This is the one thing that just makes no sense to me. If their ranking system (Trueskill 2.whatever) works as well as everyone claims, it is completely unnecessary to carry it over from one playlist to another.
Many people, including myself play other playlists totally different than the ranked playlist and for one to affect the other will provide very misleading data. Especially if they start adding playlists with game types like Griffball or Infection.
ABSOLUTELY this is another BIG issue! How do you quantify effort. You can have more deaths but singlehandedly won the match for your team.
I’m guilty of not going for oddball, I’ll be located to respond to a collapse if my team decides to make a push, but I won’t feed kills unless I’m super confident I can nuke whoever is on the other team.
It really depends on what the other players on my team decide to do. If I see them not pushing, I’ll focus up on kills and playing my life to save my rank.
People that don’t play video games or halo made this system and that’s what the real problem is. On paper it looks good, but to play it is garbage and makes no sense. fortnite does it better, old halos do it better. This system needs serious work and I ain’t kidding around, if they don’t this game is done.
By Super Sweats do you mean players ranked above you? Because that would be significant.
But if you mean players that are playing at your level and you have to play hard to keep up - that’s what is supposed to happen.
Maybe running around and stomping bots just made you a bit soft and it took a game or two to get back into the swing of things 
If social and bot match form has an effect on the global part of your MMR - it may not be that important.
It depends on what it is affecting… and to what weight.
People are (anecdotally) describing a game or two of slightly softer opponents. This may not be everyone else’s experience (it’s not mine).
And as long as it doesn’t have a significant bearing on your actual ranked MMR it is really not that important. Especially if there are significant gains in terms of algorithm efficiency (speed and data size) by sharing that data between playlist MMR’s.
A tragic list of events. I agree that something needs to be done with the way quits (or crashes are handled). But that’s an issue that isn’t intrinsic to the ranking system.
But fingers crossed 343 are going to do something about it.
A brief look through your games yesterday…
A rough breakdown of your opponents in 8 games were; 3 onyx, 19 diamond, 6 platinum, and 2 gold.
Initially I thought that is a bit tough for a poor old Platinum player…
And then I noticed you are buddied up with an Onyx player.
An Onyx 1620!
FFS.
Um, yeah, that is.
No, they are better than me, I specifically looked them up and they are ranked higher.
And even if they weren’t, I shouldn’t have to be sweating to win a CASUAL playlist.
When I play Halo, I just want to relax, not be on the edge of my seat so I can “clutch” the win.
Past Halo’s were not like this (except H5 but at least they owned up their SBMM)
That might be my brother if it was demo, 1620 onyx -1 match at the start of the night and he wasn’t partnered up for the rest of the night.
Even still I do okay all things considered.
I still had this issue last night running with my friend peek. Pulled in an onyx 1500 and we’re both platinum and were not partnered up with an onyx.
But you CAN see that we are playing many diamond and I believe at one point an onyx1800. My issue isn’t the competition, it’s the reward/punishment for performance.
To be clear, my issue is not WHO I’m matched against -it’s the progression system that I have a problem with.
First up… the metric is kills/min and not so much a focus on k/d.
This helps them to rank players up faster, but not necessarily higher. It identifies the big fish (eg. onyx) in a small pond (eg. gold). As you play better opponents you can still go +ve K/D but it gets a lot harder to maintain a high kill rate.
Somewhat ironically players are trying to manipulate matches to get more kills. But unless they are actually increasing the kill rate - they are just wasting everyone’s time.
As for playing the objective. It would awesome if they could somehow quantify team play. As someone who isn’t that good - but prides themselves on trying to do the team thing I would love it.
But how do you quantify such plays.
And even if you could - how do you stop people becoming toxic about them? I’ve already had a couple of team-mates kill me so they can take the flag from me and score the cap.
This is why everything circles back to the win / loss. If you rank people over the journey then good team play will wash out in the end with a better rank.
You’ll occasionally get a spread of opponents.
