Pretty much, you cannot make accurate stats even in simple games like CS or COD, not sure how you could ever hope to accomplish that in NS2 lol...
Tracking damage dealt would most likely add significant overhead, especially if its per entity per round... Maybe a lump sum of damage dealt regardless of to what could be tracked.
<!--quoteo(post=1964031:date=Aug 16 2012, 07:26 PM:name=Laosh'Ra)--><div class='quotetop'>QUOTE (Laosh'Ra @ Aug 16 2012, 07:26 PM) <a href="index.php?act=findpost&pid=1964031"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->how can it lead to excessive stacking if the teams are autobalanced by ELO?<!--QuoteEnd--></div><!--QuoteEEnd-->
People don't want to play on public servers that control the teams this aggressively. It would just become frustrating, and it's impossible to stop people disconnecting and joining in the middle of the match making the teams further unbalanced.
I think a lot of people are thinking about this waaaay too seriously. These stats are not important, they are for fun!! Just because you get the top rank doesn't mean you are the best player in the world. I know that when I played dod:s none of the competitive players even cared about the stats, it was just a fun little thing for public servers. Really it's not a big deal and it's not important if it's not an exact representation of skill.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->and why would they not be stacked in comparison, if the goal is kdr and score (which is also heavily influenced by having a good team)?<!--QuoteEnd--></div><!--QuoteEEnd-->
They could be stacked but you would make less points. Since high ranking players would need to maintain a 4 or 5:1 , kd ratio in order to stay in the positive points for that round. So you could actually lose points by playing against noobs and dying a few times, even if you have a positive k:d. If it was based on wins/losses then stacking the teams would almost guarantee that you wouldn't lose any points, even if you only made a small amount from the win.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->i am concerned with very low morale (and i've seen in other threads that several people share this concern) as in people giving up risks in favor of turtling the rest of the match in order to gain some kills and score... the other team possibly trying to prolong the game as long as possible (spawnkill etc.) will further encourage this behavior.<!--QuoteEnd--></div><!--QuoteEEnd-->
If you are on the losing team and it's a lame duck situation (game is already lost) then you may as well just F4 and start a new game. If the stats encourage this then I say that's a good thing. People will give up the game and turtle etc. with or without stats.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->in fact i have seen this issue today on one of the stat-servers. our second marine base was heavily attacked (power down). i tried to motivate some teammates to come with me to take it back, 2 guys refused (they even had jetpacks) as in "it's already lost".<!--QuoteEnd--></div><!--QuoteEEnd-->
Yeah, did you ask the guys if they did that for the stats? I doubt they did. People will do that ###### anyway. It doesn't matter if you have stats. I seriously doubt anyone cares that much about their rank. I don't think having k:d stats will completely change the way people play the game. I also think it's pretty hard to farm kills without entire co operation from your team, which is unlikely. Even in that scenario, the losing team will just F4 and then start a new round. It's not a big deal.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->of course it will never be flawless. but there is no point in adding extra flaws imo. if stacking will be a problem then i'd really prefer elo-based teams, or simply randomly assigned teams. people might be annoyed as in "damn i wanted to play on the other side this time" but it's better than stacking or people putting their greed above the team effort.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't think that will ever work. The stats are just a fun thing, they shouldn't be controlling who goes on what team. People just want to join a pub server and have fun. Keep things like that for gathers et al.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->also, whats the advantage of keeping it simple?<!--QuoteEnd--></div><!--QuoteEEnd-->
Everyone understands it. Everyone knows it's imperfect, but they know why. Rather than having some mystical form of rankings where you need to do multiple formula just to work out how much a kill is worth.
It's good to have this discussion but I think people take the stats too seriously, as if the ranking actually means something. It's just a bit of fun to check your stats out and have a little rating. Not a big deal. Not a measure of your skill. Not proving who is better. Just fun.
If someone bites the extractor and gets a few points instead of you, then it's not a big deal. Move on, you didn't get +2 to your score...not the end of the world. It's just like the current score mechanic in game, do you really think people go around trying to get amazing score, no, they just play and have fun and don't worry about it.
ScardyBobScardyBobJoin Date: 2009-11-25Member: 69528Forum Admins, Forum Moderators, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Shadow
I'm a fan of a stats-system + Elo rating, but making it secret and only using it for a matchmaking/gather system (i.e. automatically matching players with similar Elo ratings). This is how games such as SC2 get around the pub-stomping that happens so frequently in NS2.
<!--quoteo(post=1964138:date=Aug 16 2012, 05:44 PM:name=ScardyBob)--><div class='quotetop'>QUOTE (ScardyBob @ Aug 16 2012, 05:44 PM) <a href="index.php?act=findpost&pid=1964138"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I'm a fan of a stats-system + Elo rating, but making it secret and only using it for a matchmaking/gather system (i.e. automatically matching players with similar Elo ratings). This is how games such as SC2 get around the pub-stomping that happens so frequently in NS2.<!--QuoteEnd--></div><!--QuoteEEnd-->
It's not very easy to match make in a timely manner when you only have 200-400 concurrent players compared to thousands or tens of thousands for sc2. I'm not sure match making in NS2 will work unless we get up to 1k+ players and even then it will have to be loose with the skill brackets.
The stat system is cool! I like it for what it is in its rawness.
The problem is that it changes the gameplay on the servers that have stats enabled. Games become horrendously stacked, and besides... IP camping with a fade or egg camping as marines is hardly a good way to gauge someones skill.
After my first 3 rounds with stats enabled I had over a 20 KDR. Which then eventually leveled off to 6. Now when I join games I honestly don't care about padding my stats and like to goof off in pubs. So, now I have degraded to a 4.8 KDR.
I can't help but get frustrated knowing that people are going to pad their stats with their friends and call other people noobs for not having their KDR. If I really wanted to I could wait in the RR till one team has the upperhand then jump in game... go fade or grab a jetpack and get stupid high KDR's. That doesn't make me a good player... just clever at abusing the stats system.
