Belma site and statistics: ideas and suggestions needed

Feature requests and ideas for the new unofficial versions of Elma and general talk related to those.

Moderator: Moporators

User avatar
Zweq
34mins club
Posts: 4055
Joined: 28 Nov 2002, 15:54
Location: suo mesta

Re: Belma site and statistics: ideas and suggestions needed

Post by Zweq »

14,4k levs there. must be like 1% of all time only
Image
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

There has been some talk about the ranking system, and I think it could be improved a bit aswell. So lets discuss it :)

The way the main system works now, is that you get a certain amount of points for each player you beat in a battle, this amount of points is calculated according to the difference between you and the beated players ranking. So you get a big amount for those with higher ranking than you and a smaller amount for those with lower ranking than you, but you still get some for beating those. The ranking is "eternal" meaning you can never lose points and it will inflate.

One improvement I thought of is that older results slowly fade away, so that a battle from a year ago doesn't count as much anymore. I got this idea from the FIFA rankings, which uses a system like this:

Within the last 12 months x 1.0
12–24 months ago x 0.5
24–36 months ago x 0.3
36–48 months ago x 0.2

Maybe it should be on a smaller time scale, like half of this. Like this it will also be possible for someone who start playing in 3 years to get to the top, and someone who stops playing won't stay on top forever.

Another improvement I've thought of, though harder to implement, is that only a certain amount of battles a year counts. This inspired by Tennis rankings where only a certain amount of tournaments a year counts. This way it will be possible for someone who doesn't play 24/7 but is really good to still be number 1. So a number like 350 then it's enough to play a battle a day, but it will still encourage you to play more, beause you can improve one of your bad results with a better one.

I'm not really sure if that kinda fucks it up, because suddenly a result from earlier doesn't count anymore, and that has affected your ranking, which in turn affected the amount of points you got in all the battles.
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Belma site and statistics: ideas and suggestions needed

Post by Lukazz »

i like the first idea very much, and don't really understand the second one. so you mean that at the end of the day only the best battle result counts?
what about cups btw? imo they could be counted for rankings too, like the next master cup for example.
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
Madness
35mins club
Posts: 2168
Joined: 1 Jan 2009, 10:51
Location: UK

Re: Belma site and statistics: ideas and suggestions needed

Post by Madness »

The first idea is the most acceptable.
Image
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

Lukazz wrote:i like the first idea very much, and don't really understand the second one. so you mean that at the end of the day only the best battle result counts?
what about cups btw? imo they could be counted for rankings too, like the next master cup for example.
Second idea is like this: When you have played 350 battles and you play one more, it takes away your worst results, play one more and takes away worst results there's back ect. ect.

I don't think cups should count, cups are cups, battles are battles.
User avatar
Lousku
Kuski
Posts: 2925
Joined: 5 Feb 2010, 00:25
Team: BAP
Location: expensive land of dads

Re: Belma site and statistics: ideas and suggestions needed

Post by Lousku »

That second idea seems very nice to me. It's easier to understand, too.
then again i don't know anything
maybe easier not to think abouut alöl things thought than not things thought ... or something..=?
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Re: Belma site and statistics: ideas and suggestions needed

Post by 8-ball »

I would still go for a system where sucking ass would make you lose points but if most people oppose it, well, this ain't that bad.
39:37,91
User avatar
pawq
38mins club
Posts: 6547
Joined: 24 Aug 2008, 19:56
Team: TR
Location: Southampton, UK

Re: Belma site and statistics: ideas and suggestions needed

Post by pawq »

first idea is nice
but second - ez win 350 balles in a year... or even 1000 in eternity. so this would have to be worked on very hard to be made proper
User avatar
Bismuth
Kuski
Posts: 2777
Joined: 26 Feb 2008, 07:11
Team: <3
Location: Not on lauta
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Bismuth »

Xarthok wrote:I would still go for a system where sucking ass would make you lose points
I would be -9001

I would go for the FIFA one, fading over time.
:mrhat:
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

Pawq wrote:first idea is nice
but second - ez win 350 balles in a year... or even 1000 in eternity. so this would have to be worked on very hard to be made proper
But a win doesn't necessarily mean you get a lot of points, a third place with different opponents might give more points. If you win one a day with all the best players present you probably deserve the first place. But it should probably be more than 350 a year if we do that idea.
User avatar
NoZKeY
Kuski
Posts: 220
Joined: 13 Sep 2008, 23:53

Re: Belma site and statistics: ideas and suggestions needed

Post by NoZKeY »

The first idea is perfect i think.
And i don't understand the second one at all.. :?
User avatar
Lousku
Kuski
Posts: 2925
Joined: 5 Feb 2010, 00:25
Team: BAP
Location: expensive land of dads

Re: Belma site and statistics: ideas and suggestions needed

Post by Lousku »

<captain="obvious">

The 2nd idea means that within some time period only x amount of your battle results count, the ones that give you most ranking points. Let's shrink it to a smaller scale and say that only 10 battles per week count in your ranking. If you play 20 battles that week, your worst 10 results won't count.

