The current skill-score is to much win/loss dependet.
You can have an kd/r of 35-5 and high scores but you geting an low skill score if you lose.
I thought this is representing playerskill.
If only one in the team has positive stats and the rest is dying like flies and the team lose you get an "skill-penalty".
This makes no sense to me.
In the current state the skill ladder is a joke.
Stack with some friends marine, play 6-7 rounds, win every round (no problem on some servers), be happy to be in the top10.
If this is about playerskill the win/loss ratio should not have this huge impact on the score like it is atm.
Technically speaking an element that is not taken into account, and arguably never can be, is your field commanding. I remember field commanding a team on Summit. We were being crushed but I managed to umbra/spore my Onos, Gorge and Fade teammates over and over, whilst coaching my team not to do certain things. For example, do not go beyond the CC in atrium otherwise you will have to run through the marines to exit; Fade, make sure you only target marines - especially those with jetpacks etc etc. To an extent this might be considered skill but it's something that just cannot be quantified unless you allow people to vote for their "player of the match" which in and of itself might be too open to abuse.
As for the W:L ratio - this is something I have mentioned and it should certainly have no impact on your score. It is certainly indicative of skill that exceptional players will generally win more games than lose, but I fear the correlation coefficient is probably quite low.
TL;DR Get rid of the weighting that W:L has on the score.
This is exactly backwards. W:L is the only thing that adequately captures the complexity of NS2 since its the primary objective of each game. It doesn't have to be the only thing used to measure skill, but it should be a major component. My preference is for it to work like this:
The number of points transferred from the losing to winning team is determined by the aggregate skill ranking difference between the two where the larger the difference the less points are transferred (i.e. if you're pubstomping, winning will give you little to no points).
How those points are distributed or lost within teams is dependent on some measure of individual skill (e.g. score, k/min, etc) so the highest skilled player on a team gains the most/loses the least points for a win/loss and vice versa for the least skilled player.
The actual skill rating should be kept hidden, but players should be placed into levels/leagues which are public (e.g. skulk (bronze), gorge (silver), lerk (gold), fade (platinum), onos (master), etc)
Disproving your argument is pretty easy. Spec Virsoul commanding in pub games and notice that he probably only wins about half of them (I know comm rounds are not ranked yet).
In addition, my KD is frequently (not always) the highest on the server, and my score is usually first or second (mostly first) and yet if I lose the round (despite my field commanding - which incidentally is normally quite effective) my score will be worse than the absolute inept average player of the opposite team because a whole 1/3 was eliminated.
This is clearly how NOT to rate individual play. Think about it - you're trying to rate an individual based on the overall team.
I have already made the point that you will generally win more rounds than you lose if you're a good player (according to the rankings I have 14 wins to 2 losses - though it hasn't counted about half of my rounds played). If it should have an impact at all I think 1/3 is ridiculous. It might even encourage more stacks...
Since NS2 isn't really a deathmatch game, its how your individual skill contributes to the team victory rather than racking up your score that matters. Having a skill rating system that shows how high of a KDR you can get is a rather useless measure except in the sense of how that KDR contributes towards your team's victory.
Also, a skill rating system doesn't have to get it right every single time as long as it does the vast majority of times. Looking at the current skill rating list, the top ones are mostly D1 comp players or known highly-skilled pub players, which is what you would expect from a good skill rating system. Virsoul may not win every single games he plays, but he wins them more often than not because his skills as a player and/or comm greatly increase his team's chance of victory.
You totally didn't read what I said about VIrsoul - I mentioned the comming, not the field play.
Additionally: http://hive.naturalselection2.com/profile/23757717 Railo is a premiere level player - much better than me - yet he is lower ranked. A quick look at his ratio of wins to losses shows that he has lost 21/57 games, whereas I have lost 2/16
In addition, why did you think I only mentioned the KDR? The score IS involved (as in where you are on the scoreboard). Having a high score and KDR is probably the single most indicative thing of how valuable you are to a team - which is why Fana and Tane routinely (not always but almost always) got the highest scores with Arc.
I'm just against W:L ratio having much of an impact - an impact, yes, but not much.
