Absolutely. New players aren’t entitled to a 1 KDR and 50% win ratio when you are in the bottom 10% of players.
That is the gaming journey. You play and improve. This is just modern society, everyone wants everything easy without putting the effort it, and ultimately it is pushing away the die hard players who are the actual backbone of the playerbase
That’s oh so, 2008.
And older players aren’t entitled to more.
And you also have to appreciate that half the population are going to be low Platinum or less. They aren’t going to “get gud”. And they don’t deserve to be playthings for a bunch of Onyx players who don’t feel like trying anymore.
It’s called progress.
Not clear. But we know not having it pushes away the fresh talent who are the future of Halo. And probably from a demographic more likely to splash their cash. 
I don’t know why it overestimates your skill, but i know that it does: i don’t play infinite anymore ( waiting for 343 to fix everything) but one of my friends does, i have the stats of one of his recent matches, he sent me the screenshot because he was complaining about matchmaking
He went :39-6
teammate 8-13
teammate 2-14
teammate 1-14
enemy team:
6-17
14-13
9-10
18-11
final score: 50-47, my friend won
now do you see why many people think this particular matchmaking with sbmm seems broken?
the game literally expects you to be an mlg pro and get 39 kills and you barely win!
My friends was playing at 100% in that match, i was in discord with him
His “average skill” is definitely lower, he cannot get 39 kiss everytime, this match was an exception where he played well. If he didn’t play so well he would have surely lost ( that’s what we call “forced losses” or “almost forced losses” : You lose unless you are able to somehow carry the entire team by a lot)
And don’t get me wrong.
I’m all for choice. Have a slider where you can choose your own preference of SBMM tightness vs connection priorities.
Just don’t go back to the “good old days” or random MM.
i don’t think most people are asking for random matchmaking but when you get 39 kills and you win by JUST 3 KILLS, then you know that there’s a huge problem
It can’t over estimate everyone’s skill. Everyone would still be ranked appropriately.
Early in you CSR journey your MMR can be quite wide. And there is still a 50% chance your actual skill is below the mean of the curve.
And then there are weightings for squads. That can throw out the “on paper” balance.
Besides. The game was 50-47. So in this case match making got it right.
As for why it had such a range of performances. That was either the MM service getting desperate to give some of the players a game - or someone was running around in a squad with mates who were way below their rank.
Not really. It was a bit of an outsider of a game. And besides, this is what people are actually asking for when they want SBMM to be widened. Matched teams across a wider range of individual ranks.
Your skill is nothing more than your performance against other players. I’ve gone 40+ before in games vs lower ranked teams (and lost). It doesn’t make me MLG. It doesn’t even make me Onyx material. It just means that was what I had to contribute to the team in that game.
That logic carries to everyone in the team. The guy who was holding on for dear life and went 1-14 just as easily saved the game by not going 1-17.
And this is how it is always going to be in games with a wide range of skills.
The good players have to carry. The bad players have to survive.
The key is to balance the two teams for an exciting result.
The alternative is the good teams coast to an easy win (chill) and the bad teams have a miserable time (not so chill).
Higher ranked players have worked hard to get good at the game, there is nothing entitled about that.
No. Good on them. Their reward is to play each other and keep improving.
Lower ranked players may have worked just as hard. Maybe even harder. And don’t deserve any less.
I know I’ve worked very hard to improve. I’m not Onyx material of course. But all I ask for in return is a chance to compete.
What you said here can be true, but from your point of view there is another huge problem. This kind of “performance-based” sbmm can be easily “exploited” but at the same time it cannot be considered an exploit in any way, let me explain:
Now what do my friends do to avoid these kind of games as much as possible? when they see that they are playing a match with a high chance to lose , instead of trying to get 30-40 kills and trying to carry, they start to camp in corners with AR and shotgun/swords for the rest of the match, getting a lot less kills in the process.
It’s not griefing or trolling. they are still playing the game, not going afk or betraying teammates, they just know that getting a lot of kills is “bad” because it makes your MMR climb too high for their liking. So they start getting less kills on purpose, they still play the game but there isn’t a way to force them to play in a certain way.
They are most certainly going to lose that match anyways, so what’s the point in raising you mmr? raising you mmr in this case would only make the situation worse because then with a higher mmr, you can lose even more often against better opponents…
If the score is close or they’re winning, they play normally
Many players who complain about matchmaking know this already, you just have to read the trueskill 2 paper to know how the system works, it’s only going to get worse.
Also, let me say that other games’ sbmm is implemented a lot better than halo, i play Black Ops Cold War and Overwatch and most games are close and enjoyable, i still don’t like sweating constantly but i’m adapting to it. Halo’s sbmm is easily the worst
Nope.
For a start, you are being judged on the relative skills of your opponents. Going high K’s on a team, or bunch of players, ranked lower than you is not going to push your MMR up much at all.
If you are doing it consistently, and winning more than you are losing (against higher ranked teams), that is different. You obviously deserve to rank up in that case.
