I think this right here is what creates the matches you would feel like, are the gatekeeper matches. It cant pull in the right onyx players to get you into onyx since they hopped off, so it pulls the higher onyx players for you to play against…meanwhile spinning the slot machine for your team mates.
Its tough cause a rank decay system would alleviate this in terms of deranking players accounts that sit idle for too long (to where if you reach 1500 and stop playing for over a week or so, the csr number starts going down and whatever hidden stuff has to happen would happen as well) so…that would encourage players to keep playing but at the same time probably piss them off too.
Again. It’s a stat calculated from your past KPM vs that level of opponent.
The value will change on both the KPM and the time of the match.
They are not going to be the same in every game.
You seriously need to put some time and effort into these concepts. For many reasons. But mainly because if you don’t get your head around them you aren’t going to rank up.
Even though I think the stat is full of compromises; it only appears sporadically, it looks at CSR not MMR, and doesn’t take into account squad weightings etc.
I’ll bite.
Let’s look at YOUR last 10 or so games where the stat is available.
The favourites are 60% (W), 61% (L), 53% (W), 57% (L), 51% (L), 55% (L), 56% (L), 65% (W), 50% (?), and 67% (W).
I agree it’s a slightly wider spread than I would like. But given the population woes it looks pretty good. I think most people would be happy with that spread of games.
There are really only two outliers (the 65% and 67%) - but even those aren’t horrendous.
As for the results? The “underdog” won half of those games!
So much for you compelling evidence that the stat is proof that the games are rigged.
More than I care to say. But that’s not a prerequisite for putting a bit of effort into understanding what is going on.
I rank up just fine by playing the game at a high enough level to. Its hardly my fault that the matchmaker and the poor optimization of the game in general finds ways to screw people over.
It is rigged. Its just that sometimes the rigging doesn’t outweigh the game being a buggy dumpster fire.
Definitely weird. Keep an eye out to see if it happens again.
No the post seemed to describe a different MMR for each game mode within the playlist. So even now there would be a separate MMR for Oddball, KOH, Strong Holds, Slayer, etc. And then there was one average MMR as a whole.
It was the sort of comment that I would definitely have jumped on at the time to get clarification.
I’ll have a look later to see if it’s saved on my laptop.
It could be due to player pop limitations. It’s possible the system may not place enough emphasis on players within teams being around the same relative skill level. I highly doubt the MM is going out of it’s way to force these type of player pairings.
I never claimed to have all the answers. The point was there are other potential explanations for most of the claims being made here. Cherry picking anything and everything with even a slim chance of supporting a preconceived notion is… flawed on so many levels. Especially if it’s based off of memory. Instead of something more tangible.
That would definitely make the average move up or down. Thats actually a really good point tbh…but…still an odd choice since we’re all getting ranked in a playlist where the mmr fluctuates depending on how strong or weak players are in the upcoming game type. So someones rank could say daimond 6, but the reality is their best game type they’re about that level, their worst can be much lower…but…i guess that would be the entire point of mmr being a range especiallly in that playlist.
once again, if you actually played ranked instead of just unquestioningly believing everything 343 devs say, you’d realize what’s actually going on
but no, the tps reports say that if you divide the number of bullets that you fire by how many grenades you throw, you can calculate the rate at which the oddball matches will impact the precipitation gauge in the planters on Aquarius. Or something else nonsensical
You have, yourself, more than once in these threads, accidentally proven that the game is rigged to force a 50% loss rate by any means necessary. You’re just not realizing it.
Or does admitting the game is skewing way too many matches into the 60-40 range and attaching low level players to high level teams in order to make them lose not count?
Edit: Looking at your ranked games is proof enough. Every time you win a ranked game, the matchmaker has put an Onyx player on your team against a group of low diamond/high platinums.
Your last two losses are due to essentially the exact opposite happening.
The system is manipulating not only your win/loss but the win/loss of the teams you’re on
Good Lord, your lobbies range from gold 5 to Onyx 1700 within the same match, wtf is going on there
Found the quote. Not 100% sure of the context. And there wasn’t any real follow up questions.
Interesting.
Not sure if it tells us much about the average team MMRs. We don’t know how they are weighted or calculated.
But using different MMR’s for each game mode would certainly explain how/why they drop. But from the way the post is written it’s the average MMR across the playlist that is used for matching - so doesn’t explain what you observed.
I don’t know why you defend their fuzzy math. What about lag compensation, mixing lobbies in ranked with other time zones, and what’s fair about 2 fireteam players obviously playing with low ping on servers close to them while the other 2 fireteam members are from other server regions playing over the cloud to connect to the data center. If you disagree with the last statement try explaining why the connection to the game is always cloudapp.azure.com preceded with the region east us, west us, central us and so on. I have played a lot of multiplayer games and this one is so obvious to see. They think they have created an algorithm that is better than my brain . No algorithm outthinks people unless they are 5. You could create an algorithm today and in 20 years a 20year old has better predicted outcomes. Until they have region locks it can’t be fair. That’s why the rare match where both fireteams have an equivalent k/D is so rare. This shouldn’t be. If it were truly competitive you would see more rare matches with nearly identical K/D. This would also equal consistency in matches.
