Got a *****in bone to pick with the devs - RANKED

(OUTLAW11) #1

Okay the title was a bit rude, but did it get your attention?

CONSIDER the solo-queuing gamer. Since achieving Gold Officer I pair up with the *****tiest teams you can scrounge up. I’ve actually de-ranked to gold specialist since hitting 99.6% in gold officer en route to gold master. Your rankings system doesn’t work well for the many folks like me. My high gold rank gets paired with unranked players and bronze ALL THE TIME. This is supposed to average out the team to a certain rank that matches the opponent. Well, your algorithm sucks! My teams typically face a more stout team that are averaging more players in silver. While I understand most silvers aren’t much to worry about, it is a huge effort for a bunch of new players who have no time in ranked or are terrible players.

Take for example the match in the attached image. It is HIGHLY unfair to players like me. Essentially using a single strong rank to bring the team “rank” up for a match where we get slaughtered. I removed identifying factors for the other players.

A suggestion: Create a server pool of UNRANKED and BRONZE players that does not cross over to silver and above. This should at least address some of the mismatch issues in ranked.

(Apples) #2

I guess the problem DB have is a very very little playerpool according to steam charts, games with 25K + players everyday are already struggling with Matchmaking, so DB with its lower playerpool might have a really hard time to match you up with decent teammates and enemies, unless you want 45° minutes Q…

(ASD) #3

the last time I get downranked was after I played 5:1 … I was the one… all others disconnected
from my point there need to be things done before

[ul]
[li]reduce waittime[/li][li]realy, REALY, REALY heavy penalties for leavers[/li][/ul]
there is no order for this two… both have top prio

(Nail) #4

I’m afraid penalizing people doesn’t stop them from playing stupid, just stops them from playing, imho penalties won’t be viable until there are community servers where people are committed to actual teamwork, seems backfill is a bigger problem than leavers

(OUTLAW11) #5

I’d start with splitting up the bronze/unranked and silver/plus into separate server pools. Just to see what happened.

(Nail) #6

you’ll get bronzes crying they can’t play with their silver teammates and reddit will explode in a mountain of salt

(OUTLAW11) #7

I’m okay with salt. It’s ranked. It’s not solely about playing with pals in ranked, is it?

(OUTLAW11) #8

So another example. The flaw in the ranking system is the statistical assumption between ranks. In the attached image you will see our rankings for this match. Understanding the flow of ranks are as follows (B is bronze, S silver and G gold):

B1 B2 B3 B4 B5 S1 S2 S3 S4 S5 G1 G2 G3 G4 G5

Consider this:

My team consisted of B1, B3, B4, S5 and G3 which collectively = “silver recruit”
The opponent team was B1, S1, S4, S5, G1 which collectively = “silver agent”

SplashD assumes that having a G3 will boost the team level up to a competitive level for the match. This assumes that the G3 will cover the shortcomings of the B-level members—this is hopeless in a 5v5 match. The reason why this is hopeless is because SplashD incorrectly assumes that the magnitude between ranks is equal, but it is not! The skill difference between a B1 and a B2 is drastically different than a S5 and G1. The difference between bronzes is negligible, but between silvers and golds it can be a huge difference. As rank increases, so does the skill level between ranks.

SplashD is assuming these ranks are ratio in nature when in fact they are interval values of unequal magnitude between ranks.

(Mustang) #9

You are correct skill/Elo doesn’t doesn’t scale perfectly with ranks, that’s why it looks more “unfair” than the numbers suggest, and the numbers are the only thing we have to represent “skill”, so yeah.

Using your example with some made-up Elo numbers:

The blue shows the combination of players you had, the yellow shows what might have looked more fair considering ranks alone.

With the made-up numbers the difference in total Elo for the blue game is 50, whereas the difference for the yellow game is 150.
So on paper the blue game would have been the best case scenario.

Not trying to say the system is anywhere near perfect, just that ranks don’t show the whole picture, and it is working as intended.

In previous seasons high ranking players had problems of waiting eternally for games, because there were not enough people of nearby ranks playing at the same time, this is why it allows B1 to be in the same game as G3.

As someone already mentioned there isn’t anything that can be done to “fix” this as the “skill range” to “wait time” trade-off scales with “queuing player count”.

Your suggestion to split the pool is effectively the same solution as reducing the “skill range” which negatively impacts “wait times” to a point where they aren’t feasible with the current “queuing player count”.

Of course if you want to artificially reduce the “skill range” you could just cancel and re-queue every 30 seconds or so, but I think you’ll find you’ll be waiting forever to get a game for the aforementioned reason.

(OUTLAW11) #10

@mustang Thanks for commenting. Truly. Doesn’t seem enough constructive comments are being offered. The wait times would be horrible in the current system, but…

If 10 random players join a server like they do in pubs (silver/plus only), these players can be allocated to a team. The algorithm for piecing teams together can be applied much like it is in the pub server–perhaps a more strict algorithm. This would eliminate the long wait times. The players would already be on the server. All they need is to be placed evenly for a match.