Instruments of War
phpbb2 integraMOD
     
Go to top
Home • Forum • Check your PM • Register • Log in
Go to top
  Official site for the Enemy Territory clan Instruments of War
Calendar 
         
Navigation
Go to the homepageHome
Index of the forum.Forum
Listing of the current members of the IoW clan.IoW Members
Our good friends.IoW Regulars
Login to your iowclan.net accountE-Mail

Downloads
Topics that are top rated by forum usersLatest Ratings
See who is online at this time.Online Users
Read the rules of this websiteRules
Members of our staffStaff
View website statisticsStatistics

Frequently Asked Questions.FAQ
Search through forums.Search Forums

Donate to Instruments of WarDonate
Manage your own cookiesMy Cookies

<IoW> Server Rules

Who is Online

In total there are 61 users online ::
0 Registered,
61 Guests

Registered Users: None

[ View complete list ]


User Block
Username:

Password:

 Remember me



I forgot my password

Don't have an account yet?
You can register for FREE


Discord
Arrow IoW Discord Server

Server Monitor

Searching

Advanced Search

Style Select
Script by webmedic, v1.0.5

security

 phpBB Security ©  Has Blocked 2,346 Exploit Attempts.


Shuffle
 
View previous topic View printer-friendly version Search Display number of posts for each poster in this topic Export topic thread to a text file View next topic
Author Message
bearcat*
Major
Major
9538 Points

Canada CA British Columbia
PostYou have posted in this forum: Mon Jun 27, 2016 5:10 pm   Post subject:  Shuffle Back to top 

Last night I notice a few shuffles and it made teams even more off before the shuffle was called. I heard the shuffle was based on PWR instead of K/R. One shuffle put Allies up by 2 and with a lower K/R and Axis still rolled with a K/R advantage of just over 2.

_________________
Never trust a scarecrow wearing shades after dark
Download Post  No rating  
Sin
Admin
Admin
87019 Points

USA US Michigan
PostYou have posted in this forum: Mon Jun 27, 2016 5:56 pm   Post subject:   Back to top 

The problem that we've had is that people that look like they have an average K/D are actually really good shots, but they have average K/D because they are objective oriented, which means lots of deaths.

Look at BECK, he has an average K/D but is a really good shot. He only goes for objective and usually gets a lot of kills doing so, but also a lot of deaths. Get 2 or 3 of those kinds of players on the same team, and it usually results in lopsided teams. While BECK has a average K/D, he wins a lot of maps, thus increasing his PRW, which doesn't take into account how many kills or deaths he had, simply his will to win the map.

There are many players that fall into the category that BECK does. And there is many that are the opposite. Ones with high K/D but very low PRW. Namely, the worthless rambo meds that just go after kills. Get 2 or 3 of these on your team, and good luck winning any map.

Bottom line, I still believe that PRW shuffle is the best. I'd rather have each team have the same amount of players that care about playing the obj than have all the rambos end up on the same team and vice versa. No, it isn't perfect, but it's probably the best solution other than manually moving people from team to team, which usually causes people to just leave.

And honestly, I dont know what we have shuffle set to. Im just defending the reason why I think PRW would be the best to have it set on.

_________________
Download Post  Rating: Acknowledged  
Venom
Senior Admin
Senior Admin
104308 Points

USA US California
PostYou have posted in this forum: Mon Jun 27, 2016 9:48 pm   Post subject:   Back to top 

Couldn't have said it any better Sin

_________________
Download Post  No rating  
BECK
General of the Army
General of the Army
279757 Points

USA US Colorado
PostYou have posted in this forum: Mon Jun 27, 2016 10:42 pm   Post subject:   Back to top 

Nicely done.

In general, a shuffle isn't supposed to be some magic formula for perfect teams. I think a shake up of players is all that is needed to create a new dynamic for the next few maps. As Sin said, PRW rates how valuable a player is to the team, and on an objective oriented server like ours, it is probably a better basis for restructuring than measuring kills.

_________________
Very Happy
Download Post  No rating  
Anto*
Staff Sergeant
Staff Sergeant
3327 Points

Argentina
PostYou have posted in this forum: Mon Jun 27, 2016 11:55 pm   Post subject:  Re: Shuffle Back to top 

Like Sin said, I've noticed that good players don't have the proper K/D they should have just because they are more objective oriented players.
I think the current shuffle is based on K.R, I'm not sure but after some shuffles that's what it looks like. Maybe if it were based on PWR the shuffle would be more fair.

_________________
Download Post  No rating  
BECK
General of the Army
General of the Army
279757 Points

USA US Colorado
PostYou have posted in this forum: Tue Jun 28, 2016 9:39 am   Post subject:   Back to top 

Here are our settings, complete with the silEnT documentation that accompanies them.

