence, apart for anecdotes, that playing the objective is detrimental. But we do know that teams who play and share the objective duties win more!

You are being ranked on your team competency. That is, your personal ability to function inside a team. In a team game.

A proportion of people are being carried into Onxy by team mates who are being used to manipulate the opposition (smurfs and sandbaggers). I think this is the bigger issue.

For those genuine teams I think you’ll find that they genuinely share the objective duties. And I’ve noticed that the really good players seem to be able to carry the oddball and still get a solid K/D.

You shouldn’t try and make sense of it over the course of one game. You are more likely to lose CSR on the pull of your MMR than any individual performance.

Yes, there is so much more they could do.

If you haven’t done so - google and read Microsoft’s TS2 discussion paper. Check out the maths stuff put out by Moserware (it get’s way too hard for me). And look up on YouTube the GDC talks on Ranking (one by Menke and the other by Izquierdo)

I think it comes down to form. Matchmaking keeps track of form and then gives you harder or softer games to give you a chance to change your rank. If you haven’t actually improved in skill you will win or lose these games and keep going with your 50% win-loss.

The alternative is to just have games pitched at your rank - which is even more frustrating as it makes it harder to rank up.

And if you just increased the bucket - you have to increase both the harder and softer games. So you still end up 50%. But just a random spread of increasingly mis-matched games.

1 Like

Solo queue is the only way I see this system kind of working. Doesn’t work with friends. Might work with a stacked team. But I play ranked mainly for the no radar and br start, not the csr, already know where I lie roughly in the grand scheme.

2 Likes

It works if your friends are similarly ranked and you work well together. Which is kind of unlikely.

It doesn’t work so well if you’re across a wider range of ranks.

The squads that work really well for grinding rank are manipulating match making. They usually have two Onyx players with a mid-range player (often a smurf account) and a really low ranked player. The latter Bronze or Silver player “sandbags” the average MMR of the team and draws a bunch of mid to high Platinums for the Onyx players to grind against.

1 Like

Welcome to the club.

Its working slightly better than it was at launch.
Where going positive and winning games was causing you to de rank, because the post game screen sometimes put you on the wrong team.

Take a look here.
https://www.youtube.com/watch?v=geWwIwWv2Gc&t=1s

What’s worse is when they put ranked into the Progression challenge mix.

Like I don’t want to play ranked. I’m not bad but I’m also not the best, and I’m sure the Dedicated Ranked players don’t want casual players like me to go into the ranked system.

But we all know why they do it. They want you to buy those challenge swaps.

But the last time I used a challenge swap for a Get X amount of kills in ranked … it swapped it for a Win x Amount of Ranked kills, next swap, earn X amount of cummulative score in Ranked…

I’m like is this system Brain dead, clearly I don’t want to participate in ranked games in order to progress to a weekly reward?

is it this?

https://www.microsoft.com/en-us/research/uploads/prod/2018/03/trueskill2.pdf

Just looking at the trueskill2 explanation, it seems like if the math is all there, there should be a rudimentary way to look up:

  1. how the matchup came to be
  2. an explanation of gains and loss based on performance

It doesn’t have to be in-game either. An online breakdown and service record on PERFORMANCE or COMMENDATIONS (think Halo: Reach and even Halo 5) not just the battle pass progress.

If you’re playing a TEAM based game then yeah whether or not your TEAM loses is going to be the main factor behind your ranked level.

Oh and the whole “my score’s so much higher” argument? Are you really just that much better? Are you really 1v4’ing the enemy team while your teammates cower in the corner? Or is it something much more basic like you just have no synergy with your teammates, that you can’t properly communicate with each other, or that you can’t coordinate enough to have any meaningful impact? In team based games you can’t just go off of personal scores since that perception can be skewed. You have more kills? How many assists do your teammates have? Maybe you’ve been finishing your teammates fights. Maybe they’re drawing enemy fire enough for you to kill the enemy. Maybe if your not there to help them they keep getting teamed up by 2+ players.

