View Single Post

Linuxizer's Avatar

01.03.2020 , 10:06 PM | #23
Quote: Originally Posted by DakhathKilrathi View Post
Win/loss is objectively better. Even in the circumstances you're outlining, that is a self-correcting problem. It doesn't give anyone an advantage for cheesing.

Meanwhile I've got a ship where I have a KDR of 63. That's not an exaggeration. I played it very conservatively. A more realistic look at my KDR puts it closer to 3.5 - 4 these days. Additionally, in domination I (and many of the best players) am(are) far more likely to just eat a death if it saves time. If that's the right call, it increases our chances of winning.
Kill rate (per match or per unit time) is surely a more useful metric than KDR. I suspect a lot of less good players than you have better KDR than your 3.5 - 4. Bomber aces will be misrepresented with either kill rate or KDR. Win rate is more representative for them.

I have an example of a deathmatch where one player had 24 kills, 2 assists, 10 deaths, 83000 damage, and another on the same team had 9 kills, 6 assists, 3 deaths, 54000 damage. KDR really isn't the best metric here.

Even if everybody solo-queues, I wonder if win rate is going to be a poor metric in the typical match.
- The ace pilot solo-carries and may have a win rate of 90% or more, and large sample size from thousands of matches
- The well above average pilot may have a win rate of 70%
- The slightly above average pilot may have a win rate of 60%
- The average pilot should have a win rate of 50%
These are useful and representative metrics.

On the other hand,
- The brand-new pilot has a win rate of undefined (or 0% in swtor-math)
- The 1-match pilot may have a win rate like 45%, and large sample size from many such players
- The well below average pilot may have a win rate like 46%
- The slightly below average pilot may have a win rate like 48%

I picked numbers like that, because in 8vs8 or 12vs12, an above average pilot makes a much bigger positive difference than a below average pilot makes a negative difference. So the problem is that most of the players in the typical match have the same win rate, within statistical noise. Kill rate should provide better resolution. Perhaps experience, kill rate and win rate should all be accounted at the same time, for best matchmaking results.
Starfighter videos on youtube

Referral link: