Ranked system is getting old, fast

Games Halo Infinite
141

Well yeah because there is a heavy correlation between having good stats and being a better player. Are you saying it shouldn’t work like that? Because it rewards players who do well?

In your scenario where it seems you want to stack and gain rank. You could be 1,600 and play 100 games and gain 10 CSR per win, always playing weaker teams, always as a 4 stack, always doing relatively well. You are now rated 2,600, but you’ve never had to prove it. You then play solo and seriously inhibit every team you are on because you are being plonked into a higher skilled lobby without ever proving you are capable of that level. That’s essentially what’s happening when you’re stacking at a high CSR. Unless you prove it, how could you reliably be placed there? When you solo would you want a teammate that has never played at that level before?

That is why all these players that feel they deserve a higher rank, shouldn’t be one, they’d seriously let down their teammates in a higher ranked lobby, consistently. If that wasn’t the case then you’d be able to consistently “stat out” in these easy lobbies and fast track your CSR for greater heights. But that doesn’t happen, so you don’t. If the system expects you to go 1.5 k/d in a lobby to prove you’re 1700, and you got 1.4 k/d, then you aren’t 1700 material. It’s a consistency thing.

How would Trueskill3 work for you ideally?

2 Likes
142

TruSkill3 would reward players on their ability to win. If you can win games against high ranked teams your rank should move toward their rank and their rank should move toward your previous rank.

If you play as a stack perhaps you get an alternative/separate rank for that stack that might start as the average of that stack’s rank and then move as a unit from there. That way stacks can evolve and play better stacks or randoms. More stacks could form if they added pre and post game lobbies. This could add a fun element to ranked play where random teammates might stack up for a while to see what rank they can get their stack to. This would also increase the ability of the game to match stacks against each other.

However, we shouldn’t have to look at in-game stats to figure out an individual’s contributions to a game. The system instead can test how does this player or group of players (for the stacks) contribute to the win likelihood of the match.

If at a certain rank the system determines you or your stack can win games against people at that same rank, then upgrade the rank of the individual or the stack (and vice versa). You shouldn’t play 15 games in a row against weaker teams. The system should give you a challenge match against a higher ranked opponent to see if your rank should evolve. If you get beat, maybe you are good where you are at and the system should give you some equal ranked matches before giving you more challenges/reach matches.

Like if a stack has some guy just call plays and rotate the oddball around and can win against Final
Boss they deserve a high rank as a stack. Maybe that oddball guy can’t play with randoms so his individual rank is low.

However, maybe a player can win solo but has a bad time with a stack. Let the solo rank evolve separately and then just rank the combined stack poorly until they can figure out how to win games.

This is sorta done with solo/duo queues, but not completely.

I really think stats need to be ignored. If you have a strat that wins games but stats poorly and you can execute it at the highest level that should be rewarded. If you can only execute it with your squad, then give that squad/unit a rank to recognize that. If you can execute it with randoms then your individual rank should reflect it.

This solves smurfing (as stacks would have to grind for their stack’s rank). It also rewards more individual play styles as maybe your insane comms and objective play make up for your bad aim. Vice versa maybe your insane br makeup for your allergic reaction to ball time. Or maybe you can combine the two, or maybe you think of something else to do that’s equally effective. All we need to do is figure out if you can win games. For instance, think about chess. No one cares if you captured all the pawns before you won the match, the rating system only cares if you can win games.

143

I don’t think the correlation between stating out and being a better player is there within similar ranks. Some players tend to lead engagements and other players follow. It’s harder to stat out as the first one into a setup. You have the least information. It’s far easier to cleanup the double after bating your teammates. This doesn’t mean you’ll win more, however if you do win it gives you a bigger boost and if you lose, the game assumes it wasn’t your fault.

I think what I’m saying is, I can stat out at my Onyx level. That’s how I gain rank. It’s not a smart play style for winning games actually, but it works in the matchmaking meta. The actual meta of the competitive play is something completely different. If you try to stack up or play with friends, playing as a smart team isn’t rewarding rank wise. In fact, it’s quite detrimental to my rank. So instead I just play alone to gain rank, then when I’m willing to risk it, I’ll try out the potential for good teamwork/chemistry with a stack. However, quite cruelly, the game punishes me hard for losses and rewards me little for wins, leaving me to go back to being a match making stat troll to get my Onyx rank back up.

This encourages silly behavior and is quite anti-social and isn’t actually creating good competitors or encouraging of good team work.

So why not just have two separate ranks within the same playlist. Give your squad a rank that can evolve, then allow players to have their own separate rank.

1 Like
144

That’s how Trueskill2 works.

1 Like
145

It won’t as the greatest weighting is winning games you are expected to lose. That’s the easiest way for the system to determine your/the teams MMR has an imbalance. Do it consistently and the system can determine you are the main contributing factor, increase MMR.

Statting out doesn’t make any sense if it isn’t directly correlated to winning more games and doing better than expected.

1 Like
146

Holy bleep. It doesn’t. It also looks at KDA. That isn’t your ability to win games.