I avoid the stats servers now. People play like cowards and never push out of base. The gameplay is much less enjoyable when you know everything you're doing is being recorded. Besides NS2 isn't the kind of game where stats matter. Getting kills does matter a lot, but if you sacrifice your life over and over to kill harvesters or shells then you would be an incredible team player but have a terrible KDR (At least more often than not).
I think stats make even less sense for competitive play... where teamplay should be the number 1 priority. Not individual stats.
That's my two cents. I honestly think the stats are overall hurting the public gameplay of NS. I love just hopping into a server and doing fun strats with people and not caring about having a super good KDR. Public servers are for fun, and ever since stats I've seen people hardcore raging at eachother and lots of drama. All of the really good players can easily go 40-1 or 90-4 or even better. We have all done rounds like that so I don't see why having stat proof matters to anyone. I have a screenshot library of all my scores over 7:1 if anyone ever needs proof :-P
Btw it does suck that most of my best KDR rounds didn't even get saved XD
<!--quoteo(post=1964077:date=Aug 16 2012, 03:02 PM:name=Khyron)--><div class='quotetop'>QUOTE (Khyron @ Aug 16 2012, 03:02 PM) <a href="index.php?act=findpost&pid=1964077"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->...The way to mitigate these problems is to better measure what is actually going on. The best way to measure what's going on is to measure more things. So for me the question is what things can be measured that reflect how effectively a player is contributing to the team's objective.
There are two pieces of information I'd be interested in tracking in addition to the currently available stats. I'm not sure what's possible in terms of technical limitations, so I'm just going to throw this out there...
...The first is damage dealt...
...The next piece of info I'd be interested in tracking is something that measures the relative worth of my kills...<!--QuoteEnd--></div><!--QuoteEEnd-->
I agree with him.
We shouldn't just track KDR but also damage dealt and even healed, gorges and welding marines contribute to the success of a match even though they are not killing.
This game is much deeper than just killing, there is constructing, repairing, blocking, healing etc. all of those should be tracked to make the actual estimated representation of a player level more accurate.
<!--quoteo(post=1964183:date=Aug 16 2012, 07:17 PM:name=.ADHd)--><div class='quotetop'>QUOTE (.ADHd @ Aug 16 2012, 07:17 PM) <a href="index.php?act=findpost&pid=1964183"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->The stat system is cool! I like it for what it is in its rawness.
The problem is that it changes the gameplay on the servers that have stats enabled. Games become horrendously stacked, and besides... IP camping with a fade or egg camping as marines is hardly a good way to gauge someones skill.
After my first 3 rounds with stats enabled I had over a 20 KDR. Which then eventually leveled off to 6. Now when I join games I honestly don't care about padding my stats and like to goof off in pubs. So, now I have degraded to a 4.8 KDR.
I can't help but get frustrated knowing that people are going to pad their stats with their friends and call other people noobs for not having their KDR. If I really wanted to I could wait in the RR till one team has the upperhand then jump in game... go fade or grab a jetpack and get stupid high KDR's. That doesn't make me a good player... just clever at abusing the stats system.
I avoid the stats servers now. People play like cowards and never push out of base. The gameplay is much less enjoyable when you know everything you're doing is being recorded. Besides NS2 isn't the kind of game where stats matter. Getting kills does matter a lot, but if you sacrifice your life over and over to kill harvesters or shells then you would be an incredible team player but have a terrible KDR (At least more often than not).
I think stats make even less sense for competitive play... where teamplay should be the number 1 priority. Not individual stats.
That's my two cents. I honestly think the stats are overall hurting the public gameplay of NS. I love just hopping into a server and doing fun strats with people and not caring about having a super good KDR. Public servers are for fun, and ever since stats I've seen people hardcore raging at eachother and lots of drama. All of the really good players can easily go 40-1 or 90-4 or even better. We have all done rounds like that so I don't see why having stat proof matters to anyone. I have a screenshot library of all my scores over 7:1 if anyone ever needs proof :-P
Btw it does suck that most of my best KDR rounds didn't even get saved XD<!--QuoteEnd--></div><!--QuoteEEnd-->
I haven't seen any of that.
Games play the same way now as they did before stats were enabled. I don't notice any difference between stats servers and non-stats servers. There is a lot of both for people to choose from.
I also haven't seen anyone play the "your KDR is ###### therefore you are ######" card. Not a single person.
fanaticThis post has been edited.Join Date: 2003-07-23Member: 18377Members, Constellation, Squad Five Blue
edited August 2012
<!--quoteo(post=1964183:date=Aug 17 2012, 12:17 AM:name=.ADHd)--><div class='quotetop'>QUOTE (.ADHd @ Aug 17 2012, 12:17 AM) <a href="index.php?act=findpost&pid=1964183"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I have a screenshot library of all my scores over 7:1 if anyone ever needs proof :-P<!--QuoteEnd--></div><!--QuoteEEnd--> The saddest post.
DghelneshiAims to surpass Fana in post edits.Join Date: 2011-11-01Member: 130634Members, Squad Five Blue, Reinforced - Shadow
ELO ratings don't even seem to work for a game as simple as Quake (everything below ~2200 Rating is essentially random from my experience), so unfortunately I can't see it working at all in NS2.
SintJoin Date: 2007-01-09Member: 59540Members, Squad Five Blue
Thanks for all ideas and feedback!
<b>Hiding stats</b> For those who do not want to see stats on any servers we'll add option to disable them. Simply by using 'stats off' and they will be disabled on every server. We could also hide your personal page at website same time. Data will stay at database and can be enabled again at any point by 'stats on', anyway command is not available yet, but should be at end of next week. So this way nobody can see your kpd or other stats if you don't want to.
