So SBMM is designed to decide when you lose…?

Games Halo Infinite
367

From the sound of it, most of them just need to learn HOW to relax.

The intensity to which they play is entirely up to them.

You will only attract opponents to match your level of output. Play at a chill level and you will get comparatively chill team mates.

The problem, as always, is when said player wants to go at 100% all the time - and then expect that “social” gives them opponents that don’t.

Agree 100%.

But it’s silly to have the exact same maps, rules, and set ups in Social and somehow expect people to play differently.

There is so much more they could do with Social to encourage EVERYONE to have fun. Multi-teams, themed games (eg. infection), more asymmetrical maps with random pieces, and (my favourite) handicapping games.

368

That’s the issue though. With a coin flip, no matter how many heads or tails you’ve gotten, the chance is always 50/50 that you will get one or the other. With this game’s system it directly forces mismatched teams in order to artificially influence the win rate so you have to have a close to 50% win/loss over time. That’s why instead of it being a 50% chance to win with a small window of deviation due to availability variables, etc. the game will consistently make matches 60/40 or worse in order to try and encourage you to lose when you’ve won too much lately.

It’s not a coin flip, its RNG.

Multi team was the best goof off playlist of all time. You could literally just go in and get 50 kills a match using anything you wanted.

1 Like
369

And honestly, it’s the best anti-sweat mechanism ever.

3v3v3 was perfect. It’s amazing how that one extra player added chaos to the map.

All the sweaty tactics you used in 4v4 went out the door. You were always outnumbered 2:1. Equipment and power weapons are still handy - but you can’t dominate the map with them like you could in 4v4. There is pretty much no map control in 3v3v3.

And best of all… when one team got out in front the other two teams would kind of team up on you. Which meant games were contested for longer, and blow outs less.

This.

There was always someone to shoot at. None of this ranked sweaty cat and mouse stuff.

Lots of kills. Lots of medals up for grabs.

In Social games - chaos = fun.

If I was 343 I would literally make 3v3v3 the staple Social playlist.

1 Like
370

Nope.

The match maker would prefer to have a smaller range… but low population is making it harder.

As for the 60/40 stuff - we can’t use HaloTracker for that. It’s based on poor data and too inconsistent… and as I pointed out from your playlist - half the “underdogs” got up - which suggests it’s closer to a 50:50 then their numbers imply.

I’ve played 17 games to date (Season 2.5) and the average team MMR difference is 8.8. And the biggest difference was in placement; 35 points.

Aren’t they the same thing? Or are we using a different acronym?

371

It was and idk why it’s still not in the game.
Multi team on boardwalk was goated. Especially on crazy king.
If they ever bring it back, it shouldn’t have any SBMM involved as it’s supposed to be fun chaos, not something to sweat over.

1 Like
372

That’s the thing.

Non sweaty game types plus loose SBMM = win win for everyone.

The problem is when sweaty players are allowed to sweat… it doesn’t matter what else you do - they’ll sweat until the cows come home (and then complain about it).

373

I think we have a different definition of sweaty, because I still wouldn’t be taking it easy on people in Multi-Team.

A few games ago I had a match in which the enemy team had a 73% chance to win, your argument, although succinct and therefore welcome, is wrong.

No, it just means ranks aren’t accurate to people’s skills and most of that is due to the matchmaker gatekeeping people out of ranks they should be at with the weird way MMR influences everything.

They’re ranked onyx and therefore have a higher chance to win, but you destroy them because they’re bad onyx players and you’re good diamond players, as an example. That means the ranks aren’t accurate.

A coin flip logistically has just two outcomes, making every flip a 50/50 chance between heads or tails. Given the wide array of variables in these matches, that doesn’t produce anywhere near a logical picture. Sometimes you catch the coin and throw it back in their face.

1 Like
374

Multi-team kind of washes away the sweat with chaos. :slight_smile:

You have room to sweat when it’s always 2:1 - and even less so when the other two teams naturally tend to gang up on you if you get out to an early lead. It kind of let’s you be you without necessarily taking the result out of anyone’s grasp (which is essentially the idea of SBMM).

When we had custom nights we loved playing 3v3v3. Short and sweet. I think it was to 35 points or something. Every game was intense and quick. No time for blow outs. Fiesta types were multi-kill feasts.

I can’t wait to try with equipment.

But overall - it’s a perfect game type that welcomes everyone. It should be the default for “social”.

It’s also very hard for you to be “right” when you are basing everything on a stat with somewhat dubious origins.

The ranks are accurate to the individual. The games are then balanced to the team’s average (plus weightings).

And that’s all the match maker can do.

There is no “gate keeping”.

No. It doesn’t mean that the individual ranks are inaccurate.

It doesn’t mean necessarily mean that they had a higher chance to win. That down to average MMRs and weightings for squads etc (which is why the HaloTracker is only guessing with the CSR).

And if, for whatever reason, your Diamond ranked team “destroys” an Onyx ranked team - the woo hoo. Your MMR will surely go up! And if you can sustain that level of performance and show over a period of time that you can do it 50% of the time - then you will soon be the Onyx team.

