SBMM Has got to go

Games Halo Infinite
82

In ranked, I only play when I’m fresh, wear turtle beach headphones, listen to footsteps, callout and ping a bunch, focus hard on what I’m doing and the team. I always open with meta strategies and focus hard on map control. The games are often extremely tense and extremely close. I’m always thinking ahead about next steps. I’m in Onyx and have a 61% win ratio in Crossplay.

In social, I play whenever, sometimes very tired, listen to music or documentaries, never use my headset. Sometimes I will open with a meta strategy but often I’ll just do whatever I feel like doing. Regularly I’ll be going for challenges so instead of contesting rockets I’m running straight to the needler.

My win ratio on social is as follows:

Swat - 59%
Slayer - 63%
Quickplay - 59%
BTB - 51%
Fiesta - 64%

Most the time I’m playing social I’m doing challenges (sometimes that means being an absolute filthy tryhard and sometimes that means doing awfully as I try to do a basic action). The crux of it is I don’t have to put much effort into social compared to ranked. Usually ranked is highly intense, whereas social is often not. Some games are, but most games are fairly casual. I have great win ratios in social despite taking a very casual approach to them most of the time. It’s not like I’m trying to lose but the level of intensity is much lower.

BTB is one of my favourite modes and my lowest win ratio which is still 51%. I spend most of my time playing BTB fooling around with teammates and vehicles. I use a lot more of the Sandbox and particularly love getting kills with the fusion coils and turret on Fragmentation. Sometimes I turn it on, usually if there are other players on the enemy team beating up my teammates a little too much. I tend to only do really well if I am doing challenges.

It’s just clear as day to me that whether I’m top of the leaderboard or near the bottom, the game is still fun for me. I’m just playing how I want and freely. Compare it to ranks where I have to be 100% on at all times or we lose most the time. The games are so close so often it’s just a fact you can’t ease up. In social half the games I don’t turn it on at all, or I only need to turn it on in the final quarter to take the win back.

1 Like
83

Interesting. Reading your response, I can’t say I understand your point of view that you want SBMM to remain strict in social. Everything you just said suggests that you see social as casual… sometimes sweaty sometimes a breeze. Sometimes doing challenges and sometimes just mucking around.

Does it bother you that your MMR is shared across social and ranked? ie if you jump on and warm up in social then switch to ranked that your win streak in social affects your ranked games?

84

I don’t believe it should be strict in social. I’m saying it isn’t that strict right now. Any looser though and I might take issue with it if I’m constantly playing much worse players. The level now is pretty good. I still have to play fairly well each game but it’s nothing like ranked.

To put it lightly, your MMR number isn’t the same for every playlist. Being 1,700 MMR in ranked won’t necessarily be putting you in 1,700 MMR games for Grifball. The correlation between playlists is driven by player data. Most playlists we have now likely have a fairly high correlation but still doesn’t mean it will be exactly the same for every playlist and that games will have a major effect on the outcomes. I play a bunch of BTB (my worst playlist) yet that didn’t noticeably slow me down from climbing a few hundred CSR points.

Even if it did, I can’t say I care. I don’t care what rank I am provided I’m getting good close games in ranked. I’d rather be 1600 Onyx having 4-3, and 3-2 CTF games all the time than 2000 Onyx having 4-0 games. I’m not in ranked to get a higher rank, I’m there because the games are more intense and have higher stakes. Of course I like ranking up, but it’s not my primary purpose for playing ranked.

85

Actually MMR is your hidden matchmaking ranking. You can’t see what it is. But it is how the game ranks you when matching. (regardless of playlist).
Your playlist CSR ranking is the playlist specific rank. It doesn’t correlate to MMR which is why you can see a gold player who did badly when being placed in the first 10 games, and is slowly getting his CSR raised whilst being matched with Diamond players.

In rank playlists everyone agrees that tight competitive games is the goal.

I haven’t actually played infinite for a while… but it was tight on release. Maybe it’s more relaxed now. But to this day I cannot match in halo 5 because it was too strict. I simply could not find a low ping match. I would find a match when I played with a friend who is no good at the game. Or once I made a smurf account and I got plenty of games… but it only lasted a little whilst until the MMR was too high again. Halo 5 was definitely too strict.

86

Your MMR won’t necessarily be the same across every playlist. There is an offset skill distribution per game mode based on level of correlation which is determined from player data.

They did an update and I’m certain it got made looser after. The game was super strict at launch in social. Try it out.