<b>About ELO autobalance</b> Personally I would like usually to play on servers where teams would be always automatically balanced. Ofc it would need to work decently for it to be good. I understand that some people would not like it very much. If there would be some kind of fast vote for it, it could work on most servers.
<b>ELO rating</b> So problems with it so far are team stacking and rating farming. Team stacking also happens without it, but rating purely based on wins could make it worse. To address these things there are some options: - Limit how many points per round you can get (this is must have imo) - Limit how many points per round you can lose (also must have) - If your average team ELO rating is significantly higher than enemy team you get lot less points from winning. People who stack might generally have higher ELO ratings so stacking would make team have higher ELO and would not be so good idea to stack. Also losing team would lose lot less points. - Calculate each players points in team individually based on their performance. (Would be tricky to take all things in consideration) Thou this would allow "best" players maybe to even get points in losing team.
If we want to keep it simple it should be just win/lose with some rules to discourage team stacking. Or to have it enabled only when teams are arranged by ELO rating. Anyway we are still thinking and accepting ideas.
Looking pretty good. I am especially interested in the stats hiding. I prefer them for my own edification and don't really want them floating out there so people can step up next to me and have a measuring contest. This has always been my only real hang up with stat tracking.
<!--quoteo(post=1964356:date=Aug 17 2012, 05:34 AM:name=Sint)--><div class='quotetop'>QUOTE (Sint @ Aug 17 2012, 05:34 AM) <a href="index.php?act=findpost&pid=1964356"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec--><b>Hiding stats</b> For those who do not want to see stats on any servers we'll add option to disable them. Simply by using 'stats off' and they will be disabled on every server. We could also hide your personal page at website same time. Data will stay at database and can be enabled again at any point by 'stats on', anyway command is not available yet, but should be at end of next week. So this way nobody can see your kpd or other stats if you don't want to.<!--QuoteEnd--></div><!--QuoteEEnd-->
I tell you one thing the stats site is good for, i was gettin trolled the other day by someone who is angry at the nexzil clan, and by looking at who i thought it was on the ns2stats site, i was able to to see that the trolls name was the only other alias for that person i suspected.
I dont have much of an opinion on the stats, not sure the commander stats work properly tho, i have tried to test it by going against gorgeous and rantoloy for a couple of rounds, just so i actually have a commanding loss for each side, that worked well. But the time played for marine commander seems to be broken.
Good effort gettin the site up. Certainly helped me scout out a rata.
i highly doubt this is actually happening, but just in case: could feign death count as an extra kill?
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->If there would be some kind of fast vote for it, it could work on most servers.<!--QuoteEnd--></div><!--QuoteEEnd--> dePARA has a "voterandom" option on HBZ, maybe you could work with that. you could also make it a server setting instead of voting, so people can pick their personal preference (by picking a server) without things changing every round (depending on the players' voting mood).
ScardyBobScardyBobJoin Date: 2009-11-25Member: 69528Forum Admins, Forum Moderators, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Shadow
<!--quoteo(post=1964176:date=Aug 16 2012, 04:00 PM:name=GORGEous)--><div class='quotetop'>QUOTE (GORGEous @ Aug 16 2012, 04:00 PM) <a href="index.php?act=findpost&pid=1964176"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->It's not very easy to match make in a timely manner when you only have 200-400 concurrent players compared to thousands or tens of thousands for sc2. I'm not sure match making in NS2 will work unless we get up to 1k+ players and even then it will have to be loose with the skill brackets.<!--QuoteEnd--></div><!--QuoteEEnd--> I'm going on the presumption that NS2 will be a successful game on launch, such that we could expect 1000+ concurrent players (roughly 500-750 will get you on <a href="http://store.steampowered.com/stats/" target="_blank">Steam's Top 100 played games list</a>)
<!--quoteo(post=1964344:date=Aug 17 2012, 04:00 AM:name=Dghelneshi)--><div class='quotetop'>QUOTE (Dghelneshi @ Aug 17 2012, 04:00 AM) <a href="index.php?act=findpost&pid=1964344"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->ELO ratings don't even seem to work for a game as simple as Quake (everything below ~2200 Rating is essentially random from my experience), so unfortunately I can't see it working at all in NS2.<!--QuoteEnd--></div><!--QuoteEEnd--> I'm a bit curious why this didn't work so well (got any links on how they calculated the Elo ratings)? As far as I can tell, SC2 uses a modified form of Elo ratings with good success. I've figured that NS2 could probably use something similar.
<!--quoteo(post=1964453:date=Aug 17 2012, 10:18 AM:name=Hunter.S.T.)--><div class='quotetop'>QUOTE (Hunter.S.T. @ Aug 17 2012, 10:18 AM) <a href="index.php?act=findpost&pid=1964453"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I tell you one thing the stats site is good for, i was gettin trolled the other day by someone who is angry at the nexzil clan, and by looking at who i thought it was on the ns2stats site, i was able to to see that the trolls name was the only other alias for that person i suspected.<!--QuoteEnd--></div><!--QuoteEEnd--> I must admit, the thing I like about the site is that its an effective smurf-name detector :)
I've been thinking about it and I'm interested to see if a win/loss system does work. Perhaps I'm just too attached to old ways which I've seen used before. It's no fun too be on the opposite side of a heavily stacked team, so if you could prevent that by quickly randomizing the teams based on rank then it could be awesome.
I guess I sort of feel like it's a bit harsh to lose points for something outside your control, like if you have a bad com, or if your team mates suck etc.
DghelneshiAims to surpass Fana in post edits.Join Date: 2011-11-01Member: 130634Members, Squad Five Blue, Reinforced - Shadow
edited August 2012
ELO for 1v1 games/game modes mostly works decently.