It’s like judging Tom Brady solely on his yardage. He brings something else to the table. You judge competitors on their ability to win games, not put up massive stats.

Like everyone is telling you that stating out is how you gain rank. Right now the system promotes playing alone and stating out. You’ll win more than 50% of your games because that’s the long term expectation of a player who deserves a higher rank, but your losses will sting less than the wins.

So now at the top you don’t have the best players, you have a collection of the best lone wolfs, the best baiter of teammates, the best stealer of kills, etc

147

How dense are you bro? Everyone has spelled out their issues with the system.

In one post you admit the TruSkill2 looks at stats, then in another you say it is based solely on your ability to win games.

What I am saying is quite clear:

Your stack should have a rating separate for your individual rating. If your stack rates to Onyx 1700 but the individual are all Diamond 5s so be it. We know that stack can beat another team of Onyx 1700s (either individually ranked or a stack that has the stack rating).

However now with the stack rating, the individuals don’t have to sacrifice their individual ranks if they don’t gel well and the possibility of smurfing disappears.

When playing alone (outside of a stack) winning and losing is all that should matter. If I match up with all Onyx 1505 on my team against a team of all Onyx 1560s and win, we should gain the same and they should all lose the same. The best loser, shouldn’t have an easier punishment.

Again think about chess, all that matters is winning or losing agains higher or lower ELO’s. TruSkill3 should be the same system but with a separate stack ranking system.

148

No, I didn’t.

There’s no need to start name calling.

149

But when they aren’t in the team it screws their new teammates over.

Your idea is lovely, but it would result in severely inconsistent games.

1 Like
150

I do understand.

I just don’t think it’s THAT important.

The vast bulk of your rank reflects your skill. Pure and simple. Your ability contributing to the win.

There is a weighting to kills per minute (not KD or KDA) that is there to help fast track people from lower ranks to their actual rank - that people are now gaming to eke out a few more MMR points.

But that doesn’t mean the system is “broken”.

It just means that people are focussing too hard on a stat that’s just simply can’t be as accurate as they want it to be.

I imagine the skill distribution of your local population (time-zone / server) is more influential on your final MMR than going all out to grab a few extra kills.

People want to look at their MMR and say that their value of 1850 means they are a better player than someone on the other side of the world (a distant skill population) with a score of 1849. But you just can’t.

If I was 343 I would just remove the MMR fullstop. Divide Onyx up into divisions so that the really good players have something to show for it (default to 1 to 6… or for nostalgia add in a military rank or the good old 1 to 50). Then add in a skill based XP for people to grind (weighted to wins).

And finally give us skill stats to grind for. Show something specific for each game type that reflects a person’s skill and effort. eg. Effective damage per game (damage that leads to a kill) for Slayer. Metres gained per Flag run in CTF. Scoring time / total time for zones. etc.

1 Like
151

I wonder how Chess would go if you played in teams of four?

1 Like
152

Just trying to get my head around something…

People are trying to grind out more MMR by getting more kills.

But the metric is kills per minute. So you may be getting more kills - but at the cost of prolonging the game. Are you really increasing your kills/minute significantly?

And do good teams really fall for the trap anyway.

1 Like
153

I play in fireteam of 3 usually with the other 2 having higher ranks than I. Up until about 1 week or 2 ago things were generally fine. Though it doesnt make sense in slayer when I go 10-6 and they go 12-11 and get more points. thats a whole other debate too.

However, I want to talk about the last week or two where I have gotten 0 points for a win 3 times. And maybe thats not a big deal- except every single loss no matter what I have been going down double digits. Even on 3v4 I am losing 3-13 points. With that said, other fire team members appear to get t more points on wins and lose less on losses even if I take second. And dont get me started on assists and damage. Again, slayer is best example but if I go 8-8 with 12 assists I say its a better game than 10-8 with 2 assists but nope, bottom of leader board no points.

Even more so, we had two separate occassions of going on mad win sprees- 20 out of 25 or more and I am not even getting back to even from the few losses we take. Something is broken and needs to be fixed before even the most loyal players start to leave. Its no fun if you cant rank and have no interest in BTB.

This also creates an incentive to run out and die and try to end up on top of the leaderboard rather than play right. It should be an easy fix but it hasnt even been admitted there is a problem yet so Id love to hear other feedback regarding this.

154

i generally play in a fire team but today great example. 2 wins playing on my own. one i got 0 points and one i got like 6. next game was a loss 3v4 and i lose 10. makes no sense.

the peak level thing maybe makes sense when youre going up but vice versa going down isnt entirely true because the system wont be able to tell if you were just “role playing”. If i am defending the flag in CTF and have lowest score it doesnt mean i contributed the least amount toward the win

155

Flag camping is not a very effective strategy so you should stop doing that in 99% of cases, as you’re leaving your team to 3v4 over other parts of the map. However, I agree with your philosophy. Just because you fulfill a role in a game that equates to less kills, or means sacrificing your life more to get the W doesn’t mean you didn’t contribute to the win. However, this system works over every game, so you’re being judged on every game you have ever played. You won’t consistently be the lowest scorer or lowest kill getter, but if you are then that is a correlation of skill in most cases. Again, it’s a consistency thing.