Let me ask you, what multiplayer game nowadays that features matchmaking isn’t dealing with lag compensation and teams with players from mixed time zones?
Can you name one?
Do you really want a game with matchmaking to restrict people from joining together unless they reside in the same time zone with internet connectivities that are on par with one-another?
I hope you realize that there are multiple systems in place that work together to make matchmaking happen.
There’s the Skill Assessor system (TrueSkill2) which does nothing more than build an understanding of player skill, makes adjustments to those skills when partying up, and then provides predictions based on its understandings of player skill. It’s constantly adjusting its understandings based on match results. To note, there’s not a more accurate predictive skill system publicly published than TrueSkill2. Its accuracy is ~70% which was a major improvement within the industry at the time of its public release…
There’s the Team Balancer system, which does exactly as its name suggests, it places players that the matchmaker has gathered into teams while evenly dispersing the skill balance as best as possible. This does not mean every match created will feature 50:50 odds, but that the goal is to achieve 50:50 odds or as close to it as possible.
There’s the Matchmaker system, in Infinite it’s a machine learning system called TrueMatch, and its job is to pull in players who are actively searching into a created lobby per parameters it establishes. The machine learning aspect gains a predictive understanding of the active population. In other words, it predicts what player skills and connectivity are likely available in relation to the necessary search time. It’ll automatically self-adjust search parameters in order to find the optimal settings per the time of day. Here again, Halo is using cutting edge technology within the industry.
Within the ranked environment you have CSRs which are designed to provide visual representations of your TrueSkill per a playlist, but in a way that’s more relatable. It’s goal is to offer a pretty accurate representation within a fairly short period of time. It’s not really meant to be a progressive grind fest, but at the higher skill levels it does becomes a bit of a grind due to the rank-in cap being set at Diamond 3 or whatever it’s set at nowadays. Players with high skill levels need to grind out noticeably more wins in order to converge their CSR to their MMR.
GOW 1 had a match making lobby. You saw who started a multiplayer lobby and his ping. If it was “White”, very low ping, i joined one of the lobbies. If it was orange or red, i started myself a multiplayer lobby.
A very simple and effective system.
It choses from the available players and combines them in a squad with higher percentage to win or to lose the match. To many time most of my team mates went negative, when we lost. Strangely, when we win most of my team mates are positive.
No, thats not what halowaypoint shows. Its 6 kills expected for the whole match, not for a certain period of time.
When the match start, im not aware of the predicted kills. But i can imagine how it ends when i see the ping on the right upper corner and how reliable it registers my BR bursts.
If the system predicts that im getting only 6 kills, because im playing +130ms ping, in which gamemode are 6 kills helpful? In slayer, koth, strongholds or oddball? In none of them.
When you play in a match with +130ms ping, while other player are on their local server with 20-30ms ping, then you have a hard time.
You know exactly what the problem is. Me with 130ms ping, the opponent has 20 ms ping, i start shoting, lag, desync, a few burts dont register, i have to stay longer engaged in the fight, that means that i become an objective for one of his teammates and get killed. If all 4 burst registered correctly, i would have moved on.
High ping triggers a chain reaction, you stay longer engaged in a fight and get killed more often, because if you try to evade in mid fight, you will die anyway.
Thats one of the match where i end with a k/d of 0.5, unless that i would begin to hide and stop playing the match. Thats not what im doing, never.
I said it more than once, if one team gets stomped, doesnt matter my team or the other team, then something went terrible wrong with the match making system. This 2 teams had never be put in a match againts each other.
Its one thing, if 1 of the player has not the same level / skill as the other 3 player. Thats ok. But a whole other thing, if all 4 player of a team are stomped, going very negative. Coincidence? No way.
supposed to yeah. But the reality is most of the people who are getting ranked are in platinum. So we are seeing platinums matching onyx players consistently.
Now there are many reasons why things aren’t what they are supposed to. Perhaps more significant than population decrease is how active ranked players are and how many are ranked. Right now those numbers are almost certainly low as ranks were reset recently. Also how is this system working for people who will never play ranked. is SBMM accurate for people who only play social?
Lastly, with this recent rank reset, 343 prepared a classified experiment with CSR and deployed it for the rest of this season. The scope can’t be discussed. But there is active work being done to improve upon menke’s build.
Social MM for SBMM is way looser. I won every doubles match probably too easily. I hop into ranked and im immediately plagued by
ping over 140
unstable packet loss
unstable jitter
Man the matches end up being more difficult. But for all the wrong reasons. I’m doing no damage at times, teleporting at times, rubber banding…in Social though?..SMOOTH sailing as far as network stability went. Population for ranked must be really low.
I don’t believe that anything related to Damage Per Minute (DPM) was taken into account. Accuracy and damage is part of the post game stats, but I’m unsure if either play a role in your CSR or SBMM.