Once you get to higher player numbers per team, it is effectively random until every player has played at least a few hundred games in total since there are too many variables for a team game to level out in a short amount of time. In NS2 it's even worse because there are no set goals for a player to achieve. In Quake you do can just count damage done per time and make the first third move up and the last third move down in your system (this is how the official system works; QLRanks.com seems to use wins/losses alone, which can be pretty stupid when you lose 160 ELO points from a single 9-10 game), but in NS2 there are just so many things you can do, which you can either not influence yourself or not quantify properly (especially the relations between those actions, i.e. how much a building kill is worth in relation to a player kill).
Additionally, there are even completely different roles like the commanders or Gorges. It's pretty much impossible to quantify commander performance other than by win percentage, which will be pretty random until you've played a lot of games with all the different pub players.
ScardyBobScardyBobJoin Date: 2009-11-25Member: 69528Forum Admins, Forum Moderators, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Shadow
<!--quoteo(post=1964562:date=Aug 17 2012, 07:25 PM:name=Wilson)--><div class='quotetop'>QUOTE (Wilson @ Aug 17 2012, 07:25 PM) <a href="index.php?act=findpost&pid=1964562"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->SC2 is slightly different though, since it's 1v1.
I've been thinking about it and I'm interested to see if a win/loss system does work. Perhaps I'm just too attached to old ways which I've seen used before. It's no fun too be on the opposite side of a heavily stacked team, so if you could prevent that by quickly randomizing the teams based on rank then it could be awesome.
I guess I sort of feel like it's a bit harsh to lose points for something outside your control, like if you have a bad com, or if your team mates suck etc.<!--QuoteEnd--></div><!--QuoteEEnd--> SC2 does its ranking/rating system for up to 4v4's afaik. Also, it sucks that you lose a game, let alone Elo points, for a bad comm. That's just something you have to accept when playing a team-based online multiplayer game.
<!--quoteo(post=1964729:date=Aug 18 2012, 10:27 AM:name=Dghelneshi)--><div class='quotetop'>QUOTE (Dghelneshi @ Aug 18 2012, 10:27 AM) <a href="index.php?act=findpost&pid=1964729"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->ELO for 1v1 games/game modes mostly works decently.
Once you get to higher player numbers per team, it is effectively random until every player has played at least a few hundred games in total since there are too many variables for a team game to level out in a short amount of time. In NS2 it's even worse because there are no set goals for a player to achieve. In Quake you do can just count damage done per time and make the first third move up and the last third move down in your system (this is how the official system works; QLRanks.com seems to use wins/losses alone, which can be pretty stupid when you lose 160 ELO points from a single 9-10 game), but in NS2 there are just so many things you can do, which you can either not influence yourself or not quantify properly (especially the relations between those actions, i.e. how much a building kill is worth in relation to a player kill).
Additionally, there are even completely different roles like the commanders or Gorges. It's pretty much impossible to quantify commander performance other than by win percentage, which will be pretty random until you've played a lot of games with all the different pub players.<!--QuoteEnd--></div><!--QuoteEEnd--> This is why I think it should stick to wins/losses only, since that is the ultimate goal of NS2 matches, with the total points distributed or taken evenly from each player on the winning/losing team.
I'm not saying the system is perfect, but I do think it'll do a better job of making more evenly matched games than the current system (which is really no system). At a minimum, it would reduce the number of (to borrow SC2-wording) platinum players against bronze players and vice versa.
i think an ELO-type system should compute points for: accurate shooting, life expectancy (adjusted per lifeform), win/loss, and resources destroyed.
#1 is important since it is a good indicator of fps skill (or having an aimbot) #2 is important since it somewhat shows good game sense/ability to protect your investments #3 is important since it's the goal of the game #4 is important so that people don't get credit for standing around while the team carries them.
<!--quoteo(post=1964822:date=Aug 19 2012, 02:12 AM:name=ScardyBob)--><div class='quotetop'>QUOTE (ScardyBob @ Aug 19 2012, 02:12 AM) <a href="index.php?act=findpost&pid=1964822"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->SC2 does its ranking/rating system for up to 4v4's afaik. Also, it sucks that you lose a game, let alone Elo points, for a bad comm. That's just something you have to accept when playing a team-based online multiplayer game.
This is why I think it should stick to wins/losses only, since that is the ultimate goal of NS2 matches, with the total points distributed or taken evenly from each player on the winning/losing team.
I'm not saying the system is perfect, but I do think it'll do a better job of making more evenly matched games than the current system (which is really no system). At a minimum, it would reduce the number of (to borrow SC2-wording) platinum players against bronze players and vice versa.<!--QuoteEnd--></div><!--QuoteEEnd-->
With potentially 16 players per team, win/loss ratios lose (almost) all meaning for the individual. At best you'll see a trend for how much a player influences the outcome after several hundred played games. I play another game with 15 players per team that tracks stats, and there the difference in wins between terrible and exceptional players is not more than 15%.
It's incredibly hard to rate individual skill in team games. I can say from experience that systems like Elo and TrueSkill become problematic at 3vs3 games already.
<!--quoteo(post=1964836:date=Aug 19 2012, 02:09 AM:name=Raza.)--><div class='quotetop'>QUOTE (Raza. @ Aug 19 2012, 02:09 AM) <a href="index.php?act=findpost&pid=1964836"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I play another game with 15 players per team that tracks stats, and there the difference in wins between terrible and exceptional players is not more than 15%.<!--QuoteEnd--></div><!--QuoteEEnd-->
Care to share what game that is? I'd be interested in seeing the stats. Also, if it really doesn't work as you say it does then it might give the makers of ns2stats some valuable information before they spend time working on the ranking system.