Edit: Railo's score when he loses is often half of when he wins. While this may be because he might just be playing worse sometimes, I would put forward the hypothesis that it might be that the weighting is too heavy.
Just wanted to highlight that accuracy and total damage inflicted (can show teamwork on marine side, and can account for gorges biling and destroying structures, as examples) should probably have heavier weightings in the ranking algorithm... over K:D and win-loss records (which can be skewed over time with team/skill-stacking... which never happens, right? *sarcasm*).
I might be wrong, but I'm pretty sure that, other than hackers, there are no good players that have low accuracy. Additionally, good team-work and more experienced/knowledgeable players will generally have higher than average damage infliction, on both the enemy and structures.
Also, once you start tracking accuracy, it can be used as a tool to identify potential hackers.
And if possible, separate skill profiles should be created for marines and aliens. Personally, I am a much better marine than alien. I know some who are the opposite. And then, there are those who are good whether they are marines or alien.
Personally, all I want is for "stronger" players to be split up relatively equally on both sides, since the hope is that they will counter each other. The biggest problem, in my honest opinion, is skill/team-stacking, which is no fun at all, especially since it leads to more skill stacking in subsequent rounds...for which the only counter is to rage quit... LOL. But, I've had good games that had equal amounts of strong people on each team, and the deciding factor then becomes teamwork.
Skill bar needs to be enabled in game, and tied to random... I can't stand when it when the other team calls stacks game after game (when in each game teams are vastly different). I'll be the last to join and join the weaker team but it's still a stack.
I understand some people would abuse this to stack. But to show that a round was skill balanced would be nice.
From what I'm reading, player ranking is extremely basic.
Want a high rank? Stack teams, the system cannot account for this sort of simplistic exploitation. Conversely players like Mad Max will be underranked as they try to balance teams.
We really need someone with at least a few hundred hours of game time, and an understanding of data analysis doing the skill system. Look for trends and compare them between players, evaluate value of players on a losing team fairly, don't just tack on negativity because you can't evaluate skilled play in that scenario.
Once again, if there is real interest, I am willing to share what I've come up with for the game.
Server needs to display 50 matches played and has to be vanilla or using mods considered vanilla (However I'm not sure what those mods are, I think some of the admin tools pass the filter)
Server needs to display 50 matches played and has to be vanilla or using mods considered vanilla (However I'm not sure what those mods are, I think some of the admin tools pass the filter)
As far as i know only BNS2 and [Shine] NS2Stats. And no if mods don't override the playerranking.lua functions the server won't send stats to "hive". So also admin tools as Dak and Shine dont work atm with hive.
Surly there's a list of accepted mods out there by now?
How does adding/removing mods affect this by the way ..
eg. lets say
1. server is whitlisted ..
2. server gets mod X installed for some reason, thats not on approved list..
3. few games are played
4. mod gets removed and now only approved mods are on
How does the ranking deal with this?
a) once a server is whitelisted, its whitelisted and always included?
b) doesnt record data for matches with unapproved mods running (so might record a match on vanilla maps, but not on a custom map?
c) removes server from whitelist until 50 games are played with approved mods (with no break of unapproved mods)
October 27, 2013 Heidis Ziegenfarm [HBZ] ns2_summit 23min Loser 375.13[/b]
21 kills and 18 assists with 3 deaths. Also:
Edit: took the link out because it just looks at ALL of my screenshots rather than just one.
Highest damage and structure damage.
In another round I did about the same and had October 27, 2013 Survival of the Fattest ns2_mineshaft 25min Winner 741.96
Anyway the system probably cares too much about the result. In any case, I'm still happy somewhat with the progress as I can now see which servers have the "better" players.
Surly there's a list of accepted mods out there by now?
How does adding/removing mods affect this by the way ..
eg. lets say
1. server is whitlisted ..
2. server gets mod X installed for some reason, thats not on approved list..
