It can't be solved, that's what i'm saying :<

I'm not quite clear on what some of the suggestions are looking for. I'm guessing you mean the
overall rank should be seperated out. So for example:

Under a players profile. for example BuLaT
http://stats.quakeworld.nu/index.php?a=qwplayer&currentPlayer=BuLaT
If you click on the Ranks tab you can see all of his individual ranks.

You see a breakdown of his rank for each of the modes. Most of them make sense.
The overall ranking contains all matches you have played. So im guessing that fog and rikroll are saying that your overall rank should be scrapped and and I should use these seperate ranks and then add new ranks for 4on4 and 2on2 ???
So when you view the entire players list you should be able to sort by 4on4 rank or a 1on1 rank and it will filter out players that dont play 4on4 etc etc? I think this makes sense to me. The Overall Ranking does seem to get scewed if you play a game mode and are not good at it and your rank for that mode goes below 1.



There was a system in place (based on IP addresses) before to address this issue but that is not within the scope of the stats site (at the moment anyway). There was a site that allowed you detect fake nickers (and dodgy bras ) but its no longer online. If this site was running it would sort out that issue.

This is a problem that exists regardless of the stats site. People will fake nick on servers
regardless and to be honest as is stated on the 'About' section on the website, the real test of
a players rank and performace will be how they perform in Competitions, ladders and tournaments. Not based on the site. The site wont ever give an accurate or decent rank as its based on subjective analysis and can also be fooled as we dont have a proper account system. So its only there to give an indication and can be handy way of getting a rough players skill level. This was one of the reasons I stayed away from ranking for 4on4 as it would be even less accurate. But I can add in a rank if some people would like it.

The argument was made when the site first went live that by having a stats which has ranks would
turn people of pracccing. I think in fairness this is not really the case. People wont take a rank seriously on the stats site anway and every one knows that its Owneage or EQL that will determine the real top dogs. But I have offered to remove the rankings before and replace with a different system which would help people determine a similiar skill level player. But need suggestions on this ?


I have a plan for that but thats in the pipeline.


At the moment in fairness these players are excluded from the rankings. With the exception for 'unnamed' as this was a nick used by a div 1 player and was left in purposely

Also as Ake Vader says

That is true. It would be good to improve it from what it is though for those that do like to look at the ranks. There was talks of a login system and all that but I dont see that ever getting agreement anyway.

I'd like to see the Date field use 24hr clock instead of 12hr clock. Currently it is highly misleading because it doesn't even indicate am/pm. e.g. the below two games that I played back to back yet appear to be 11 hours 37 minutes apart

30-09-2010 01:03 dm2 mix 209 179 tks View
30-09-2010 12:40 dm3 mix 136 223 tks View

Yes hangtime, that was one of the things I forgot... 24h clock ffs!

Medar:

Individual stats:
- Last match info is almost never what I care for, but it takes most of the page all the time -Complete

4on4 (2on2) scoreboard:
- Sort players withing their teams by frags - Complete
- Show the winning team - Complete
- Disconnected players are not shown at all - Can't be fixed
- Something to show on which team the player was in 4on4/2on2 would be useful - Player will always be on the right hand side.

HangTime:
Matches are now in 24 hour clock. - Complete

Zalon:
When looking at a player with awards, add a title to the award image - Complete
When searching, maybe make it always include wildcards. - Complete



More to follow:

Been looking at the ranks and should mention the following:

1) The ranking system takes into account if you play a weaker player.
2) It rewards high activity. So the more you play the higher your ranking will be.
3) Your rank will converge to an arbitrary value of 400 (I think) over time. This formula can be
set to have this convergence occur slowly or quickly.
4) For the Overall Rank if you are inactive for a week or a month, your ranking is reduced. Then
after 3 months of inactivity your rank is set back to one. This explains how some players like
Milton have a low Overall Rank but a high individual Map Rank. The Individual Map rank doesnt ever get reduced due to inactivity.
5) The Overall Rank for each player is calculated based on TB5 only.
6) The amount by which you lost a game is taken into account. So if we look at two examples (as mentioned above):

Example 1:

[tVS]Ihminen Pov Matches
Matches Won: 1699 Matches Lost: 92 Matches Drawn: 3

.urban Pov Matches
Matches Won: 1885 Matches Lost: 756 Matches Drawn: 14

[tVS]Ihminen has a Pov Rank of 119.319
.urban has a Pov Rank of 135.937

You would imagine that [tVS]Ihminen should be way ahead based on win/loss ratio. However the reason .urban has a higher rank is that he has played 2641 games (as against 1794 matches for [tVS]Ihminen ). So he is firstly rewarded for higher activity. For the 756 games he has lost, the amount by which he lost by is taken into account. So by the nature of Pov games they are usually very close. So therefore he will be gaining in rank even for games he has lost. Whereas in a lot of duel matches there can be a bigger loss margin and hence usually a lost game would not help you gain in rank. On one hand this can seem unfair. However if you lose against a high rank person then you should get some credit. So basically he is being rewared on two counts. A) Activity B) Closeness of defeats. You will also find that if [tVS]Ihminen had played the same number of games as .urban and with the same win/ratio he currently has. Then his ranking would clearly surpass .urban's.

Also the second example is for say [tVS]Milton. His overall ranking is only 4.9323 whereas his DM4 ranking is 57.9373. In this situation his Overall ranking is only at 4.9 as it would have been reset during the summer when he was inactive. If you are inactive for a period of 3 months your ranking gets reset. All the other rankings currently stay the same and are not reduced due to inactivity.

So anyway I'm not sure that the ranking is really all that screwed up. It rewards activity and takes into account the margin by which you win/lose a game. If it were to just take into account a win or loss then you would see [tVS]Ihminen and [tVS]Milton ahead of others. Maybe that would be a better ranking but then that system could lead to anomolies as well. Because certain sections of players may never play each other in the small Community so Its fair to say that no ranking system can give an overall caccurate picture.

Does that make any sense ?

Anyway in my previous post I've updated bugs that have been fixed now. I will be adding a few nice features in the next few weeks as well. So I want to have the ranking thing tied down a bit though so any suggestion on a better system would be great.

Something is really wrong with logging the frags in some games, I have noticed at least two games that display my frags wrong, and _always_ to my deficit!

http://stats.quakeworld.nu/index.php?a=4on4_ScoreBoard&matchId=16601
qw.foppa.dk - Here I scored something like 95-98 frags, not 77

http://stats.quakeworld.nu/index.php?a=ffa_ScoreBoard&matchId=131683
xs4all - Here I won the FFA with 150 frags and remember quite vividly how unnamed "whined" a bit at losing with one frag when it was such a tight round. So in addition to losing the win and frag, I actually tie for first place and end up getting the Silver medal?

I didn't drop in any of these games and they were played on different servers.

I don't really care about the medals and winning rounds that much, I just care that the stats are accurate and I don't see anything in these games that would be linked to the inherent limitations in getting the stats? Also these are just two games I happened to check out, how many more games are screwed up?

hey niomic

Thanks for spotting that. The first issue is one I introduced recently when I changed the scores to
"Sort players withing their teams by frags" The match score is in the datsbase correctly however it is just sorted and then displayed incorrectly.

16601 1288308204 89.104.194.145 27501 qw.foppa.dk #1 - ktx dm2 20 353 fu-buggy 77 -fu- meshuggah 77 -fu- niomic 96 -fu- fu-kassim 103 -fu- 92 ddk 15 tks PolloX 18 tks HangTime 29 tks Medar 30 tks 43 2010 29

I will fix that later today. So that it displays correctly.

The second issue unfortunately is that the FFA bot only scans the server at regular intervals. So not every single frag will be detected. As the servers don't report in scores we have to scan the FFA servers every few seconds. Generally the score is sccurate but there will be an occasional frag lost unfortunately. The other bot which handles the 1on1, 2on2 and 4on4 matches does get the actual score as it detects the countdown to the end of the game and then takes the end of game screenshot.

Can look into improving the FFA bot.

Thanks for the quick reply. At least xs4all has a countdown for when the map ends, it's not the same as the newer countdown, but couldn't that be detected?

You can see that countdown because you are a player, that is you are connected to the server and you are receiving 77 packets per second, some of those contain the countdown text.