I do believe you could still have fairly balanced games by just using the k:d elo ranking. I feel like before thinking about auto balancing of teams you should just experiment a little and see how each method works. You already record the k:d so it shouldn't be too hard to add or take away elo points for each one and see how it plays out. Obviously gorges and coms won't be ranked properly in this system but you'll probably need to do some sort of separate stats for them anyway. Having a small section with "top 10 gorges" and "top 10 coms" that are based on different stats from the main rankings.
i think there's something quite wrong with the commander stats. just had my first match as marine commander, stats show the game but not me commanding. we had two more commanders (things were pretty chaotic) until i took over a second time and stayed there for the rest of the match (pretty sure it was longer than half the match-time, although i left the chair to defend the base a few times, maybe that could be the reason?)
what it does show (which it did not before), is alien command stats which seem quite off: i did not play as alien commander on the stat servers yet, but it shows first hive upgrades: 25x crag hive, 3x shift hive and 1x shade hive. that's more than win+loss recorded matches i got... the time played stats are 5s not-commanding for marines (and no commanding) and more than 4 hours of alien not-commanding (and no commanding stat). there are neither wins nor losses displayed. here is the match: <a href="http://ns2stats.org/round/round/1426" target="_blank">http://ns2stats.org/round/round/1426</a>
ZeikkoJoin Date: 2007-12-16Member: 63179Members, Squad Five Blue, NS2 Map Tester
I'm back from holidays and added some new features including <a href="http://ns2stats.org/player/index" target="_blank">better player</a> search. This search finds players by their ingame nick and steam nick. There's still some work to be done for finding very short nicks.
I also added a round length pie diagram to the <a href="http://ns2stats.org/map/map/1" target="_blank">maps</a>. I will be adding more of these diagrams soon.
I also worked on some other bigger features but they're not ready yet.
Thanks for the bug report Laosh'Ra. I will take look into it soon. Probably need to get some sleep first. :)
A lot of games on stats enabled servers don't seem to show up. Even with nearly full 9v9 combat for most of the game, I just saw an hour long game not get parsed by the stats. Could have been a problem with the server's implementation of stats, but I've seen other games where it has skipped a game in between games that it parsed.
supsuJoin Date: 2012-04-24Member: 151023Members, Squad Five Blue
<!--quoteo(post=1965067:date=Aug 20 2012, 02:57 AM:name=GORGEous)--><div class='quotetop'>QUOTE (GORGEous @ Aug 20 2012, 02:57 AM) <a href="index.php?act=findpost&pid=1965067"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->A lot of games on stats enabled servers don't seem to show up. Even with nearly full 9v9 combat for most of the game, I just saw an hour long game not get parsed by the stats. Could have been a problem with the server's implementation of stats, but I've seen other games where it has skipped a game in between games that it parsed.<!--QuoteEnd--></div><!--QuoteEEnd--> Interesting. We've got the log from your game that didn't show up on the web page but did show up in our dump site. It stopped adding log lines at some point even though the log wasn't even near its limits, we have to see what's up with this.
commander stats seem much better now, nice job with that :) just one more thing: it shows crag hive as first upgrade although it also says i've never been alien commander. maybe i was, but not long enough for the game to count it as commanding? (i don't remember tbh)
ZeikkoJoin Date: 2007-12-16Member: 63179Members, Squad Five Blue, NS2 Map Tester
<!--quoteo(post=1965154:date=Aug 20 2012, 11:39 AM:name=Laosh'Ra)--><div class='quotetop'>QUOTE (Laosh'Ra @ Aug 20 2012, 11:39 AM) <a href="index.php?act=findpost&pid=1965154"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->commander stats seem much better now, nice job with that :) just one more thing: it shows crag hive as first upgrade although it also says i've never been alien commander. maybe i was, but not long enough for the game to count it as commanding? (i don't remember tbh)<!--QuoteEnd--></div><!--QuoteEEnd-->
Yeah i fixed the marine commander stats last night. Thanks for reporting the bugs. The alien commander first hive upgrade stats are still not working quite right. Currently it shows stats from rounds where you never even entered the hive. I will fix that as soon as i can. That is hopefully this evening.
ZeikkoJoin Date: 2007-12-16Member: 63179Members, Squad Five Blue, NS2 Map Tester
I fixed the alien commander first hive upgrade graph now. I hope i got it right this time. If it proves to be correct i'll proceed on adding more charts about upgrade choices of the commanders.
Comments
Tracking damage dealt would most likely add significant overhead, especially if its per entity per round... Maybe a lump sum of damage dealt regardless of to what could be tracked.
People don't want to play on public servers that control the teams this aggressively. It would just become frustrating, and it's impossible to stop people disconnecting and joining in the middle of the match making the teams further unbalanced.
I think a lot of people are thinking about this waaaay too seriously. These stats are not important, they are for fun!! Just because you get the top rank doesn't mean you are the best player in the world. I know that when I played dod:s none of the competitive players even cared about the stats, it was just a fun little thing for public servers. Really it's not a big deal and it's not important if it's not an exact representation of skill.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->and why would they not be stacked in comparison, if the goal is kdr and score (which is also heavily influenced by having a good team)?<!--QuoteEnd--></div><!--QuoteEEnd-->
They could be stacked but you would make less points. Since high ranking players would need to maintain a 4 or 5:1 , kd ratio in order to stay in the positive points for that round. So you could actually lose points by playing against noobs and dying a few times, even if you have a positive k:d. If it was based on wins/losses then stacking the teams would almost guarantee that you wouldn't lose any points, even if you only made a small amount from the win.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->i am concerned with very low morale (and i've seen in other threads that several people share this concern) as in people giving up risks in favor of turtling the rest of the match in order to gain some kills and score...