3. few games are played
4. mod gets removed and now only approved mods are on
How does the ranking deal with this?
a) once a server is whitelisted, its whitelisted and always included?
b) doesnt record data for matches with unapproved mods running (so might record a match on vanilla maps, but not on a custom map?
c) removes server from whitelist until 50 games are played with approved mods (with no break of unapproved mods)
I am just going to answer your questions:
a) yes unless it gets reported and is removed from the whitelist
b) either it reports all rounds or non (by default), but e.g i build some protections into [Shine] ns2stats so botgames, rounds with cheats on etc are not recorded
c) see a) + b)
@BestProfileName could you explain your statement about which server has "better" player?
Surly there's a list of accepted mods out there by now?
How does adding/removing mods affect this by the way ..
eg. lets say
1. server is whitlisted ..
2. server gets mod X installed for some reason, thats not on approved list..
3. few games are played
4. mod gets removed and now only approved mods are on
How does the ranking deal with this?
a) once a server is whitelisted, its whitelisted and always included?
b) doesnt record data for matches with unapproved mods running (so might record a match on vanilla maps, but not on a custom map?
c) removes server from whitelist until 50 games are played with approved mods (with no break of unapproved mods)
I am just going to answer your questions:
a) yes unless it gets reported and is removed from the whitelist
b) either it reports all rounds or non (by default), but e.g i build some protections into [Shine] ns2stats so botgames, rounds with cheats on etc are not recorded
c) see a) + b)
@BestProfileName could you explain your statement about which server has "better" player?
WHat I meant was that the higher the skill level bar, on average, the better the players there.
Having said that I just played a marine round in which my score was 100 points better than the best on the alien side, and I ended up with 389 points because it was a loss. Crazy, right
The current iteration of the skill rankings is greatly skewed by being so heavily influenced by W/L. It really does need to be a factor but it seems like losing gets close to cutting your skill rating in half just because your team lost.
As the hive ranking currently is it rewards those who play against players below their skill level (not there are many servers with good players around on a routine basis) and punishes players who play vs. other skilled players. If these rankings are going to be used in the Sabot system then they still need to be refined in order for the Sabot system to be effective.
I'm playing more occasionally. Once a week or 2 weeks. The casual kind that get on the top 3 scores (and commander blah). I don't like to wait so public is fine.
It says here i'm level 1 and in the game menu and profile : level 5.... His level 16..???...
Mysterious ways of doing math in there. or maybe i left something behind.
It doesn't show all my games. Maybe because some servers are modded.
That would be a good idea to have a filter on moded server in the game. Just saying.
That would be a good idea to have a filter on moded server in the game. Just saying.
Not a good idea. People are mod scared enough. Back when they were yellow it was impossible to have mods on a server. Admin mods, quality of life mods, etc. MODS ARE GOOD ya hear.
Comments
You totally didn't read what I said about VIrsoul - I mentioned the comming, not the field play.
Additionally: http://hive.naturalselection2.com/profile/23757717 Railo is a premiere level player - much better than me - yet he is lower ranked. A quick look at his ratio of wins to losses shows that he has lost 21/57 games, whereas I have lost 2/16
In addition, why did you think I only mentioned the KDR? The score IS involved (as in where you are on the scoreboard). Having a high score and KDR is probably the single most indicative thing of how valuable you are to a team - which is why Fana and Tane routinely (not always but almost always) got the highest scores with Arc.
I'm just against W:L ratio having much of an impact - an impact, yes, but not much.
Edit: Railo's score when he loses is often half of when he wins. While this may be because he might just be playing worse sometimes, I would put forward the hypothesis that it might be that the weighting is too heavy.
Just wanted to highlight that accuracy and total damage inflicted (can show teamwork on marine side, and can account for gorges biling and destroying structures, as examples) should probably have heavier weightings in the ranking algorithm... over K:D and win-loss records (which can be skewed over time with team/skill-stacking... which never happens, right? *sarcasm*).
I might be wrong, but I'm pretty sure that, other than hackers, there are no good players that have low accuracy. Additionally, good team-work and more experienced/knowledgeable players will generally have higher than average damage infliction, on both the enemy and structures.
Also, once you start tracking accuracy, it can be used as a tool to identify potential hackers.
And if possible, separate skill profiles should be created for marines and aliens. Personally, I am a much better marine than alien. I know some who are the opposite. And then, there are those who are good whether they are marines or alien.