Keep in mind that a Diamond player should be beating an Onxy player 1 in 4 times. That doesn’t make them Onyx. They need to do in 1 in 2 times to be that. But it also demonstrates that there isn’t some magic barrier between an Onyx and a Diamond player - they can be very close in skill. And it isn’t hard for a Diamond player on a good day to destroy an Onyx player having a bad one.

There are indeed a lot of variables. Way to many to guarantee any form of team cohesion. This is always the lament of any match making system.

So you rely on RNG to smooth it all out over time.

But the algorithm isn’t specifically selecting out any of these variables to make any one individual suffer. We are all suffering bad team-mates, bad ping, desync, etc. It’s the same RNG mix for everyone.

1 Like
375

@M1STA_WU1FY

I don’t know if you saw my posts earlier with Pheinted.

But it looks, from preliminary review, that the game keeps track of your KPM in different game modes (eg. Oddball, KOTH, etc). I clearly have a KPM of 1.5 in Strongholds for example, but only 1.2 in Oddball.

Infinite may be following the H5 model of having a separate MMR for each game mode and then using the average of these for your playlist MMR - which your CSR then “chases”.

So, if you are on the cusp of Onyx - the low hanging fruit to improve your average would be to find your weakest game mode and specifically look for ways to improve your performance in that game. Getting your MMR up in those modes will have the greatest affect on your average (and thus CSR).

ie. Specifically look at your W/L and KPM for each game mode.

Mine appear to be Oddball (poor KPM) and CTF (poor W/L).

1 Like
376

You have no idea how it works I guess lol. Rank means very little

There are plenty of people in low Diamond that don’t deserve it.

It’s called the MMR system that is forcing a 50% win/loss rate

It depends on how good you’re doing whether it gets better or worse from match to match.

1 Like
377

Ok.

Ok?

As in they should be ranked higher or lower?

And don’t confuse individual skill with team skill. I’m sure there are a few players that are probably better as an individual but have no idea how to get the best out of working with a team - and similarly plenty who don’t quite hold up in the individual skills, but excel in a team structure.

It’s called a skill ceiling.

Or how frustrated slash paranoid you are.

It varies from match to match. You do slightly better in the matches that suit you. Not so good in the matches that don’t.

Your rank is the average of your performance.

You stop ranking up when you hit your performance ceiling. Not at the whim of some vindictive software engineer at 343.

1 Like
378

Both. The ranks are not accurate across the board. I’m sure there are plenty of silvers that should be gold, and gold that should be platinum, etc.

No such thing as a skill ceiling.

It depends on how the matchmaker has set up the circumstances. You play a match against onyx players with a platinum teammate and you’re supposed to lose. You can try and maybe win in a long shot, but the game WANTS you to lose.

2 Likes
379

ranked arena
Matching is so bad that the number of active players has decreased significantly.
They should be matched with others who are as close in CSR as possible.
Everyone feels negative when they lose, but if the matching is satisfactory, there is still salvation.
If there are too many unreasonable matches, the number of people will decrease even more.

380

A very subjective view.

And you are basing this entirely on one game of experience playing with and against these players? A couple of days ago you lost a game of Oddball with a score of 21 & 35. No judgement from my side (I’ve done a lot worse) - but I wonder what your random team-mates thought of you and your ranking from that sample size of one.

If the overall W/L of the games matches the expected from the ranking - then the ranking is correct (for the players relative to each other).

And no, this method of self validation does not mean the results are rigged.

I consider myself a low Diamond. Possibly to mid-Diamond as my absolute skill ceiling. I ended H5 in that vicinity. I started Infinite in that vicinity. I am still ranking in that vicinity. I am getting even matches.

Bottom line is the ranking seems consistent.

Is anyone you know not ranking consistently compared to previous ranks? Taking into account that everyone was two to three tiers over-ranked in Season 1?

Google it.

There is also a skill floor.

If there wasn’t such thing as a skill ceiling there is no reason I couldn’t be a professional tennis player. Number 1 if I wanted it bad enough.

Why does it want you to lose?

What were the MMRs of both sides?

The system only uses the “expected” result to allocate rank. If the result was expected you don’t gain or lose much rank. But if the result was unexpected then your MMR will move.

If you think you deserve to rank up then you should be embracing these “expected” losses. It’s a chance for you to show the system you are better than it thinks and rank up!

Or, you can perform as expected (ie. commiserate with your actual rank) and tread water where you are.

Again. Let’s have a look at real life examples. Your last 10 games where HaloTracker guestimates the % chance of winning.

The last ten game your team had (upsets marked by an asterix);

  • 40% chance and lost.
  • 61% chance and lost. *
  • 47% chance and lost.
  • 43% chance and won. *
  • 51% chance and lost. *
  • 45% chance and won. *
  • 44% chance and won. *
  • 35% chance and lost (but 3v4 out of the gate).
  • 50% chance and won.
  • 33% chance and lost.