the other team possibly trying to prolong the game as long as possible (spawnkill etc.) will further encourage this behavior.<!--QuoteEnd--></div><!--QuoteEEnd-->
If you are on the losing team and it's a lame duck situation (game is already lost) then you may as well just F4 and start a new game. If the stats encourage this then I say that's a good thing. People will give up the game and turtle etc. with or without stats.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->in fact i have seen this issue today on one of the stat-servers. our second marine base was heavily attacked (power down). i tried to motivate some teammates to come with me to take it back, 2 guys refused (they even had jetpacks) as in "it's already lost".<!--QuoteEnd--></div><!--QuoteEEnd-->
Yeah, did you ask the guys if they did that for the stats? I doubt they did. People will do that ###### anyway. It doesn't matter if you have stats. I seriously doubt anyone cares that much about their rank. I don't think having k:d stats will completely change the way people play the game. I also think it's pretty hard to farm kills without entire co operation from your team, which is unlikely. Even in that scenario, the losing team will just F4 and then start a new round. It's not a big deal.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->of course it will never be flawless. but there is no point in adding extra flaws imo. if stacking will be a problem then i'd really prefer elo-based teams, or simply randomly assigned teams. people might be annoyed as in "damn i wanted to play on the other side this time" but it's better than stacking or people putting their greed above the team effort.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't think that will ever work. The stats are just a fun thing, they shouldn't be controlling who goes on what team. People just want to join a pub server and have fun. Keep things like that for gathers et al.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->also, whats the advantage of keeping it simple?<!--QuoteEnd--></div><!--QuoteEEnd-->
Everyone understands it. Everyone knows it's imperfect, but they know why. Rather than having some mystical form of rankings where you need to do multiple formula just to work out how much a kill is worth.
It's good to have this discussion but I think people take the stats too seriously, as if the ranking actually means something. It's just a bit of fun to check your stats out and have a little rating. Not a big deal. Not a measure of your skill. Not proving who is better. Just fun.
If someone bites the extractor and gets a few points instead of you, then it's not a big deal. Move on, you didn't get +2 to your score...not the end of the world. It's just like the current score mechanic in game, do you really think people go around trying to get amazing score, no, they just play and have fun and don't worry about it.
It's not very easy to match make in a timely manner when you only have 200-400 concurrent players compared to thousands or tens of thousands for sc2. I'm not sure match making in NS2 will work unless we get up to 1k+ players and even then it will have to be loose with the skill brackets.
The problem is that it changes the gameplay on the servers that have stats enabled. Games become horrendously stacked, and besides... IP camping with a fade or egg camping as marines is hardly a good way to gauge someones skill.
After my first 3 rounds with stats enabled I had over a 20 KDR. Which then eventually leveled off to 6. Now when I join games I honestly don't care about padding my stats and like to goof off in pubs. So, now I have degraded to a 4.8 KDR.
I can't help but get frustrated knowing that people are going to pad their stats with their friends and call other people noobs for not having their KDR. If I really wanted to I could wait in the RR till one team has the upperhand then jump in game... go fade or grab a jetpack and get stupid high KDR's. That doesn't make me a good player... just clever at abusing the stats system.
I avoid the stats servers now. People play like cowards and never push out of base. The gameplay is much less enjoyable when you know everything you're doing is being recorded. Besides NS2 isn't the kind of game where stats matter. Getting kills does matter a lot, but if you sacrifice your life over and over to kill harvesters or shells then you would be an incredible team player but have a terrible KDR (At least more often than not).
I think stats make even less sense for competitive play... where teamplay should be the number 1 priority. Not individual stats.
That's my two cents. I honestly think the stats are overall hurting the public gameplay of NS. I love just hopping into a server and doing fun strats with people and not caring about having a super good KDR. Public servers are for fun, and ever since stats I've seen people hardcore raging at eachother and lots of drama. All of the really good players can easily go 40-1 or 90-4 or even better. We have all done rounds like that so I don't see why having stat proof matters to anyone. I have a screenshot library of all my scores over 7:1 if anyone ever needs proof :-P
Btw it does suck that most of my best KDR rounds didn't even get saved XD
There are two pieces of information I'd be interested in tracking in addition to the currently available stats. I'm not sure what's possible in terms of technical limitations, so I'm just going to throw this out there...
...The first is damage dealt...
...The next piece of info I'd be interested in tracking is something that measures the relative worth of my kills...<!--QuoteEnd--></div><!--QuoteEEnd-->
I agree with him.
We shouldn't just track KDR but also damage dealt and even healed, gorges and welding marines contribute to the success of a match even though they are not killing.
This game is much deeper than just killing, there is constructing, repairing, blocking, healing etc. all of those should be tracked to make the actual estimated representation of a player level more accurate.
The problem is that it changes the gameplay on the servers that have stats enabled. Games become horrendously stacked, and besides... IP camping with a fade or egg camping as marines is hardly a good way to gauge someones skill.
After my first 3 rounds with stats enabled I had over a 20 KDR. Which then eventually leveled off to 6. Now when I join games I honestly don't care about padding my stats and like to goof off in pubs. So, now I have degraded to a 4.8 KDR.
I can't help but get frustrated knowing that people are going to pad their stats with their friends and call other people noobs for not having their KDR. If I really wanted to I could wait in the RR till one team has the upperhand then jump in game... go fade or grab a jetpack and get stupid high KDR's. That doesn't make me a good player... just clever at abusing the stats system.
I avoid the stats servers now. People play like cowards and never push out of base. The gameplay is much less enjoyable when you know everything you're doing is being recorded. Besides NS2 isn't the kind of game where stats matter. Getting kills does matter a lot, but if you sacrifice your life over and over to kill harvesters or shells then you would be an incredible team player but have a terrible KDR (At least more often than not).
I think stats make even less sense for competitive play... where teamplay should be the number 1 priority. Not individual stats.
That's my two cents. I honestly think the stats are overall hurting the public gameplay of NS. I love just hopping into a server and doing fun strats with people and not caring about having a super good KDR. Public servers are for fun, and ever since stats I've seen people hardcore raging at eachother and lots of drama. All of the really good players can easily go 40-1 or 90-4 or even better. We have all done rounds like that so I don't see why having stat proof matters to anyone. I have a screenshot library of all my scores over 7:1 if anyone ever needs proof :-P
Btw it does suck that most of my best KDR rounds didn't even get saved XD<!--QuoteEnd--></div><!--QuoteEEnd-->
I haven't seen any of that.
