Was that an oddball match by any chance?
It seems to be weighted quite a bit.
An example of a match I had last night: Strongholds on Training grounds. We lost by 104 points. I went 14-9-7. I gained 1/3rd a bar LOL. My teammates stayed the same, they did “ok”.
The enemy team barely went up any except for the diamond 2 they had, and he went up half.
My team ranks: D3, d3, d5, d2
Enemy: Onyx 1556, Onyx 1610, d2, d5
Im ok with that. The win could be more weighted for Diamond+ and higher, but for lower rank platinum and lower, the personal performance and objective should be more weighted.
You are right if we are talking about a stacked team. Very few matches i play with friends and we communicate.
But most of my match are with 3 random player. Sure, even the best player of the losing team lost the match. But why should he be penalized when his performance were good? Like playing the objective, getting capture, stopping the carrier, holding the ball, going positive K/D in a Slayer match?
I dont recall but my kd wasnt negative. Probably enemy team had lower ranked people so I get punished for it. Zzz
1 Like
As far as I’m aware you always rank up if you win, even if your k/d is negative, as it simply looks at if you win or not. Individual performance counts for nothing.
Unless it’s oddball…
It’s possible you could win 2 of the 3 rounds and therefore the match but still rank down.
This is because it’s bugged and checks the total ball time for the two teams rather than the number of rounds won.
Example:
100-90
50-100
100-90
You win round 1 and 3 and therefore the match but the game sees your total time as 250 and the opposition’s is 280 and ranks you down!
Hopefully 343 can fix this sometime.
Oddball is the most frustrating ranked gamemode. Not only because it puts you 1 man down, the system it used to calculate your MMR is busted… but the two glitches to ruin the whole experience.
1.) You can use the drop and throw buttons next to a wall, and have the oddball sit inside a wall the entire match. You can score 1 point, and win the entire oddball game because it’s stuck.
2.) Cause the oddball to spin around the map and respawn after 2 minutes, by crouching over the spawn before it spawns.
Wow, have never heard of those. If this happens in ranked that’s some serious issue right there
It does, youtube some videos on it. It’s absolutely disgusting this has not been hot fixed ASAP.
Yet if there was an issue with the $$ store that caused people to get free items you can bet that would be hotfixed immediately.
Sorry, I’m being cynical now - ignore me
1 Like
I honestly think that sometimes the visual is bugged. It may be red or blue depending on whether you lost your match but it may actually go up or down based on your performance within that match.
Keep in mind that if the system expects you to win a game by a landslide but you only win by a hair, you may also lose MMR, although it’s less likely.
That’s one way of thinking about it, but the goal of a matchmaking algorithm is to 1) evaluate individual skill and 2) create matches that are as even as possible. In order to do both 1 and 2, you have to consider individual performance.
If you were to create ranking solely based on team statistics, the algorithm would have a much, much noisier picture of everyone’s skills and would create much less balanced matches as a result.
If Ranked were like a tournament where teams were a permanent construct and ranks were at the team level, it would be different. But it’s not, your ranking is yours alone.
Keep in mind the point 2 is far more important than point 1.
From reading various articles / posts / papers etc… they have looked at lots of ways to evaluate the individual player. But it keeps coming back to the win. Nothing else adds any significant accuracy to the data.
With “accuracy” being defined as the ability of the ranking to predict the result of the match.
It also has the benefit of being consistent across objective games. We all know that KD etc is only a small part of the picture in chasing the objective.
And when you think about it… what would be a better reflection of team orientated playstyles that contribute to the win than your W/L ratio.
Halo 2 tried a system where they ranked players within the match against each other - and you could definitely rank up being the best player on the opposing team. But it clearly created a selfish game style where people would protect their K:D even if it meant losing the match.
As a high Platinum player I get to play with a lot of Diamonds / Onxy players - who are clearly very good. You can tell which ones are the “team players” and know that part of their job to win is to nurture / protect the lower ranked players… versus those who are in it just for their K:D (and just scream at you to be better).
1 Like
I think people get frustrated when they don’t feel rewarded for their personal efforts.
I get that. I’m often sitting in that very same boat. It sux.
But keep in mind that a lot of very clever people have been working on ELO type ranking systems for decades. They have looked at lots of ways of doing it… and to date, TrueSkill2 is one of the best out there.
It’s complicated enough when you are dealing with a solo sport (eg. Chess or Tennis) but even harder when you are trying to reflect an individual in a team sport.
And the problem here is that we have no personal connection to the “team” - it’s just a bunch of people we played with for 10 minutes.
Subjective data doesn’t work - and is open to all sorts of misuse. The algorithm needs simple objective data points.
The purpose of the game is to win. And keeping win based rankings hopefully keeps all the players focussed on just that.
1 Like
I don’t agree with number 2.