Point is, there’s too many variables in a game to accurately tell your performance based solely on personal score in a TEAM game. The problems you’re facing are no different than most other ranked online games out there, and if you’re losing almost every game it’s not just your teammates that are the problem. Blaming all your problems on everyone else is the exact type of toxic personality that makes ranked unenjoyable.

Here’s more than likely the hard truths:

-You’re not a pro gamer, just some average shmuck.
-You had a run of bad luck, it happens.
-That rank is probably a lot more accurate than you care to admit.
-Even if you got a stacked team with your friends you would probably still be running into these problems.

The TrueSkill2 paper is very light on the maths.

Moserware goes into it in more detail;

http://www.moserware.com/2010/03/computing-your-skill.html

And there are links to even more maths there.

It made my head hurt.

I’m not sure the general public are accepting of your skill being a curve, let alone getting knee deep in Bayesian probabilities, expectation propagation, and factor graphing.

So how do you display it?

I agree that we should have more access to our info. As you suggest, make it out of game.

Maybe a diagram that shows our MMR curves - global and for each playlist. A bit like what Menke used to generate on request for Halo 5. I couldn’t find any of his waypoint images… but there are similar things on this webpage;

https://esports.gg/guides/halo/halo-infinite-ranks/

It at least demonstrates that you can show MMRs graphically.

Do we need much more than average team MMRs?

They just need better communication.

It is quite clear that the per-game gains or losses are almost entirely win/loss, the relative skill of you vs your opponent, and the influence of MMR on CSR.

Your K/D etc is pretty minimal.

343 just need to be better at communicating that.

And again. I would;

  1. Change the scale of the CSR. It doesn’t need to be 0-1800 like the MMR. Make it 1-50, 1-100, or 1-117 (someone else’s awesome idea).

  2. Cap the CSR. It doesn’t need to be open ended. Call the top level Champion to make it worth striving for and then let it go. Stop this stupid (and toxic) grind for MMR.

  3. You only need to show the CSR change for the player. Don’t show the other’s. It just upsets people when they seen someone else do better.

No kidding… jeeze. It reads like a card counting scheme only the system wants generally equal outcomes (an assumption that 50% winning is fun or presents a challenge. Woke equity in competition isn’t the right way to go about things since fairness in competition is only judged by equal starting points and following the same rules, skill, proximity to similarly skilled athletes, and athleticism determine outcomes, not an algorithm pulling strings to get equitable average win loss chances.)

Pro sports wouldn’t be entertaining if the leagues handicapped the exceptional players by pairing them on squads and teams where all teams perform roughly at a 50% win rate. We want to experience elation from crushing opponents, the nail-biting of a close game, the outrage at perceived injustice, the grief of losing, bending rules, breaking rules and getting away with it, creating new ways to push the envelope.

By predicating MMR and CSR on the assumption that predicting a 50% chance of winning or losing is fair undoes the notion of outcomes being judged on your abilities only by comparison to others instead of by accomplishments or class of skill.

I experienced machine/algorithmic bias in real time where I let my wife play on my account for a few matches. She kept getting paired with people who were increasingly better than her as allies and then playing against opponents that greatly outclassed her, it wasn’t fun for her. The average odds of each team winning… wasn’t really 50% and the algorithm kept paring her with better people only to make it harder for her to have fun.

In contrast, when I picked up the controller, I was paired with people who I outclassed, but the team was being paired up against 3-4 people who were much better than my allies but mathematically it seemed a good match. I kept getting paired with worse and worse allies (I’m not that good…) and opponents who were, in the end, more likely to win since I can’t carry any team on my best day.

The only fair matchups are allies and opponents of similar measurable performance/class, not a mathematician’s convoluted guess (programmers who attempted to write an unbiased algorithm) attempting to get equal outcomes (which is itself done by being bias).

The matches evened out eventually, I’m back to the usual intentional 50% chance win/loss cadence.