Code: ›
set g_shuffle_rating "4"
            // Sets the rating system used by shuffle.
           
        set g_killRating "3"
            // When not set to 0 silEnT will track a player's killing ability using an ELO type
            // statistic similar to chess. Each kill will increase a player's kill rating based on
            // how hard the kill was. Killing unskilled players will result in very few points
            // whereas an unskilled player killing a skilled player will result in more points. Kill
            // rating also takes into account the weapon used (less for arty, more for knife).
            //
            // See also g_shuffle_rating.
           
        set g_playerRating "3"
            // When not set to non zero positive value, silEnT will track a player's ability to win
            // against good teams using an ELO type statistic similar to chess. Each win will
            // increase a player's rating based on how hard the opposing team was. Skilled teams
            // defeating less skilled teams will result in very few points whereas an relatively
            // low- rated team defeating a good team will result in more points. Player rating also
            // takes into account the number of players on each team (less points when a very large
            // defeats a small one) and which side tends to win more often on the current map (e.g.
            // Allies tend to win tc_base).
            //
            // See also g_shuffle_rating.



silEnT docs:
Spoilers: 

g_shuffle_rating
Description:
Sets the rating system used by shuffle.
Type: [ 1 | 2 | 3 | 4 | 5 ]
Parameters:

1 Use player XP
2 Use the rate at which players have gained XP since connecting.
3 Use the killRating (g_killRating must be nonzero)
4 Use the playerRating (g_playerRating must be nonzero)
5 Use the playerRating (g_playerRating must be nonzero), but take the map into account and give the teams closest to 50/50
Default: 3

g_killRating
Description:
When not set to 0, silEnT will track a player's killing ability using an ELO type statistic similar to chess. Each kill will increase a player's kill rating based on how hard the kill was. Killing unskilled players will result in very few points whereas an unskilled player killing a skilled player will result in more points.

See also g_shuffle_rating.

Type: bitmask
Parameters:

1 Use kill rating. A dummy flag since any non-zero value for g_killRating will enable it.
2 Make kill rating visible. Players can use /killrating and at the end of each map a message will display the top killers for the map and overall.
4 Log files will include the GUIDs of the players involved in kills. This makes kill tracking a lot easier.
8 Disable kill rating from the score board. This can prevent players from playing for higher kill rating alone. This does not prevent the "killrating" client command enabled with flag 2 from working.


Default: 3

g_playerRating
Description:
When not set to non zero positive value, silEnT will track a player's ability to win against good teams using an ELO type statistic similar to chess. Each win will increase a player's rating based on how hard the opposing team was. Skilled teams defeating less skilled teams will result in very few points whereas an relatively low- rated team defeating a good team will result in more points. Player rating also takes into account the number of players on each team (less points when a very large defeats a small one) and which side tends to win more often on the current map (e.g. Allies tend to win tc_base).

See also g_shuffle_rating.

Type: bitmask
Parameters:

1 Use player rating. A dummy flag since any non-zero value for g_playerRating will enable it.
2 Make player rating visible. Players can use /playerrating and at the end of each map a message will display the top player.
4 Put player rating on the scoreboard instead of the XP.
8 Print out extra info at the end of a round that can be used to further refine the playerrating model.
16 The same as flag 8 except a lot more info is output.
Default: 3


So our settings already use player rating for shuffle. We've had it that way for years and honestly we will probably not change it now. I'm not sure what flag 5 means, and how it would change how it arranges teams, but I do know that every map we play is stored in a database with measures that effect this, for example:

Code: ›
[serverstat]
rating            = -0.109171
rating_variance   = 0.000413
distance_rating   = 129.424225
distance_variance = 0.021879

[mapstat]
name              = battery
rating            = 2.299107
rating_variance   = 0.140633
spree_record      = 34
spree_name        = ^0Mike^7Slayer


That is the real data - all of this information is unique to our server and our history of how maps are played and won - so these factors (there are factors for every single map and the server as a whole) probably weigh into this as well. Perhaps if we actually used setting '5' instead of '4' for our shuffle settings, it might factor the map data into the shuffle better and could actually produce better teams.

Also note that the other mode shuffles on kill rating - not kill/death ratio. This is different, as is explained in the docs. People are often confused by this and wonder why their "K/D" is wrong. Well, it's not "K/D" that you are actually looking at. I also think that the map and server factors above affect how kill rating is calculated, because I'm often told by people that our settings are weird compared to other servers. I guess that is why relatively new players to our server like you guys are noticing this stuff Wink

_________________
Very Happy
Download Post  No rating  
sum1else
Brigadier General
Brigadier General
67080 Points

USA US Ohio
PostYou have posted in this forum: Tue Jun 28, 2016 5:03 pm   Post subject:  Re: Shuffle Back to top 

I think changing it to 5 might be worth a shot. It doesn't sound like it would harm anything too much in any case. Are we set at the default 3 for killrating and playerrating?

Download Post  No rating  
BECK
General of the Army
General of the Army
279757 Points

USA US Colorado
PostYou have posted in this forum: Tue Jun 28, 2016 8:51 pm   Post subject:   Back to top 

Might be worth considering.

We'll adjourn to our ivory tower to discuss.

_________________
Very Happy
Download Post  No rating  
Sgt. Hobbs
Retired
Retired
32504 Points

USA US New Hampshire
PostYou have posted in this forum: Wed Jun 29, 2016 7:00 pm   Post subject:  Re: Shuffle Back to top 

better question is how many "New" players are you still getting. (Might be time for a new who's contributing and who's Rambo move and adjust freely rather than random mix up based on a math equation)

_________________
I reject your reality and substitute my own.
Download Post  No rating  
Display posts from previous:      
 


 Jump to:   



  View previous topic View printer-friendly version Search Display number of posts for each poster in this topic Export topic thread to a text file View next topic

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
You cannot attach files in this forum
You can download files in this forum

Lo-Fi Version
Protected by CBACK CrackerTracker :: 1190544 Attacks blocked.
Protected by phpBB Security © phpBB-Amod :: phpBB Security ©  Has Blocked 2,346 Exploit Attempts.
 @ 2007 The Integra Team
 @ 2007 phpBB Group
:: FI Theme :: All times are GMT - 5 Hours ::
[Page generation time: 0.6348s (PHP: 63% | SQL: 37%) | SQL queries: 83 | GZIP disabled | Debug on]