Games play the same way now as they did before stats were enabled. I don't notice any difference between stats servers and non-stats servers. There is a lot of both for people to choose from.
I also haven't seen anyone play the "your KDR is ###### therefore you are ######" card. Not a single person.
The saddest post.
<b>Hiding stats</b>
For those who do not want to see stats on any servers we'll add option to disable them. Simply by using 'stats off' and they will be disabled on every server. We could also hide your personal page at website same time. Data will stay at database and can be enabled again at any point by 'stats on', anyway command is not available yet, but should be at end of next week. So this way nobody can see your kpd or other stats if you don't want to.
<b>About ELO autobalance</b>
Personally I would like usually to play on servers where teams would be always automatically balanced. Ofc it would need to work decently for it to be good. I understand that some people would not like it very much. If there would be some kind of fast vote for it, it could work on most servers.
<b>ELO rating</b>
So problems with it so far are team stacking and rating farming. Team stacking also happens without it, but rating purely based on wins could make it worse. To address these things there are some options:
- Limit how many points per round you can get (this is must have imo)
- Limit how many points per round you can lose (also must have)
- If your average team ELO rating is significantly higher than enemy team you get lot less points from winning. People who stack might generally have higher ELO ratings so stacking would make team have higher ELO and would not be so good idea to stack. Also losing team would lose lot less points.
- Calculate each players points in team individually based on their performance. (Would be tricky to take all things in consideration) Thou this would allow "best" players maybe to even get points in losing team.
If we want to keep it simple it should be just win/lose with some rules to discourage team stacking. Or to have it enabled only when teams are arranged by ELO rating. Anyway we are still thinking and accepting ideas.
<!--quoteo(post=1964356:date=Aug 17 2012, 05:34 AM:name=Sint)--><div class='quotetop'>QUOTE (Sint @ Aug 17 2012, 05:34 AM) <a href="index.php?act=findpost&pid=1964356"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec--><b>Hiding stats</b>
For those who do not want to see stats on any servers we'll add option to disable them. Simply by using 'stats off' and they will be disabled on every server. We could also hide your personal page at website same time. Data will stay at database and can be enabled again at any point by 'stats on', anyway command is not available yet, but should be at end of next week. So this way nobody can see your kpd or other stats if you don't want to.<!--QuoteEnd--></div><!--QuoteEEnd-->
I dont have much of an opinion on the stats, not sure the commander stats work properly tho, i have tried to test it by going against gorgeous and rantoloy for a couple of rounds, just so i actually have a commanding loss for each side, that worked well. But the time played for marine commander seems to be broken.
Good effort gettin the site up. Certainly helped me scout out a rata.
Hunter
i highly doubt this is actually happening, but just in case: could feign death count as an extra kill?
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->If there would be some kind of fast vote for it, it could work on most servers.<!--QuoteEnd--></div><!--QuoteEEnd-->
dePARA has a "voterandom" option on HBZ, maybe you could work with that.
you could also make it a server setting instead of voting, so people can pick their personal preference (by picking a server) without things changing every round (depending on the players' voting mood).
I'm going on the presumption that NS2 will be a successful game on launch, such that we could expect 1000+ concurrent players (roughly 500-750 will get you on <a href="http://store.steampowered.com/stats/" target="_blank">Steam's Top 100 played games list</a>)
<!--quoteo(post=1964344:date=Aug 17 2012, 04:00 AM:name=Dghelneshi)--><div class='quotetop'>QUOTE (Dghelneshi @ Aug 17 2012, 04:00 AM) <a href="index.php?act=findpost&pid=1964344"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->ELO ratings don't even seem to work for a game as simple as Quake (everything below ~2200 Rating is essentially random from my experience), so unfortunately I can't see it working at all in NS2.<!--QuoteEnd--></div><!--QuoteEEnd-->
I'm a bit curious why this didn't work so well (got any links on how they calculated the Elo ratings)? As far as I can tell, SC2 uses a modified form of Elo ratings with good success. I've figured that NS2 could probably use something similar.
<!--quoteo(post=1964453:date=Aug 17 2012, 10:18 AM:name=Hunter.S.T.)--><div class='quotetop'>QUOTE (Hunter.S.T. @ Aug 17 2012, 10:18 AM) <a href="index.php?act=findpost&pid=1964453"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I tell you one thing the stats site is good for, i was gettin trolled the other day by someone who is angry at the nexzil clan, and by looking at who i thought it was on the ns2stats site, i was able to to see that the trolls name was the only other alias for that person i suspected.<!--QuoteEnd--></div><!--QuoteEEnd-->
I must admit, the thing I like about the site is that its an effective smurf-name detector :)
I've been thinking about it and I'm interested to see if a win/loss system does work. Perhaps I'm just too attached to old ways which I've seen used before. It's no fun too be on the opposite side of a heavily stacked team, so if you could prevent that by quickly randomizing the teams based on rank then it could be awesome.
I guess I sort of feel like it's a bit harsh to lose points for something outside your control, like if you have a bad com, or if your team mates suck etc.
Once you get to higher player numbers per team, it is effectively random until every player has played at least a few hundred games in total since there are too many variables for a team game to level out in a short amount of time. In NS2 it's even worse because there are no set goals for a player to achieve. In Quake you do can just count damage done per time and make the first third move up and the last third move down in your system (this is how the official system works; QLRanks.com seems to use wins/losses alone, which can be pretty stupid when you lose 160 ELO points from a single 9-10 game), but in NS2 there are just so many things you can do, which you can either not influence yourself or not quantify properly (especially the relations between those actions, i.e. how much a building kill is worth in relation to a player kill).
Additionally, there are even completely different roles like the commanders or Gorges. It's pretty much impossible to quantify commander performance other than by win percentage, which will be pretty random until you've played a lot of games with all the different pub players.