The algorithmic bias is what’s frustrating. The equality of class/skill of players competing with rules at the onset is what is fair in real life and at the professional level, winner takes all and winners build momentum in their class/skill brackets and become the best in class. An algorithm’s efforts to force a 50% chance of winning or losing frustrates really good players since to bring down their average chance to win, they’d have to be paired with people who aren’t as good as THEY ARE.

Matching this year’s champion lightweight boxer against last year’s heavyweight boxer isn’t a fair fight despite if mathematically it should be fair based on recent performance. the algorithm can’t take into account factors the programmers couldn’t predict would be factors that affect the guesswork mathematics like if the heavyweight lands one hit, the game is over.

(I don’t know much about boxing, so forgive the analogy if it doesn’t fit.)

What would be fair are matchups between boxers of similar class that, though their own agency, prepared or trained or had better opposition research to gain advantages for the competition of boxing.

If I could, I would play against better opponents than me every time. I learn a lot from really good opponents and learning from what they do.

I think yes. For example, if online ads were honest, they’d tell you which cookie or metric signaled the ad so you knew the stalking was, at least, something that made sense.

“We’re serving you this milk ad on Google because of a post you liked on Facebook.”

An honest explanation for a matchup wouldn’t have to get into all the details, rather, simple statements or notices or infographics to convey:

“You’re paired up against combatants with a 45% chance of your team winning (show MMR/CSR averages?). Due to the low pool of available players and to favor playing on a server closer to you, you’re playing against more experienced players.”

“You’re paired up against combatants with a 51% chance of winning based on average MMR scores and a high likelihood of good communication as a fireteam grouped in a party against a similar fireteam in a party. High availability in your region selected the server closest to you in Virginia with the best connection speed for 7/8 of combatants.”

Post match explanations could follow similar formats:

“In a match where you were expected to lose with a 45% chance of winning, your team won. Your performance was better than expected with 10 more kills than your average for the past season. Combined with your improved performance over the past 5 matches, you’re awarded 18% more rank advancement than had you won against a team with a 50/50 chance of winning.”

“In a match where you team’s chances of winning were 51%, your team lost. Over the past 7 matches with 3/4 of the same fireteam participants for each match, your individual performance improved overall but decreased for objective-based games. Given this information, your team overall loses CSR value, but yours increased slightly.”

If I remember reach, didn’t it have a rudimentary calculation of credits earned after each match?

Obviously, no one wants to read that nonsense above, but this information, if presented in a pleasing way, is what frustrated players lack, I think: an explanation for the change in CSR/MMR, how it was deemed fair to be paired up for the match against opponents or with allies.

Honestly, OP.

I’m more impressed by your four betrayals at Platinum.

Honestly I’m not trying to write an entire novella of a response to any of the massive walls of text anybody has put here but I think it helps my case here to note that my overall average K/D is of course going to be skewed downwards given that in objective gamemodes I’m not trying to play Slayer like my team, I’m on the objective as that’s the point. Of course I’m going to die more and get less kills when my goal isn’t to boost my K/D. Raw stats also arent a good indicator of my team and I’s gamesense and who was where at what time doing what action. But if anything objective time is the best indicator of that, of which mine was repeatedly high.

When you look at purely slayer in those last 10 games, and the objective is for me to kill, I personally think I did fantastic. My regret being that it only gave me slayer once and thus I cant use it to further illustrate my points.

I guess I could sum up my original argument by saying that I am no esports god. I am no epic gamer. I am slightly above average as all of my Halo stats throughout history have been. But its really annoying when I get placed with teammates who cant wipe themselves after leaving the bathroom let alone play a competent game of Halo. I’m annoyed that my stats reflect something that I can’t control as a solo queue-er and in turn that lowers my supposed measure of skill despite the system sandbagging me and any other competent players that get matched with me with deadweight teammates seemingly every game.

Most important and constructive part I guess


I dislike my skewed stats and I think the metrics the system uses should be changed to better reflect true player skill and promote quality teams at all levels.