Like do you always perform the worst against higher level opposition and better against lower opposition. Have a look at your games on Halotracker, if the system can predict a winner it will say on the top right when you click on a game Eagle has 75% chance of winning. Of course if you win a game like that it doesn’t prove your rank should be higher, your team was expected to win easily. You need to consistently perform at a higher level against opponents the game sees as equal to or above you, that proves that you are deserving of a higher rank.

I’m sorry to all the people that are hitting their current ceiling. You are frustrated with the system, but honestly you will start ranking up when you get better at the game (which yes does mean winning more games expected to lose and performing better in general stat wise). Doing well in games you are expected to do well in is unfortunately not proof you are ready to go up in rank just yet.

2 Likes
156

I play in 1600-1900 lobbies on a regular basis and also almost to a 60% win rate yet I’m at 1530 currently. My highest CSR was 1595, and since then my Win rate has went up 3% and my CSR has went down. You can make all the excuses you want about why I’m not ranking up but this is clearly showing that Wins do not matter enough in this system.

1 Like
157

Yeah. The ranking system doesn’t work at the highest level, full stop.

There isn’t an incentive or a reward for making heads up plays. Wins and effective play style don’t matter enough.

I strongly believe stacks need to be ranked as a unit and matched based on that stack ranking.

Individual ranks when solo queuing should be based on your ability to win games or contribute to the winning expectation of that team.

The system shouldn’t have to look at any in game stats to do this. Instead, it should look at the relative change in win rate when a player is in the game.

This system rewards the trolliest behavior. Imagine playing team slayer, literally watching your team mate get two people weak, letting them die just so you can get two kills. Well congrats, you are now a good Halo player according to TruSkill2. To be frank, it’s not the worst play (at least you can get the 2-1 exchange in kills), but it’s not the best play (the 2-0 exchange in kills). If we looked at winning expectation contribution for each player, that would be born out.

I honestly can’t trust people who say this system is optimal. There is no reason to be playing strongholds with Halo players who ought to be in the top 5% and see everyone running around like dogs chasing their tails and solo capping B on recharge off spawn.

If you want better team mates and players who understand strats and game situations you have to reward it. We need a better way to estimate Game IQ. Anyone can aim in this game, the BR is literally brain dead. Anyone who puts 10 minutes in can super slide. Anyone who puts 30 minutes in can hit useful jumps. The ability to recognize and correctly play situations is not born out in any stat this game currently tracks.

If you want Onyx players who run to the oddball one after the other solo pushing and otherwise run around like headless chickens, then congrats you have Halo Infinite.

It’s also very frustrating to see stack play discouraged by the ranking system. Playing in a stack and being magically expected to be better really messes with the process of trying to find good potential teammates.

There is a reason why the top players play 8s in Discord and laugh at Onyx rankings. I just think y’all aren’t in on the joke.

158

Don’t insult people by saying they are hitting their ceiling. Can you honestly tell me that players are rewarded enough for making heads up plays and good teamwork in the current ranking system?

Why does ranked play look nothing like 8s play or other competitive play? Have you even played 8s?

159

Why does it matter? We can rank Machine learning algorithms in chess, we can rank a team as a unit in chess as well.

Also re: kills per minute

A game that can be won with 15 total kills in four minutes could be extended to 4x the amount of kills in 2x the amount of time.

Imagine a CTF game where instead of capping the flag, you just troll and get more kills. In the proper effective way of playing the game, you’d have 5 caps in 8 minutes: game over. In the other way that is rewarded, you cap once or twice then just slay for the rest of the game.

160

Yes you can.

But it still comes back to the problem at hand. How do you rank those team members as individuals?

I can imagine this scenario if one team is super jacked compared to the other team.

If you have two teams of pro-players… how much milking is going to happen?

I don’t think anyone was trying to be insulting.

But I also think people are hitting the ceiling so to speak. At least in terms of the accuracy of the system.

The actual number of your MMR is probably not accurate enough to compare directly to your housemate… let alone someone else in your immediate region. There are simply not enough games vs each other to sort the data enough (you really need a tournament type structure). The numbers become less relatable with each population apart (servers / time-zones). To the point where someone in England really can’t compare their MMR to someone in New Zealand.

So why have we set it up as some sort of badge of honour for people to grind?

You are given an MMR of value X. This will be accurate to plus or minus Y. The problem is that Y is probably a lot higher than a lot of people think - and increasing towards the higher values of M). And it’s especially frustrating if you are sitting, through no fault of your own, the lower range of X - Y. And then, as we mentioned, the fact that X in one region may not be the same as X in another.

Yet, despite this, people lose their minds that someone else may have a few extra MMR points than them.

Genuine team players are rewarded with the win.

You are already telling me how people are abusing the kills per minute… so should 343 take it back to purely wins?

But the data clearly tells us that factors such as kills per minute and a weighting for squads improves the predictive ability of the system (which is what it is intended for).