So, out of 10 games… there was one that was a write off (3 v 4) and one genuine 50:50. And then out of the remaining eight games five of them went against the Halo Tracker odds. Including one that was a 61% upset.

How does this in anyway show that 343 are rigging the games?

Over half the games (including a game you lost as the 61% favourite) were won by the side you say 343 are setting up and “forcing” to lose.

1 Like
381

I ranked Gold 5 when I first did my qualifiers in season 1. Oops

382

:slight_smile:

I think I was Gold 4!

I blame it on getting completely lost in all the new maps (I get disoriented easily) and dying a lot when my thrusters (ala Halo 5) didn’t work.

383

Well pros are getting placed low platinum in doubles since the reset. A lot of people are finding it easier to rank up since they fixed some of the issues this time around apparently. Maybe this discussion about mismatching teams is moot because several friends said they aren’t getting wildly mixed lobbies lately.

And everyone I know got to the same point or higher in season 2. I didn’t play much ranked until the end of the season or I would have gotten higher myself. I only hopped back on the game when I heard ranks were resetting soon.

Sounds like you just don’t want it bad enough.

To achieve a 50% win/loss

Why does it set up matches it expects you to lose?

Because almost none of the matches are within 5 points of 50%, which would be a reasonable variation.

It helps to play with friends who communicate and work well together. That % chance is irrelevant then, because its not based on the ranks and its not based on the matchmaker throwing random groups together with a dud on the team specifically to cause a loss.

I mean, one game we were searching in a group of 3 and had an actual ex pro on our team and destroyed the other guys. One of the enemies was even an onyx over 1700. Funnily enough it gave us only a 43% chance of winning.

Or maybe I’m just Han Solo

384

Do you have a link - or a name? I googled Infinite, doubles, rank etc and got zilch.

Just need a bit of context. Which pros? How many. Who were they partnered with. How long did they stay Platinum for - and what are they now?

Seems like it was probably just the match maker struggling with a lower populations all along.

So, consistent ranking then.

By which stage your rank would have been relatively stabilised - and you didn’t perform above expectation after a couple of months off.

This surprised you?

Oh, I want it.

I still haven’t found any of these matches you are expected to lose.

Or win.

You seem to be in upsets at least half of your games.

So the game is producing even matches most of the time.

Good work.

So, another revelation. An organised communicating team performs better than a bunch of strangers thrown together at a minutes noticed.

Colour me something.

Depends on a lot of things.

How ex- was the pro. How pre- were your opponents. They may have been a high functioning, in form, squad of four.

And that’s before we even begin discussing how useless the CSR are for estimating the win/loss of the game.

You do seem like the shoot first type.

385

No they’re just not forcing bad matches as much, because people don’t want low ranked anchors on their team and have been complaining about it for a while now.

No, you said people were overranked in season 1. It can’t be both.

No, your math is wrong if you think that. Having a 40% chance to win is not an even match. I’d concede that a 48% chance is about as even as you can expect with so many variables. But you can’t sit there and be honest with yourself that an actual algorithm arranged 33% chance to win is an even match.

More than you know, less than you think

In Infinite you have to, the majority of fights are solely determined by who pulled the trigger first.

386

It doesn’t make any sense for them to deliberately sabotage the basic principle of ranked matchmaking.

Why would they do this?

They are not forcing bad matches.

They have only really been complaining of it recently, when the population started to drop precipitously.

Previously there was just the odd murmuring when you came up against squads that forced wider ranking into the game. And then I guess a bit more prominently late in Season 1 when teams worked out the could grind CSR.

The population was still ranked correctly in regards to each other. Just the ranks were skewed to the right.

It’s like if they only let Onyx players into Season 3. After a while the players would spread themselves Bronze to Onyx. The ranking system sorts the players it has into order - but the scale is completely arbitrary.

Currently that sub-population is spread from 1500-2200+. It’s no less relevant to re-scale them from 0-2200+.

Nobody said it was.

All I am saying is that the match maker produces the most even match it can with the players it has available.

And if the match is lopsided then the prediction is taken into account. You still have to beat the better side to rank up. And losing to a much higher rated team is not going to do much to your MMR at all (unless you weave some magic KPM wise).

But regardless - it can keep ranking the players accurately if the predictions are correct. It actually needs a difference to work - as in it needs to know how big the upset was to work out how much to change the ranks.

My arguments are that;

  1. The bias for and against is equal over time for all players.
  2. The % chance is a lot closer than you are stating.
  3. HaloTracker’s estimate is a guess at best (based on CSR with no weightings for form or squads).

And largely irrelevant if those factors are working equally for and against all the players.

It’s like ping. If you could make everyone play every match with 200ms ping - it would by yoink of course - but players would adapt and the rankings would spread out pretty much as they should. An Onyx player at 200ms ping is likely to beat a Diamond player at 200ms ping by the same margin.

Your ranking is your average performance over time. It doesn’t pivot on one game of bad ping plus or minus a team mate who doesn’t know what they are doing.