ET Server Suggestion Change KR (kill ratio) to KDR (kill/death ratio)

[Mufasa]

Currently, =F|A= servers with silEnT mod use KR (kill ratio) to track individual performance and sort teams fairly. Each time a player is killed or gets killed, KR is recalculated with the following factors:

1. The original KR of the player who made the kill and the player who was killed
2. The distance between the player who made the kill and the player who was killed
3. The average distance between players on the map

The first factor is easy to understand: if you're a high-KR player and you kill a low-KR player, your KR is increased less than if you had killed a similarly-rated player. Likewise, if you're a low-KR player who kills a higher-rated KR player, your KR will increase more than if you killed a low-KR player.

 

However, the inclusion of factors 2 and 3 doesn't make much sense to me: in general, is stabbing one player with a knife really more useful than killing two players with MP-40 who are further away? Is a one-man panzerfaust kill directly next to someone really contributing more to a team than a 3-man artillery on the other side of the map? I don't understand it in theory, and it becomes even worse in practice. Players who primarily play covert ops and get knife kills have highly-inflated KR values, causing shuffles to be highly inaccurate. Certain play-styles are rated higher even if they result in the same or fewer kills. Kill ratio is a good idea in theory, but the chosen factors simply don't work in practice. KDR is also imperfect, but I think it's intuitively better than KR.

 

TL;DR: KDR, rather than KR, should be used to shuffle teams.

Edited February 4, 2018 by TURNTLE

[Cumquat]

If it really is that way then i agree with you.

Especially since knives are bugged in silent and noobs take advantage of it and kill you with knife from 25 meters away.

[Chuckun]

Yeah I don't care at all about KD but nonetheless in the interests of shuffle, I have to agree..

 

if anything though, I think it should also consider how much of the enemys HP you actually took when killing them.. if you get lucky and 1-hit someone after another player took away 90% of their HP, then that should qualify for less +KDR than if you killed them on your own, or indeed by killing them by taking away 90% of their HP after someone else took away 10% already

 

Ps - I'm not saying the other player who dealt damage to the enemy should get any KD increase when they didn't kill them, just that if you got lucky by taking away their last remaining HP then you shouldn't get the same KD boost that you would get if you took away all or most of their HP to kill them

Edited February 4, 2018 by Chuckun

[Cumquat]

It works this way for xp,i dont know about kd.

[captnconcrete]

i think anyone who plays on a open server for only the rank is a turd..lol makes no sense.

 

shut the ranking off.. usually one or 5 admins on.. they already know what to do.this is why i play on fa servers, cuz of the nice and great admins and atmosphere, always friendly and want to help.

 

i think shuffling by kr or kdr is not  the best solution. best solution i think is to have control of your server.i know all wont agree  sorry its just my opinion!

[Mufasa]

if anything though, I think it should also consider how much of the enemys HP you actually took when killing them.. if you get lucky and 1-hit someone after another player took away 90% of their HP, then that should qualify for less +KDR than if you killed them on your own, or indeed by killing them by taking away 90% of their HP after someone else took away 10% already

 

I totally agree with that, but I wanted to limit this post to KR vs KDR because I think it's a more realistic change. I'm sure we could come up with a great formula that works much better than KR, but it would be difficult to implement without the silEnT developers doing it. KR is very broken in the meantime for certain classes and it seems like half the time I play, shuffles go terribly wrong with no admin to react.

Edited February 4, 2018 by TURNTLE

[CheepHeep]

Why havent we tried using shuffle based on PRW yet when its been suggested a billion times?

[TulsaGeoff]

I'm not a fan of the distance factoring into the equation. Theres not much skill in blasting someone's face with a fg42 or knifing them down. If anything I think it should work the other way around. The issue with only shuffling based on PRW is that you can have slayers with lower PRW than obj players and it really screws up the teams. I believe way back (8-10 years ago) with the Silent and HC servers we experimenting with it and shuffle was based on PRW.  Shuffling on a kill metric works better than a win/lose metric.

[Symfony]

PRW is better than KR for sure but I'm not sure if it's the best (Like Geoff said). Regardless something should be done about it because it's a bogus measurement.

[Mufasa]

Why havent we tried using shuffle based on PRW yet when its been suggested a billion times?

 

I'm not a fan of that. I think it could work, but we would need to reset everyone's PRW first - the reason being that currently, there are a lot of players with poor aim and high PRW just because they've had strong teammates that were able to clear the way for them to do the objective. If we shuffled based on current PRW values, it would come down to luck whether the good aimers were spread out. PRW is also very slow to adjust, so it would take weeks or months of poor shuffling before the values adjusted to a reasonable level.

Edited February 4, 2018 by TURNTLE

[RendeL]

If the kr or kdr is the issue, why not reset it like every 5 maps or so. Keep one server where stackers can rate and let others have fun on other servers. For example on silent #2 the kr is insane, as many has kr 20+ or higher, yet they bring no objective playing, only fragging and whining. I now somehow get the point why bots are not reduced there, because it'd be rating heaven otherwise.

Dont know on what shuffle is based on there (maybe prw), but when it goes off, there can be few high kr on other side and other has average kr. Therefore I tell every member/trial not to rely on kr on that server, because it is rare teams are even by kr.

[Chuckun]

Resetting doesn't work. That way shuffle is useless until people earn their kd and then it's useless because of the reasons for this post in the first place, then a reset starts the messed up cycle again..

 

Plus some people care about KD and it would be bad to reset it all the time

[Aniky]

Problem will remain regardless whatever u use, either KR or KDR, because both re inacurate in certain way, like above being mentioned already, there's a lot of fake high KDR players who literally get nowhere that skill and mostly just do knife kills to get it that high. Having a PWR would do it in terms of objective based way, considering u would have an even teams in that perspective but on other side its a lottery each time where best aimers are gonna land. U literally wont find a proper way to balance the teams that are completely even whatever way u pick up. The thing i noticed quite some time now is that when decent players move around and actually even the teams out, u have balanced both sides, but then next map comes on and there's either a useless shuffle from server side or admin side and again u end up with completely f***ed up teams.

 

Having auto shuffle should only be an option when certain team is down by a lot in terms of KDR, like 5 or even more plus, or having at least 3-4 players less, only the way should auto shuffle work, as most of the times u already have even teams and if not, decent ppl move around to even it out on their own. Second problem remains when admins still cant realize that !howfair is missleading and isn't reference to even teams, but regardless of having so many guides about it, forcefully try to even teams, even tho they are already even, specifically because there are ppl with missleading KDR at first place.

 

U could also use an option if u already are going to go with KDR or KD, to use damage dealth and recieved, which is a lot more accurate than pure number of kills in this aspect. By calculating ratio between these two statistics, u could get precise numbers as to how good certain player is and given the fact shuffle would be based on that, u would have even teams in terms of aiming at least, then it all comes down how each time bothers with objective.

 

To sum it up, whatever choice u make for setting up shuffle, teams will never be fair in certain point, either from skill ratio or objective ratio and there will always have to be a need to move around some players to even it out, adjust auto shuffle according to it and re-teach admins on how to look for even teams (with /scores and other indicators for it).

[Mufasa]

Problem will remain regardless whatever u use, either KR or KDR, because both re inacurate in certain way

 

I don't have enough data to make a definitive statement yet, but with the 39-player sample that I have, KDR correlates much more highly with PRW than KR. Still imperfect, but it's a step in the right direction. 

Edited February 4, 2018 by TURNTLE

[Symfony]

http://mygamingtalk.com/forums/topic/3118-what-is-kr-and-prw/

 

here is a link to the silent mod forum and a topic about kr.