Personally, all I want is for "stronger" players to be split up relatively equally on both sides, since the hope is that they will counter each other. The biggest problem, in my honest opinion, is skill/team-stacking, which is no fun at all, especially since it leads to more skill stacking in subsequent rounds...for which the only counter is to rage quit... LOL. But, I've had good games that had equal amounts of strong people on each team, and the deciding factor then becomes teamwork.
score / time = order number
with a minimum of 10 hours or something.
now people that play more will end up higher which seems wrong
Could you explain that a bit more?
I understand some people would abuse this to stack. But to show that a round was skill balanced would be nice.
Second this, all this time thought stats were broken because I searched for 'slowlefty' instead of 'SlowLefty'.
Want a high rank? Stack teams, the system cannot account for this sort of simplistic exploitation. Conversely players like Mad Max will be underranked as they try to balance teams.
We really need someone with at least a few hundred hours of game time, and an understanding of data analysis doing the skill system. Look for trends and compare them between players, evaluate value of players on a losing team fairly, don't just tack on negativity because you can't evaluate skilled play in that scenario.
Once again, if there is real interest, I am willing to share what I've come up with for the game.
Server needs to display 50 matches played and has to be vanilla or using mods considered vanilla (However I'm not sure what those mods are, I think some of the admin tools pass the filter)
As far as i know only BNS2 and [Shine] NS2Stats. And no if mods don't override the playerranking.lua functions the server won't send stats to "hive". So also admin tools as Dak and Shine dont work atm with hive.
That server needs to not ultra rubberband first
How does adding/removing mods affect this by the way ..
eg. lets say
1. server is whitlisted ..
2. server gets mod X installed for some reason, thats not on approved list..
3. few games are played
4. mod gets removed and now only approved mods are on
How does the ranking deal with this?
a) once a server is whitelisted, its whitelisted and always included?
b) doesnt record data for matches with unapproved mods running (so might record a match on vanilla maps, but not on a custom map?
c) removes server from whitelist until 50 games are played with approved mods (with no break of unapproved mods)
21 kills and 18 assists with 3 deaths. Also:
Edit: took the link out because it just looks at ALL of my screenshots rather than just one.
Highest damage and structure damage.
In another round I did about the same and had October 27, 2013 Survival of the Fattest ns2_mineshaft 25min Winner 741.96
Anyway the system probably cares too much about the result. In any case, I'm still happy somewhat with the progress as I can now see which servers have the "better" players.
a) yes unless it gets reported and is removed from the whitelist
b) either it reports all rounds or non (by default), but e.g i build some protections into [Shine] ns2stats so botgames, rounds with cheats on etc are not recorded
c) see a) + b)
@BestProfileName could you explain your statement about which server has "better" player?
WHat I meant was that the higher the skill level bar, on average, the better the players there.
Having said that I just played a marine round in which my score was 100 points better than the best on the alien side, and I ended up with 389 points because it was a loss. Crazy, right
As the hive ranking currently is it rewards those who play against players below their skill level (not there are many servers with good players around on a routine basis) and punishes players who play vs. other skilled players. If these rankings are going to be used in the Sabot system then they still need to be refined in order for the Sabot system to be effective.
Thanks
As far as I know only unmodded servers are used for the Hive ranking system.
By default all unmodded servers report to the hive. Then there are a few mods which also reenable hive logging (e.g. [Shine] ns2stats).
Please also notice that a server won't show up at hive unless it has more than 50 round send.
Could you please add my server to the Hive list next tuesday?
http://ns2stats.com/server/server/17961
It's got over 200 rounds under its belt.
I'm not bad at all but these guys play against higher skill than public play. So the guy should be above me.
I'm playing more occasionally. Once a week or 2 weeks. The casual kind that get on the top 3 scores (and commander blah). I don't like to wait so public is fine.
It says here i'm level 1 and in the game menu and profile : level 5.... His level 16..???...
Mysterious ways of doing math in there. or maybe i left something behind.
It doesn't show all my games. Maybe because some servers are modded.
That would be a good idea to have a filter on moded server in the game. Just saying.