"Well for one, ranked percentages are a bad measurement as even preds can still be in bronze rank and many diamond-pred skilled players don’t even touch ranked."
Yeah I mentioned that in the OP but like I said; its kinda the only estimation of the player skill distribution we have, right? I mean, it doenst take into consideration that it it can be miles of skill difference between a Diamond 4 and a Diamond 1, or that the player count distribution in especially Platinum and Diamond probably is heavily skewed to the lower tier (many in D4, few in D1 etc). I just used those numbers to give a rough model how the general skill distribution may look like, just for the sake of the argument..
"’d say connection based is the superior way to do it. Completely coincidental and on average you would have a 50% chance to run into an average player, 30% chance to run into a below average player or above average player, 10% chance to run into a new/bad player or high skilled player and so on. "
If you mean that the SBMM should just pick the last 60 players that pressed "play" and put them in a lobby, in random teams, then I could not agree more. Thats how I see causal gaming, especially in a BR. It would result in the shortest queue time for everyone and that is the ONLY model that would make K/D a real measurement of your skill. Because it doesn't take a genius to understand that your K/D depends on the skill level of your opponent in comparison to your own skills. So only if the lobbies followed the overall skill distribution of the player base would a K/D stat say something meaningful of your skills, i.e where you are skill wise among all players.
The only problem that exist now and still would, also with a "connection based" matchmaking, is ofc how to handle random solos vs premade squads, especially premade full predator trios. Hrmmm
As one of the many complainers, I will give you some possible options. Keep in mind that every option has flaws just like the current system, but hopefully would deliver an improved gaming experience for all.
1: Introduce an improved “score” that each player gets based on ALL their games, both ranked and pubs. Yes, I know this concept already exists but the current score is all over the place, resulting in the bottom 1% fighting against the top 1%. This new score could be based on the current RP in ranked but with some adjustments such as adding amount of knockdowns, gun accuracy, owned badges, etc.
For the sake of simplicity, I will use a scale of 1 to 10 and link them to current ranks so you have an idea of what kind of sbmm lobbies I am trying to create:
Bronze: 1
Silver: 2
Lower gold: 3
Upper gold: 4
Lower plat: 5
Upper plat: 6
Lower diamond: 7
Upper diamond: 8
Masters: 9
Predator: 10
The following list shows which skill-pools you can get in your lobbies:
1 fights between 1 and 3
2 fights between 1 and 4
3 fights between 1 and 5
4 fights between 2 and 7
5 fights between 3 and 10
6 fights between 4 and 10
7 fights between 5 and 10
8 fights between 5 and 10
9 fights between 5 and 10
10 fights between 5 and 10
This system protects the less skilled players, gives the above average players a healthy mix of lobbies and gives the top tier players a skill pool which is wide enough to ensure low matchmaking times and delivers a mix of easier and harder fights.
2: A hybrid between an improved SBMM system and no SBMM at all, where you get a defined percentage of SBMM games and a defined percentage of normal matchmaking games. This is not my favorite option and I feel like this would only work IF they finally step up against smurfing. Also, penalties should probably be implemented to minimize abuse of the system, such as leaving immediately or jumping in the water/lava off spawn.
3: The most simple system could be kill death ratio based. You could add account level and total amount of kills to the equation, but maybe you don’t even need to. Let the people fight against others in the same k/d range, and the higher your k/d is, the wider that range gets (to combat long matchmaking times). This automatically eliminates the smurfing problem because those accounts have the highest k/d’s out there. The main issue with this system is that it’s essentially another ranked experience, but I guess something skill-based will always have a ranked feel to it.
Apologies for the long post. Give yourself a cookie.
I think I understand your general idea of that kind of SBMM. Its an interesting idea to have a "dynamic" SBMM based on your own skill, but I think you miss my main "concern" with all these SBMM complains. I think many people forget how normal distributed the skill distribution most likely with the huge majority of players around the mean of probably a skill level correlating to gold.
So a SBMM with an algorithm based on your suggestion would be great if there were pretty much equal amount of players along the entire scale of skill, but it isnt. So I think your suggestion basically would lead to one super sweaty high elo lobby, one mid lobby where most players would be, and one low lobby for new players and super noobs.... And isn't that what the current SBMM is now?
Hheh how would i do it? No clue nothing that could be done will make everyone happy. There will still be those who gripe they cant kill someone or cant win a match because its broke. It is impossible to please everyone. As the stats show with the current system in place ranked has returned to season 1-2 top percentage of who earned to be there. There is no denying this 100% reset pred lobbies/master/diamonds exactly where they was many seasons ago as it should of been.
This is not me saying there are not flaws but numbers dont lie when comes to the outcome of the system used.
Community Manager
