Forum Discussion
EOMM vs. SBMM - A Case Study by UCLA with the Assistance of Electronic Arts
Every heard of EOMM? It stands for engagement optimized matchmaking, which is different than skill based matchmaking.
This study by UCLA, with the assistance of data provided by four employees of Electronic Arts, the publisher for Respawn Entertainment and Apex Legends, took data from over 36.9 million matches from 1.68 million unique players.
It discusses the EOMM model using variable inputs such as churn rate to optimize player retention, and discusses similarities to SBMM.
In my own conclusion after reading through this, it appears that we actually live in a state of EOMM more than SBMM, but one that employs known SBMM logic and systems, which would explain why you have three excellent games in a row, which appear to be SBMM-related, only to get steamrolled on your fourth, and so on.
Before you comment and rail Respawn, Electronic Arts, or anyone in particular, I suggest you read the length of the paper (8 pages including citations, not a long read at all). I'm posting this here for discussion purposes, not to ignite a fire in the community.
If presented as EOMM versus SBMM, after reading this paper, would you feel better or worse playing under the parameters of this matchmaking algorithm?
Here is the frame work - https://core.ac.uk/download/pdf/326834229.pdf
This one will probably make you feel unpleasant
https://www.pcgamesn.com/ea-matchmaking-microtransactions-eomm-engagement-patent
@ComicBookGuy2727It doesn’t surprise me nor makes me unpleasant it’s expected… I always knew theese systems aren’t for fair play but a way for devs to manipulate their playerbase…
But it doesn’t suit and fit the image that ea try’s to be, so much talk about human rights dignity fair play and all kind of other great words…yet they manipulate their playerbase and abuse human emotion for their gain.
it also doesn’t matter if they use sbmm or what ever, it’s the same result for the players. Sbmm is not fair and it never will be and it even can’t be. The only fair thing would be a truely 100% random matchmaking
- Midnight97462 years agoHero+
I'm only saying it looks like EOMM and not SBMM, because I'd get two "noobs" as teammates for several games in a row, (and by noobs, I mean they play like bots), and then after losing 10-20 times, I get good teammates where suddenly the enemy teams are playing just like how my randoms played like those 10-20 games prior, like bots.
That's EOMM.
It was trying to get me to think I gotta do better for 10-20 games in a row and then once I got better teammates and ended up with noob/bot level enemies, it was suppose to make me feel like I got better and to thus keep playing since I was on a killing spree and/or winning high.
That might work for other people, however I'm quite aware that I'm in ALGs pred level lobbies then get put into a "bot lobby" before being thrown back into the ALGs pred lobby just for winning a match, so its just not fun.- 2 years ago
I think you are right in your assessment.
Please try the following experiment: leave the dropship about five times in a row and return to the lobby. And then observe your opponents in the following round (which you then actually play). What level they have and whether they are actually weaker. See here: Apex Legends S18: The Secret to Getting into Easy Lobbies Revealed!
Of course, this is only meant to gain knowledge and to find out if the MM is practically only based on the survived time of a player.
If the MM can be influenced by this, then that's pretty weak of the developers.
- Midnight97462 years agoHero+
@saurerDropsXXL I've already did this back in S16 when me and a friend had "play 10 times as this legend" challenges. We qued up into duos and kept throwing ourselves off the map and on the 11th game, we played as our mains for real and all the enemies were low leveled accounts running into walls, randomly stopping midfight to stand there, AFK, etc.
In short, considering that matchmaking has gotten worse since S16, its safe to assume this system has gotten worse as it puts more emphasis on EOMM.
The hidden MMR had one goal, just one, which was to give weight to a player's skill to then accurately adjust them in the matchmaking to ensure even and fair fights. Instead, its being used to see which teammates a player should get and who to go up against to insure a predetermined win or loss.
This is observable by the fact that we get teammates who can't even do 100-300 damage, they die first in any fight, and then after losing matches too many times due to teammates, we get teammates who suddenly know what they're doing and in most cases, they do more damage than us, and we end up winning or getting closer to winning games.
Edited 1: I understand that ranks no longer "represent skills", however why did I get a Rookie I player at all? They had no gun skills or game awareness and was mainly looting when a fight was starting to occur. The Valkyrie wasn't too great either, but they tried at least. It was harder for Valkyrie and me to really do anything when the team we died to had a Revenant using aimbot, (he beamed all 3 of us while their two teammates were running around like how the Bangalore random was).
The next game gave me a Gold IV and Gold III teammates, they were better then the last two teammates. While the first team we ran into had a Mirage acting like a new player, an Octane pushing too far out from the team, and a Seer unaware of their surroundings. This match was lost on our 3rd fight, as there was a pred player who gunned us all down.
Edited 2: The next game after the two golds, I had actual decent players. Enemy teams were still a challenge to fight, although some were just bad and stood there while firing their guns, however we won that match, then the next game, I died to a level 51 aimbotter.
How is this even SBMM?
- 2 years ago
Well, even if I agree that the MM isnt designed to put 60 players of equal skill in terms of shooting, moving, reading the map etc together, which most of us would define as a "good" SBMM, I STRONGLY disagree with your assessment of the MM. Your fundamental mistake in your assessment is that you only base it on YOUR experience, but totally forget that the same rules must apply for EVERY OTHER PLAYER OF ALL SKILLS in the matchmaking queue. This makes your theory logically impossible.
How can a MM system logically be designed to put every player pressing FIND MATCH, from the noobiest noob to ALGS pros, with two significantly worse players against 57 significantly better opponents for "10-20" matches, and then put them in the next match in a lobby with 57 significantly worse opponents with two significantly better teammates? I dont know if you studies any type of programming or just understand mathematical logic but you must realize that this would be IMPOSSIBLE unless you have an unlimited active player pool of players at all skill levels. You dont appear like an idiot so just give it some thought and I am sure you will realize that this cant be the case, from a mathematical POV.
And when you say "thats EOMM", then I am afraid you are just plain wrong. We must just acknowledge that EOMM refers to a type of game MM algorithm, derived and patented by EA. Its not a general broad term for "a matchmaking that makes you lose by design and then lets you win". But anyhow; this is the last time I will say anything on the subject. If people wanna keep refering to EOMM then I couldnt care less. But then dont complain that Respawn doesnt care what is being said here. Why would they listen to blabbering fools?
- Midnight97462 years agoHero+
Good points you have there, however looking at things from your point of view it makes me think more about bot accounts.
A lot of players report seeing Lifelines and Bangalores all landing in the same spot and not moving. That's because there's players who run multiple instances of the game just to level up the accounts to then use them to cheat in rank.
Even if we ignore the bots put in by cheaters, Respawn has their own bots in the game. These bots are supposed to be in the orientation matches, but I wouldn't be surprised if they're placed in normal matches.
Although a lot of bot looking players have badges and stats, some of which that I've seen were over level 500, however its not hard to equip fake stats onto a bot account and then throw that bot into a match. Which bots in games, (such as NPCs in Skyrim), are good enough to follow players, pick things up, etc. but the bots/NPCs lack things such as common sense.
In short, (from what I'm understanding and seeing), everyone has good and bad games due to bots being present in the match as well as the hidden MMR and it being EOMM based.
There's definitely more things to consider, look into, and test.
About Apex Legends Feedback
Community Highlights
Recent Discussions
- 9 hours ago
- 12 hours ago
- 13 hours ago
- 2 days ago
- 2 days ago