Vell0cet wrote:Several points here:
1. We need an optional 2nd tier academy mode for players under around 5 mill SP. This is critical, but I consider this separate from the matchmaking topic.
2. We don't need tiers by gear or SP (other than point 1) this fractures the player base and will result in crazy wait times, or would require removing game modes to make work.
3. Because the number of players is low, the most important aspect to focus on is making team A and team B have relatively equal power. Too often I see two full squads from good corps on one side, and randoms on the other. I always wonder how much differently the match would have gone if those squads were fighting against each other instead of being on the same team. This won't fix all of the problems, but getting this part right should help a lot and at least make matches less one-sided.
4. The big picture solution is to give each player a "Matchmaking Quotient." This is a number between 0.0 and 1.0 designed to represent how talented you are as a player. The number is invisible to us, but CCP's servers use it to make good matches. It factors in SP, average WPs per match, weighted K/D (the better the player is that you kill, the more that kill is weighted), ISK efficiency, average rank at the end of a match, win/loss, etc. There should be a strong bias for average rank (in an ideal matchmaking system most players should be in the top half 50% of the time and the bottom half the other 50% of the time--that means it's working).
There can also be penalties built in for pub stomping. If your final WP tally exceeds the median WP from your team by some percent (say 400%) then you will get a boost to your matchmaking quotient (forcing you to play harder opponents in the future). Likewise, if you do poorly your Quotient will drop and you'll be matched with easier players. Eventually you will hit an equilibrium and your quotient should be a reasonably good barometer for how good you are at DUST. If we had a stat like this, coding the matchmaker itself should be pretty easy, as there are widely published algorithms for this sort of thing.
I do want to stress though that focusing on getting team A and team B to be as even as possible should be the first priority, along with getting an academy mode in place to help NPE.