I've been thinking about it and I'm interested to see if a win/loss system does work. Perhaps I'm just too attached to old ways which I've seen used before. It's no fun too be on the opposite side of a heavily stacked team, so if you could prevent that by quickly randomizing the teams based on rank then it could be awesome.
I guess I sort of feel like it's a bit harsh to lose points for something outside your control, like if you have a bad com, or if your team mates suck etc.<!--QuoteEnd--></div><!--QuoteEEnd-->
SC2 does its ranking/rating system for up to 4v4's afaik. Also, it sucks that you lose a game, let alone Elo points, for a bad comm. That's just something you have to accept when playing a team-based online multiplayer game.
<!--quoteo(post=1964729:date=Aug 18 2012, 10:27 AM:name=Dghelneshi)--><div class='quotetop'>QUOTE (Dghelneshi @ Aug 18 2012, 10:27 AM) <a href="index.php?act=findpost&pid=1964729"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->ELO for 1v1 games/game modes mostly works decently.
Once you get to higher player numbers per team, it is effectively random until every player has played at least a few hundred games in total since there are too many variables for a team game to level out in a short amount of time. In NS2 it's even worse because there are no set goals for a player to achieve. In Quake you do can just count damage done per time and make the first third move up and the last third move down in your system (this is how the official system works; QLRanks.com seems to use wins/losses alone, which can be pretty stupid when you lose 160 ELO points from a single 9-10 game), but in NS2 there are just so many things you can do, which you can either not influence yourself or not quantify properly (especially the relations between those actions, i.e. how much a building kill is worth in relation to a player kill).
Additionally, there are even completely different roles like the commanders or Gorges. It's pretty much impossible to quantify commander performance other than by win percentage, which will be pretty random until you've played a lot of games with all the different pub players.<!--QuoteEnd--></div><!--QuoteEEnd-->
This is why I think it should stick to wins/losses only, since that is the ultimate goal of NS2 matches, with the total points distributed or taken evenly from each player on the winning/losing team.
I'm not saying the system is perfect, but I do think it'll do a better job of making more evenly matched games than the current system (which is really no system). At a minimum, it would reduce the number of (to borrow SC2-wording) platinum players against bronze players and vice versa.
#1 is important since it is a good indicator of fps skill (or having an aimbot)
#2 is important since it somewhat shows good game sense/ability to protect your investments
#3 is important since it's the goal of the game
#4 is important so that people don't get credit for standing around while the team carries them.
This is why I think it should stick to wins/losses only, since that is the ultimate goal of NS2 matches, with the total points distributed or taken evenly from each player on the winning/losing team.
I'm not saying the system is perfect, but I do think it'll do a better job of making more evenly matched games than the current system (which is really no system). At a minimum, it would reduce the number of (to borrow SC2-wording) platinum players against bronze players and vice versa.<!--QuoteEnd--></div><!--QuoteEEnd-->
With potentially 16 players per team, win/loss ratios lose (almost) all meaning for the individual. At best you'll see a trend for how much a player influences the outcome after several hundred played games. I play another game with 15 players per team that tracks stats, and there the difference in wins between terrible and exceptional players is not more than 15%.
It's incredibly hard to rate individual skill in team games. I can say from experience that systems like Elo and TrueSkill become problematic at 3vs3 games already.
Care to share what game that is? I'd be interested in seeing the stats. Also, if it really doesn't work as you say it does then it might give the makers of ns2stats some valuable information before they spend time working on the ranking system.
I do believe you could still have fairly balanced games by just using the k:d elo ranking. I feel like before thinking about auto balancing of teams you should just experiment a little and see how each method works. You already record the k:d so it shouldn't be too hard to add or take away elo points for each one and see how it plays out. Obviously gorges and coms won't be ranked properly in this system but you'll probably need to do some sort of separate stats for them anyway. Having a small section with "top 10 gorges" and "top 10 coms" that are based on different stats from the main rankings.
just had my first match as marine commander, stats show the game but not me commanding. we had two more commanders (things were pretty chaotic) until i took over a second time and stayed there for the rest of the match (pretty sure it was longer than half the match-time, although i left the chair to defend the base a few times, maybe that could be the reason?)
what it does show (which it did not before), is alien command stats which seem quite off: i did not play as alien commander on the stat servers yet, but it shows first hive upgrades: 25x crag hive, 3x shift hive and 1x shade hive. that's more than win+loss recorded matches i got...
the time played stats are 5s not-commanding for marines (and no commanding) and more than 4 hours of alien not-commanding (and no commanding stat). there are neither wins nor losses displayed.
here is the match:
<a href="http://ns2stats.org/round/round/1426" target="_blank">http://ns2stats.org/round/round/1426</a>
I also added a round length pie diagram to the <a href="http://ns2stats.org/map/map/1" target="_blank">maps</a>. I will be adding more of these diagrams soon.
I also worked on some other bigger features but they're not ready yet.
Thanks for the bug report Laosh'Ra. I will take look into it soon. Probably need to get some sleep first. :)
Interesting. We've got the log from your game that didn't show up on the web page but did show up in our dump site. It stopped adding log lines at some point even though the log wasn't even near its limits, we have to see what's up with this.
just one more thing: it shows crag hive as first upgrade although it also says i've never been alien commander. maybe i was, but not long enough for the game to count it as commanding? (i don't remember tbh)
just one more thing: it shows crag hive as first upgrade although it also says i've never been alien commander. maybe i was, but not long enough for the game to count it as commanding? (i don't remember tbh)<!--QuoteEnd--></div><!--QuoteEEnd-->
Yeah i fixed the marine commander stats last night. Thanks for reporting the bugs.
The alien commander first hive upgrade stats are still not working quite right. Currently it shows stats from rounds where you never even entered the hive. I will fix that as soon as i can. That is hopefully this evening.