The countdown on KTX servers is also visible by querying the server with a special packet. That way we don't need to connect to the server as a (pseudo) player, we just send one small packet and reive the necessary info. Then we calculate when we need to ask the server in "the right time" so that we catch scores on final scoreboard.

Neither the FTEQW server nor the FFAMOD provide any info about when the map is going to end, nor how to get results from the last match. And also there is the next_map thing. You simply have to ask the server instantly about current scores and if suddenly all scores are zero and map has changed, you know the last ones you got are closest to the final scoreboard. At the moment the bot asks every 15 seconds.

hey guys, alot of servers are not correct in the "game views" like this:
http://stats.quakeworld.nu/index.php?a=4on4_ScoreBoard&matchId=16596

the info is:
Server Name: The Netherworld
Address: 127.0.0.1:27500

when it should be the foppa servers took me a while until I found the damn demos

That is a known issue. The bot sends in the IP address/Port of where the match was. There is another bot that does the lookup and gets the rules and name for the server. Need to set it running again and enure that it updates tables. The 127.0.0.1 is a default in case the bot is not running and has not looked up a server.

Good to see you back btw

This has been fixed now.

Thanks pleura

What part, his fragcount... or the bug?

The first part. I.e the frag count, The second part can't be fixed for now unless we have a server change. His frag count was a bug in the sorting and displaying

Stats site do not record any information about matches with more than 4 players in each team.
Last night few 7on7 and 10on10 matches were played on playground.ru , but there are no any information about it on stats website.

Hey B1aze

Thanks for that. Will look into it and see if we can do something about it. We dont support it at the moment for various reasons but could try and trow something together. Pity to miss out those games.

One bug that we noticed lately is that when you are stripping away certain coloring codes in team names for example, you end up with a game like this:

http://stats.quakeworld.nu/index.php?a=4on4_ScoreBoard&matchId=18751

Teams were actually as below and of course the score is a lot tighter as well. Hopefully you don't strip the names too early so that you can't do team matching :/?
para vs murdoc
strik vs niomic
krab vs moje
lakso vs phren

Arhhh I feel like i'm going the wrong way up an escalator... one step forward and two steps back :-)
Will have a look today. Thanks for posting niomic.

Won't fix.

The bot strips Quake colors and therefore it cannot distinguish between ]sr[ and ]sr[ in different Quake colors. The site is then probably mistaken when it sees 8 players (all in one team) and tries to somehow split them into two teams.

Simply use team names that differ in letters, not only in colors. I'm sure even commentators would appreciate that. "I think SR are going to win." "Which SR?" "The orange ones" "Orange brackets or letters?" "Letters ... no wait"

I will add in a warning for matches like that. So a disclaimer like

B1aze: I plan to make changes soon enough to handle larger matches.

Commentators would call them SR2 as that is their name, It might not be their tag... But officially they are called Slackers2. I can see the problem for outside spectators watching a game with commentators tho.

But why not sort the teams before stripping the colors?

Every spectator is "outside". People who actually watch the QW scene 24/7 like you do, know "who is who" make only a tiny minority of the scene.

Player comparison needs some work. Right now when you even go to that section, it immediately starts loading megabytes of data, it shouldn't probably do so. Also, after the loading has finished, accessing either player name box (by clicking) will cause Opera crash most of the time. Also the comparison didn't work last time at all, maybe I need to check it now if it does, but getting there could prove difficult due to crashing.

Oops I thought I disabled that. That section never worked fully. Its been part of the plan to be rewritten.

Still didn't work.
Maybe it is a good idea to put all unrecognized matches to unsorted group?
Sometimes we played CTF or team arena and it will be nice to see those matches somewhere.

I haven't read the whole thread but I've noticed that when you spec an ffa game, you'll be displayed as a player on stats.quakeworld.nu match results page with 0 frags. I think spectators should be ignored in calculating match stats. I can imagine how this also changes the spectators general stats, which is stupid.

Sounds entirely correct to me, it rewards activity!1!1!1!!!

Also, sometimes a server address is just 127.0.0.1:27500.
Example: http://stats.quakeworld.nu/index.php?a=1on1_ScoreBoard&matchId=216364