But the occasional day out, or day off, won’t affect a mature MMR very much at all. The curve narrows and becomes resistant to moving. It doesn’t lock down like Halo 3 used to. But it’s enough that you need quite a few games to prove to the system you should be changed.
The weighting for kills and deaths isn’t that strong.
Did we read different papers?
Your MMR mainly moves on the W/L and the relative strength of your opponents.
I’m not sure where this idea that kills are all that matters comes from? I guess you get those sorts of games in placement - when your MMR curve is wide and volatile - and people assume that behaviour carries on?
Can’t vouch for other games.
Personally I find it works just fine. I play very differently in Social than I do in Ranked. And I get close games in both - both pitched at very different MMRs.
The problem only comes when people play Social the same way they do in Ranked. The same tactics. The same map control. The same weapons. So they end up with similar MMR’s in both modes and then wonder why both games feels the same.
Josh Menke literally said that MMR can fluctuate a lot more than CSR. He also said that MMR changes a lot based on your performance, while CSR only on wins&losses because large and rapid fluctuations in CSR would “confuse” players.
i don’t have “proof” of this because i didn’t save his post but i remember very well. He said that changes in MMR even in a day can be “significant”, even some hundreds points. He used to show the graphs of some players’ MMR and they were fluctuating a lot in some cases. I doubt the large fluctuations were only because of wins and losses because in halo 5, you couldn’t lose or gain more than 30csr in a single match. ( this was in halo 5, but halo 5 uses trueskill 2.0 for sure)
And why would they randomly change the system when it was working as they intended in halo5, a very similar game to infinite? Unfortunately Menke doesn’t work at 343 anymore, so we won’t have an answer
let’s take an example of this previous game i told you about.
So my friend got 39 kills, you could say (but we don’t know exactly) that the sbmm expected him to get 35-40 kills in order to “balance” the match.
If he got let’s say, 10 kills, because of his new tactic of camping in corners and getting less kills whenever his team was losing badly, his team would have been completely destroyed.
At this point the mmr should “think”:
– " oh this guy was supposed to get 40 kills but he couldn’t even get a third of them, let’s lower his MMR by a moderate amount ( not too much because everyone has bad matches) and see what he manages to do in the next match" –
the point i want to make is, if you don’t get as many kills as the system wants, against lower skilled players in our case, your mmr should lower.
It makes 100% sense to have large changes in MMR based on performance.
Yesterday i had a bad day on black ops because i didn’t sleep well the night before. Today i expect to win more matches than i lose, if i play as usual.
Should someone lose (or win) a large amount of matches just because the MMR can’t keep up with his performance? it wouldn’t make sense at all…
I think the sbmm works better in COD or overwatch because they also have a lot more players than halo, but i can’t be sure…
And this is mainly relevant early in the season (placement and soon after).
Your MMR curve stabilises as it narrows and tends to resist moving. It takes 50 games or more though.
Your CSR is more consistent in it’s behaviour. It starts off on the left hand side of your MMR curve and then chases it - but only to a maximum of 15 points. That’s why early season wins gets you lots of points but losses are rarely punished.
And the MMR can be VERY volatile in the first handful of games. That’s why they don’t even give you a CSR for the first ten games.
I think this was more along the lines of your personal form. It can vary by +/- 150 points depending on the day.
But if he was talking about your MMR I suspect he was still talking about how it behaves early in your ranking journey. A mature MMR should resist a lot of movement unless you put in a few good (or bad) days in a row.
You can see this when a mature MMR oscillates around a point. Wins against evenly matched teams only gives you 1 to 3 CSR points - which build up but are then quickly taken off with your next loss (invariably when someone quits out on you). This infuriates people… but it’s just that your MMR has become relatively locked to your skill level and it’s going to take a lot of good wins (or bad losses) to change.
How we could do with a Menke type on the forums now…
It would be unlikely they do much with TrueSkill2 itself. But they are always tweaking with the CSR behaviour. They even changed things up a couple of time in Halo 5.
Not really. The ranking system pretty much only needs the W/L. The weighting for personal performance adds in a bit of extra accuracy - but I think it mainly works at getting to your rank faster, not so much higher.
To put it in context, have a look at the prediction tables in the TS2 discussion paper. A kill rate of 1.4 is the 50:50 mark. Doubling your kill rate to 2.8 only increases your win expectation by 4%!
Pushing you kill rate out to 4 per minute (wow) is still only 60:40.
I don’t think it’s a huge weighting.
I do not view playing an absolute sweat fest against yourself as being a reward for your time and effort. I view it as the opposite.
Further, because there literally are not enough players above onyx to get regular games in social, it just pulls in borderline onyx and diamond players in to games with people rated onyx 1700 or Ive even got a couple of onyx 1900s in my matches because I am the closest to their rank in the search pool. That is just a straight up punishment.