Im a high diamond. But my friends are golds. I don’t care what the odds are, I just want to play some ranked.
In overwatch I could never play with my friends because we would be so far apart. This made the experience really suck.
It’s not really the odds or people wanting to play with their friends. It’s that, your friend will invite you to ranked, knowing you suck or vice versa, for an easy W to boost. Or they have a burner account that’s low ranked, and they have a onyx or high diamond play it… for another easy W. Plus it doesn’t hurt them as bad because they have a low rank on their team if they somehow lose, but it hurts the other team worse because they have a lower MMR player.
I agree 2 is more important than 1, but they’re impossible to disconnect from one another. The whole point of TrueSkill2 (which we’re assuming is still relevant, but we don’t actually know for sure) is that it incorporates individual performance (primarily kills per minute), which vastly improved its ability to predict win rates compared to original TrueSkill which only cared about wins/losses. You can describe it as “minor”, and that’s true in that the win/loss is the biggest variable, as it should be. But the paper is clear about how much better the model is when it starts caring about individual performance.
I like the idea of your last paragraph but I’m not sure how it works in practice. As an Onyx myself, the best way for me to help lower-ranked teammates is to carry the team by slaying as much as I can (and, of course, playing objectives where appropriate). I’m not sure what nurturing or protecting means in the context of a game–am I supposed to shadow them or act as a human shield? The best way to win is almost always outslay, in virtually all gametypes. Of course, that certainly doesn’t mean being a jerk!
I don’t think that’s what he means. I think he just means slow down, and help out your teammates. Slaying is all good, but if you aren’t running flags, and protecting and putting bods on objective, it’s all for naught. I’ve sacrificed my KD more than anything in this game to win matches compared to past Halos. Sometimes teammates straight can’t pick up the ball or run the flag. You got to step up there.
I’ve had teammates not pick up the ball when we had 5 points left to win… it’s insane.
Maybe. In my opinion, if one person on the team is substantially better at slaying, the best thing for the team is for them to focus on racking up kills and allowing their teammates to hold ball, run flag, etc. Objective modes still come down to slaying for the most part. You can only run the flag if you’ve knocked the other team, can only hold strongholds if you’re killing the other team, and can only hold the oddball if you’re consistently keeping the other team from coming at you with full strength.
Of course, I’m not saying ignore the ball, ignore the flag, etc. But if I’m in a game and we have a really skilled player, like an Onyx 1800, sitting in police station or something with the ball, we’re not operating well as a team. That person should drop the ball and let someone else hold it. It’s about comparative advantage.
All this being said, yes, it’s important to communicate, be a positive teammate, and all of these things. I don’t want to come off like some hardo who only cares about their K/D. Team play is important to win, objective play is important to win, and all that jazz. I just think that if someone is strongest at killing the other team, they should focus on that as much as they can.
The paper uses the term “much better”… but doesn’t seem to put a value on it. Except I guess the table presented.
I remember the 343 guy on the old Waypoint forum saying that it does help - but is effectively drowned out by your W/L. The gist was that it was there but being on a team and beating a higher ranked opponent (or losing to a lower ranked one) tended to drown it out.
And someone else, with better maths than I, was discussing that the way the bias was applied meant that it was far more influential in a win than a loss (which is kind of frustrating for those looking to salvage something from one).
To me, the most interesting point of the paper is that they talk about having variables for other performance values. Obviously kill rates are useful in Slayer. I would love to see them use data for objective scores (maybe they are already) and in Slayer experiment with damage scores as opposed to kills and assists.
The type of Onxy player definitely makes a difference.
There are “lone wolves” out there - who you barely see all match. They really don’t care about anything than their K:D - and abusing you when we lose.
Then there are those who protect / nurture.
Yes, it’s really just basic team strategies - but some players are better at adjusting their playstyle to help (make up for) other player’s inadequacies.
Appreciation that their high K:D is better when my K:D is closer to 1.
It’s better communication. Sharing of power weapons and equipment. Waiting for you to catch up before entering the room. Drawing fire from the better players.
I’m not expecting them to dive in front me to take a bullet like a Presidential body guard. Just help me to be part of the team.
I am happy to be the damage sponge to help my sons. I’ll enter the room first and then take the hits so they have a better idea of who is the room, with what weapons. They can then enter and clean up.
I am more than happy to draw out the camo player so they can take them down.
I played a game the other day with a much better player. They would actually engage first and then step back for a few seconds. I would move forward to fight to distract the enemy while they repositioned / reloaded and then they would join in again. I got a heap of assists that game!
It really makes a difference. I’ve had games where I am the only Platinum in the team - where I’ve gone horrendously minus because my team-mates haven’t given me the time of day. And others where I’ve actually gone positive K:D because of the way my team-mates have included me in the play.