It may be the only player that was available to fill the match.
It’s probably that they were buddied up with some lower ranked mates.
I’m sure some of your opponents are sitting there going why are we playing against an Onyx player (your brother or mate Demo).
As for your progression. It’s based on the shape of your curve and who you play (for and against).
I think the system does try and resist the lower ranked players in a squad moving up. But as per the posts above we are still seeing a lot of over-ranked players who were probably dragged up the food chain by very good teams.
On the same token you shouldn’t be chilling to an easy win either.
And that would make sense as to why some of these “diamond” teammates or enemies aren’t performing.
Some on my matches I’m FAR AND AWAY above diamonds on either my team or enemy team. Occasionally you will encounter players at their proper level, but it’s the exception NOT THE RULE in the majority of matches.
It’s creating lopsided matches, our diamonds either won’t perform like the systems is expecting or the enemy team has the same issue. It’s frustrating.
I hope the reset fixes some of this
It’s tricky.
You can’t necessarily judge a player on their performance in a one or one… or in a particular match.
I’ve had games where I’ve gone negative K:D but walked away knowing that I’ve contributed to the win. Similarly I’ve been in losses where I’ve come out with a respectable K:D but have to acknowledge I was one of the reasons we lost.
Even within games players can struggle with team combinations. Put four specialist snipers together and you probably aren’t going to have a great outcome - especially if the map or game-mode favours close quarters combat.
But at it’s essence the current ranked playlist is a team game and your rank is a reflection of how well you fit/play in that team. Solo players are always going to be up against it (especially in objective games). And organised teams are going to be better at carrying lesser members.
IMO there’s no good reason to base skill rating on anything other than team win or loss for a team based game.
I understand that their current system is probably a “faster” alternative at determining skill compared to strict win/loss, but in the end it will always be less accurate.
I agree. And I think TrueSkill2 holds to that core value.
But the maths!
And no, I don’t confess to understanding it.
In the paper they conclude (in regards to kill rate); There is a much better agreement between actual and expected win rate, therefore the skills estimated from individual statistics are more accurate.
And we’re not talking about earth shattering forces here. The base kill rate for a 50% chance to win is 1.4. Getting out to a kill rate of 4.0 only influences your win rate to 59%.
It’s just disappointing that people get fixated on the kills.
In discussing what performance variables to use they discuss that some game modes don’t favour raw kills; Even in modes where the objective is to score kills, there may be teamwork effects where players can help their team win without scoring kills themselves. We ultimately want player skill to reflect a player’s ability to win, not their ability to score kills.
How about the drm campaign problem that they just ignore
It would be. If it worked like that.
It’s not one big global MMR.
It some shared data between playlists (which makes sense from an efficiency and size point of view) with offsets for each playlist. We don’t know the break up of data but I imagine each playlist offset has it’s own mean and variance (which means they can function as quite separate skill curves).
So you still have functional MMR’s for each playlist.
Halo 5 had the same set up. People were quite happy to be Diamond 3 in Slayer, Platinum 1 in Snipers, and Gold 4 in FFA. It worked well despite sharing the same common central data.
And that central data will have some influence. There is weighting for time away from the game for instance - which would reflect recent form. It makes sense that is shared between playlists.
The problem is that we only have one ranked playlist. People are starting to jump at ghosts.
Source, please.
Different CSR, yes. Different MMR, no.
The TrueSkill2 paper is clearly talking about MMR - and it describes shared data with offsets for each playlist.
Halo 5 had multiple ranked playlists and clearly had MMR’s for each that functioned quite independently.
If it was one big MMR then entering a new placement in any one playlist would wreak havoc across all your playlists (when your curve is made volatile again).
Uny? Is that you?
Oh noes. The jig is up.
Ok Josh Menke.
I stand by the obvious observation that Halo 3’s rank system was excellent and well received, and Infinite’s is obtuse, inaccurate, abusable, and meaningless.
343 don’t know what they are doing here.