And this is where SBMM tends to fray a bit - at the edges of the population curve. That, and when friends of wildly different rank jump in together.
I don’t think there is an answer. At least not with conventional game types.
And definitely not for the competitive type of player that feels compelled to go 100% if the game is close and/or tough. Your social MMR will only drop if you play in a consistently relaxed manner in comparison to your ranked persona.
I think 343 need to experiment more with multi-team. Or introduce more players onto the teams (eg. 5v5). Something to break up traditional ranked tactics and game play. Make it more chaotic. Or some sort of handicapping system that allows Bronze and Onyx players (exaggerating) to meet on the same battle field.
I can’t wait for Forge and scripting. Our first script is going to be a handicapped game where every kill is that player’s K/D times 10.
And if you respawn with a K/D less than 1 you get an OS for (1-K/D)*60 seconds.
And we will tweak these numbers until we start getting close, competitive, and fun custom matches (players ranging from Silver to Onyx).
If we need to we’ll add in damage reductions for killing streaks (10% for each kill over 2) and/or limits on leading players from picking up power weapons and equipment.
The idea is for everyone to have fun. And the game will reward (or penalise) you for the effort you’re putting in on the day.
What is the point of allowing huge changes in MMR ONLY at the start of a season? If you want sbmm to succeed, ideally, you’d want to keep matches balanced at all times… MMR fluctuations could lead to unbalanced matches pretty often at every start of the season
I don’t think it’s fair for people who have a bad day to “suffer” through unwinnable matches JUST because you played well for the first 50-100 games of the season, until you lose enough to finally be placed where you belong for the rest of the day. Then after the bad day is over you are going to climb again of course. But this thinking shows another flaw in the system
Smurfs could play badly for 50 matches or so, then start playing better and stomp on many noobs before the MMR decides that it’s time to go up. Menke said that the system was banning played who did this, but someone smart could do it slow enough to the point the automated system doesn’t notice it, stomping many lower skilled players in the process
I don’t think this system will work at all, (the damage reduction system is the worst offender, 10% penalty is way too high) i wonder who will agree to play on your custom games. But let me know how it goes once you do the scripting.
Even prop players in tournaments can get 10 kill streaks, it’s rare but it happens. And they have basically the same skill…
The bigger the flux the faster the system can “ball park” your rank.
It’s essentially why systems such as TS2 can find your rank so quickly. As opposed to simple W/L sorting systems.
I don’t think they open it up quite as much with a new season as they do for a brand new account.
But by making the curve volatile they give you a chance (at least over 10 games) to prove you deserve to rank up (or down). By the end of the season you probably have a few hundred games under your belt and your MMR curve is relatively locked.
Once settled peoples ranks really don’t change. They may fluctuate with form (eg. the +/- 150). But they don’t change. It’s really hard to rank up above your skill ceiling. You have to practice. A lot.
So once the system has established your rank - it doesn’t need to worry about micro-fluctuations.
There are certainly weightings in your MMR for “form”. Probably as part of your global part of your MMR curve. I think the main aim for this is to gently push your opponents in the direction to help you rank up or down - but it would also help to even out your games overall.
Once you’ve played 50-100 games the system knows your rank. And you know your rank. It’s not going to change that much.
From that “base” I don’t think the system is ever going to feed you unwinnable matches. Yes, you might be off. But your team-mates and opponents are having similarly good / bad spreads. It should all wash out.
I think you’ve answered the question here. It does take that many games to smurf a mature account. So it just goes to show that a few good or bad games isn’t going to upset your apple cart.
Smurfs on new accounts are a different kettle of fish though.
In H5 we had a script that calculated how far you were from the lead and applied different trait profiles. It worked well. But unfortunately you could only get a scripting reference on who died. So acting on the killer is new to us.
It will be trial and error - but we thought hitting 40%-50% reduction in damage by the time you hit a killing streak would work well. The other thought was to pop your shields permanently on the third kill.
We have a weekly Halo group. So I have a captive audience of “willing” participants.
Yep. But, but in our case it’s going to be the Diamond / Onyx players going on a tear. So it’s OK to have a consistent penalty to slow them down and make it easier for the lower ranked players to fight back.
At the end of each night we used to play Shotguns on Beaver Creek. First to 50. Obviously the good players rocked. So I Forged a bunch of OS into the middle and only the designated “sponges” could use them. What we found is the good players still won - but the bad players still had a good time. And they actually got better at strafing and aiming because they weren’t dying 5 seconds after spawning.
this is my halo infinte experience, but I get somewhere between 30-20 kills and then lose…
A lot of people have been bringing up with issue, but 343i simply do not care. I stopped playing this game a while ago because of this god awful inbalance, alongside a number of other glaring issues which they refuse to acknowledge or act on. I would honestly just drop this game completely at this stage. It’s too far gone and 343 aren’t changing their ways.
I dont think you get it. If you have sbmm, youre incentivizing players to SUCK. If you suck, you can get a lower rank, and play poopier opponents.