Objective time, K/D in relation to your proximity to the objective minus of course team slayer in which its just standard K/D, and number of deaths or rather lack thereof should all be weighted based on gamemode played and should be what your “skill” is based around with a standard deviation taken into account for if you won or if you lost as that skews your stats up or down depending. I think in changing what it is to something along the lines of what i’m getting at, the system would feel more fair. W/L should MODIFY not dictate.

If theyre shooting for 50 50, my 20 80 is evidence that this current system DOESNT WORK.

Edits: ADD makes it hard for me to focus my thoughts.

Yeah, ranked is just pointless to be honest. For a single queuer you have to rely on luck to get a remotely reasonable team. It doesn’t reward objective play, it doesn’t read between the lines well to judge personal performance or different play styles. It only cares about damage per minute and win/loss. It completely ignores objective performance, accuracy, headshot percentage, and your performance relative to your teammates. To me those are equally as important to ‘skill’ as being a super aggressive kills only player. This ranking system hides behind a ‘complex’ algorithm, but actually it is pathetically simple and poorly thought through.

And most egregiously it makes the player feel cheated, on a regular basis. Playing an absolutely blinding game against much higher ranked players and watching your CSR barely move while one loss sends you tumbling back down is such a poor way to retain players. Why do you think this game has gone from ‘the best halo in decades’ straight after launch to ‘the worst halo ever’ in the space of 6 months? It is anti player, anti fun, anti fair play and the bare bones of what it had the potential to be.

2 Likes

That’s true.

But playing the objective properly should still free you up for getting kills. It kind of goes hand in hand to securing the objective. If you are dying excessively or not contributing to kills, then you probably aren’t doing it as well as you should be.

And look, you may end up with less kills per minute than some of your team-mates, but you still benefit from the win, and at worst you get slightly less MMR gain.

Don’t fall into the trap of comparing CSR gains. They don’t take into account any personal performance metrics.

It’s a tale as old as time.

How to take four random players and expect them to gel as a team. Especially sans communications.

Lucky, your opponents are probably battling the same issues.

They passed all the metrics through the Halo 5 data. That’s millions of games. And the only thing they found that added accuracy was kills per minute and (to a lesser extent) deaths per minute.

They did K, D, K-D, KDA, A, damage, objective scores, medals, etc. And essentially they found that players with good scores in those metrics did not statistically increase your chance of winning. In fact they found that some of them decreased your chance of winning (I don’t think they have ever said which ones).

Although they probably didn’t do proximity K/D.

The thing is that a lot of these things are required for the win. Teams that do them win and rank up. There is no extra ranking value to them over the win.

And the thing is, whether we like it or not, the real skill in Halo is winning a 1v1. If you can’t match an Onyx player in a dog fight, you aren’t Onyx (and I can’t, so I’m not).

And we can fall into the trap of playing the objective and “doing well”. But, for example, you need to be able to drop the Oddball and kill that Onyx player coming for you to be Onyx.

Intuitively yes. And don’t get me wrong - I’m a big fan of assists and damage.

But if you, or I, want to introduce new metrics we need to prove their influence. Microsoft have done the number on millions of data point.

That would be the nature of a team game?

I agree that we need ranked FFA for those who prefer to freestyle.

It does. If the team plays the objective you have a better chance of winning.

None of them have been shown to be useful.

KPM, DPM, and W/L.

It’s easier to deal with when you know that CSR isn’t affected by personal performance.

But unfortunately 343 have been very bad at communication.

Idk about halo infinite, but I know for halo 5 Menke would always say the match maker tried to put you in matches where you had as close to a 50% chance to win that specific match. That is different than trying to push your win rate to 50%. If it tried to push your win rate to a certain % the system would try to force you into winning or losing matches to accomplish this. Instead the system tries to fairly match you and give you an even chance at winning any given match. Given there are varying skills and amount of players queuing at a certain time the system has upper and lower limits of % chance at winning that it will match you in but it doesn’t try to force you into a 50% win rate overall.

I’m pretty sure the current system is actually doing this. If I goof off 2 games in a row (trying to complete challenges) and my teams get stomped, the next game I get matched into will be pretty much impossible to lose no matter what I do.