87

The skill gaps between individual players (like you ask for here) are already very big. Just look at my matches from yesterdat and tuesday:
Yesterday:
12-4 Quick Play, Oddball
18-3 Ranked, Slayer
23-5 Ranked, KOTH
37-8 Ranked, Strongholds
7-8 Ranked, Strongholds (it was 2v4 though, so i had was in 1v3’s or 1v4’s every time)
17-7 Ranked, Slayer

Tuesday:
13-2 KOTH
9-7 KOTH
10-4 Fiesta
21-1 Team Slayer
9-3 Quick Play, Attrition
21-2 Quick Play, CTF
23-8 Quick Play, KOTH
2-2 Quick Play, KOTH
18-12 Quick Play, CTF

Does that like a very strict SBMM with very little gap between players? It is clearly having big skill gaps, even in ranked.

A lot of people disagree with you here, they don’t want to be carried by way better players, because they are just outskilled. Often they just quit those matches. Apart of the challenge system and crashes, this is where the quitproblem comes from. People just hate being in matches where they don’t stand a chance. It doesn’t matter if they win or not, they want to be able to play and not just be cannon fodder. On top of that you also hear the better skilled players complain that they have to heavy carry to be able to win matches and don’t want to do that. The problem is not that the skill gaps between individual players is to small, but to big.

You may not be, but there have been a lot that are (this is only the millionth topic)

You might be, but i have heared a lot of entitled players ranting that they don’t want close matches.

This is deliberately framing the question, since you put it so that you won’t allow the main and most important answer.
But other (less important) differences are:

  • different starting weapons (BR in ranked, AR/SK in social)
  • weapon spawns (the spawns in the racks are not always the same as in social)
  • radar (not in ranked, available in social)
  • gamemodes (limited gamemodes in ranked)
  • melee handling (BR 2 shots+melee in ranked, 3 shots + melee in social)

So yeah, there is a clear difference. The reason why you don’t ‘feel’ that difference is because of your own mindset.

No they are not. The ‘stress’ of having to win makes it play different already. Beside that the starting weapons are also different (BR <-> AR/SK) and those weapons also make you play differently. And no, social does not mean casual. Casual is just a mindset. You can play casual in ranked and still dominate, it depends on how much ‘stress’ you put in performing. The reason many complain about ‘sweating’ in ranked is because it’s their own mindset that is in the way. They complain about not being able to play worse while performing just as well (k/d wise) or even better and win the match. Ofcourse if you play worse you perform worse. The whole point about social is that that is okay, because you don’t have to win. Yet they just can get that in their mind and therefore complain about having to ‘sweat’. They just keep the same mindset as in ranked.

You can also get stomped playing sports on the highest level. Do you remember the stomping Barcelona got in the Champions League semi-finals against Bayern München (8-2)? But also here your comparison don’t fly, since the top teams play in the top competitions and not in between the smaller teams. That too is a form of SBMM. Just like in tennis only the best player will compete at Wimbledon and the lesser players have their own tournaments. Etc. etc.

Not asked to me, but i would say that MMR shouldn’t be shared between ranked and social yes. In fact: i think MMR (except for placement matches) shouldn’t even be in ranked. Ranked should match you based on your CSR.

88

I think it would be hard for me to judge. I stopped playing infinite because of the 150 → 250ms games.

If they have indeed broadened the SBMM range then maybe I should see if I can get local games. But it possibly isn’t a valid indicator that the SBMM truly is wider because there are probably other reasons why I might continue to get 150+ms games.

89

It appears that Halo Infinite may have changed since release. Perhaps I would agree with you next time I play it.

The reason I don’t “feel” the difference is because SBMM keeps matching me against americans with 150ms ping just to place me in close MMR matches… ranked or social. Connection should trump MMR in social. SBMM should fix skill gaps whilst prioritizing connection… or simple you will lose the international crowd across the pacific. Why do you think I haven’t played infinite?

Do you agree that Halo 5 had strict SBMM?

This might help me understand if we are on the same page.

90

Since you say ‘across the pacific’ i assume you are from Australia/New Zealand or East Asia? Problem there is that there are not enough players, so either the skill gaps must be enormous or you have to play on US servers.

I am from Europe and most of my matches are on the Amsterdam server where i have just 10-12ms. But from time to time i also am being put on US servers with 120ms.

For your question about halo 5: i haven’t played that one. I don’t have an xbox (i play through steam) and H5 is not available for pc.

91

Well Halo MCC enjoys a healthy australasian population. And it has a nice social setup with a varied SBMM… Admittingly match composer and server selection also contribute to this pleasant experience.

I will have to try infinite again… but the SBMM was the nail in the coffin for my region… thanks to 343 for not adding match composer, not adding server selection… blocking geofiltering… and then employing strict SBMM.

92

Huh?

First of all, the game already gives different experiences to different players according to their skill. If I solo queue SWAT, I can pretty much guarantee I need to go +20 to win because I will get three teammates who are much worse than the enemy team. Sometimes there will be another good player on the opposite team and he also will have three terrible teammates.

So, up to six bad players get farmed all game as a direct result of this game’s SBMM implementation.

The current system doesn’t prevent worse players from losing to good players. It just makes the game frustrating in social at the higher end of the skill curve, and boring for everyone because there’s very little variety in the social matchmaking experience.

That’s what 343 needs to do with their SBMM implementation.

Lots of threads on this forum and reddit have proven it; for example people showing players who do nothing but bot boot camp and then try ranked and are placed against opposition far superior to them.

I have already said I don’t necessarily think pure random is a good idea, and some loose controls would be ideal. If I had to argue positively for something, I’d say that matchmaking should be random with some level of +/- boundary on how far apart individual skill can be, and you put fairly tight restrictions on party matchmaking. Maybe you never match an onyx against anyone below platinum 1, and a bronze never above a gold 6, for example.

My fundamental argument is not, as you apparently think, “get good or get out”. My fundamental argument is that it is good, in unranked game modes, for there to be a lot of variety in game experience, and for the player to be able to take advantage of that variety according to their relative skill. I do not believe this communicates “get out” to most players, particularly if there is a ranked mode trying to make things as tightly fair for everyone as possible.

1 Like
93

I agree with this in theory… but people would intentionally place badly in the first 10 games and get bronze… then get to enjoy easy street to their actual CSR ranking.

1 Like
94

What would the point be? Your CSR is your actual rank, so if you play bad, it will only take you longer to get to your actual rank and the last part of that journey (where you would otherwise have started) would be just as difficult as if you started their from the beginning.

95

I’m suggesting people would enjoy having players as Canon fodder as you mentioned in your first post at the top of this topic.
(ie they wouldn’t care about the destination but enjoy the journey)

1 Like
96

But that is also why i said that MMR should be there for the placement matches. So even if you did bad in those placement matches, you still wouldn’t start at the bottom and the worst will be rectified.

Remember that that is just as the current placement system is, so there is no difference.

97

I have to agree with the other guys on this one as this is a symptom of a relaxed SBMM MMR range, but with SBMM just balancing the teams.
You are just carrying more frequently as you must be at the higher end of the bell curve.

Maybe the solution is to just make the social MMR fluctuate wildy… Ie if you get on a 5 game streak then the MMR is over corrected until you get beaten… But if its sweaty or you lose then your MMR is lowered. Make it oscillate wildy to give me that variety I crave

1 Like
98

It’s a symptom of the SBMM targeting “50% winrate” as its goal.

So yeah, it relaxes the parameters for individual players in the game, but it tries to get the total skill on the team to be equal.

This is actually what it does everywhere, but you don’t see it so much in ranked unless you are at the very top of Onyx.

Do any of you really think that six players all going -10 while getting farmed by one good player on each team is more fun in any meaningful way than a more random / high variance system would be, just because each team has approximately a 50% chance to win and the final score might be 50-48?

99

No not really but what the other guys were saying is that you asking for true randomness. Which is a bad deal if you are at the bottom 10% of the bell curve where you are likely to only win 10% of the time.

Or you are saying that the range needs to be tightened then add randomness… Ie as long as the skill is close enough don’t balance the teams.

But being in a low pop area I really don’t want the range tightened as I already can’t find games in ranked for that reason… I want openness in social for range. If I’m in the middle of the bell curve then I get that juicy variety.
But the extremities on the curve both dislike this apparently.

100

I wouldn’t imagine that this is a common scenario.

I guess having a handful of players that are +20 better than everyone else on the server is going to stretch the system. Maybe these are the best games the server can set up given the available population.

Not good. Sorry.

Those threads simply demonstrated how TrueSkill2 uses a previous MMR to seed a new playlist when you are being placed for the first time. This only happens if you start a new account, play a lot of bots, and then go into ranked for the first time.

I’ve seen no evidence (apart from vague anecdotes) that it makes any difference to a mature account. The MMRs do have some shared data across playlists - so there is probably some influence. But whatever it is, it is small and temporary.

Every Tuesday I go crazy playing with mates. We lose a lot. Badly. And yet when I hit up Ranked again on Wednesday - nothing has changed.

You would have thought by now we’d have seen some blog / youtube proof of someone rigging an account via bots and social to climb the Onyx leader board. But nope.

Fair enough. At least we are roughly on the same page.

Personally I am very happy with how it is pitched at the moment. I appreciate that is coming from my level of play - and there may be tweaks to be made at other ends of the rainbow.

I don’t think the answer is in widening the range though. It doesn’t work in customs. And it’s when the ranked system starts to strain a bit at the seams.

I’d really like 343 to experiment / dabble with some games that involve handicapping of players to even the playing field. I’m certainly going to be spending a lot of time in scripting trying to get our custom games sorted out.

101

I’m def playing less because of SBMM. The sweat is unbelievable and not fun.