</captain>

This gives you a chance to höylätä your ranking within the time period. After you've reached the number of battles that count, you can only get a better ranking. It's also easier to understand what really makes up your score, whereas the 1st idea just leaves you shrugging IMO.

btw, why is this discussion in a belma topic? :o
then again i don't know anything
maybe easier not to think abouut alöl things thought than not things thought ... or something..=?
User avatar
NoZKeY
Kuski
Posts: 220
Joined: 13 Sep 2008, 23:53

Re: Belma site and statistics: ideas and suggestions needed

Post by NoZKeY »

LousQ wrote:btw, why is this discussion in a belma topic?
megalol I just noticed that :mrgreen:

Thanks for the explanation Louskyu, forgive me for my dumbness :P
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

LousQ wrote:btw, why is this discussion in a belma topic? :o
This is the topic were all ranking discussion has been taking place, so I thought it would fit (:
User avatar
pawq
38mins club
Posts: 6547
Joined: 24 Aug 2008, 19:56
Team: TR
Location: Southampton, UK

Re: Belma site and statistics: ideas and suggestions needed

Post by pawq »

DuMBeY =DD
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Belma site and statistics: ideas and suggestions needed

Post by Lukazz »

Kopaka wrote:Another improvement I've thought of, though harder to implement, is that only a certain amount of battles a year counts. This inspired by Tennis rankings where only a certain amount of tournaments a year counts. This way it will be possible for someone who doesn't play 24/7 but is really good to still be number 1. So a number like 350 then it's enough to play a battle a day, but it will still encourage you to play more, beause you can improve one of your bad results with a better one.
pls do this. it's annoying that you can reach a top position by just playing 24/7.
TT: 36:59:53 || Avg TT: 38:09:65
veezay
Kuski
Posts: 144
Joined: 15 Jun 2003, 14:13
Team: rotaplask
Location: Kouvola, Finland

Re: Belma site and statistics: ideas and suggestions needed

Post by veezay »

Kopaka wrote:One improvement I thought of is that older results slowly fade away, so that a battle from a year ago doesn't count as much anymore. Like this it will also be possible for someone who start playing in 3 years to get to the top, and someone who stops playing won't stay on top forever.

Another improvement I've thought of, though harder to implement, is that only a certain amount of battles a year counts. This inspired by Tennis rankings where only a certain amount of tournaments a year counts. This way it will be possible for someone who doesn't play 24/7 but is really good to still be number 1. So a number like 350 then it's enough to play a battle a day, but it will still encourage you to play more, beause you can improve one of your bad results with a better one.
it might be a good idea to combine these two ideas for a single one, but as hard as i try to think about all the different situations and how they should be handled, my brain shuts down.

if you implement the first idea (which isn't so bad), at least make the multiplier decrease smoothly over time, so that the first 6 months for example are counted 1.0x but after that it starts to fade until 0.1x after 4 years or somesuch. that way the importance of some older battles wouldn't suddenly go from 1x to 0.5x, changing my ranking noticeably each day when the oldest battles go over the multiplier time limits.

however i don't think it's a good idea to ignore the worst results after some magical number of battles played in a year or so. someone could just höyl like crazy so that he'd have at least some good results. then again if someone höyls like crazy, maybe he has deserved the ranking. or maybe not. he just plays a lot but might suck horribly. if the ranking list is an attempt to define one's skill level and represent it as a number, then höyling shouldn't affect it too much per se. of course if you höyl a lot but still suck horribly, after removing the worst results you'd still end up with having very sucky results. oh well.

also, i don't know if the first system alone is too fair for randomly inactive people, eventhough the timescales are pretty large. people who don't play for a year might not deserve to be in the top10, but i don't think they should be lowered too much if they happen to be just great battlers skillwise. maybe their skill has gotten better in that time and they actually should be higher in the list! </joke>

so umm yeah. some nifty combination of these two systems might work. taking into account both the activity and the amount of battles played and weighing each one according to the other. yes, this post helped you a lot.


btw the graphs on player ranking page are nice! i'm sach a graphlover.
Image
Prestigious member of 6.58 Tutor01 club
Prestigious member of 14.6x Tutor14 club

<ville_j> in addition to ""njoygin"""" footblay i ma """"""""enojying"" rap usic
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

veezay wrote: it might be a good idea to combine these two ideas for a single one, but as hard as i try to think about all the different situations and how they should be handled, my brain shuts down.
Yeah that was kinda my thought, to combine them, and my brain does the same, so I'm not really sure if it's completely crazy :)
veezay wrote:if you implement the first idea (which isn't so bad), at least make the multiplier decrease smoothly over time, so that the first 6 months for example are counted 1.0x but after that it starts to fade until 0.1x after 4 years or somesuch. that way the importance of some older battles wouldn't suddenly go from 1x to 0.5x, changing my ranking noticeably each day when the oldest battles go over the multiplier time limits.
As I see it, it will at most be one battle that goes over each "fadelimit" for each update of the ranking (if I check which battles are older than 31536000 seconds ect. at each ranking update). So it won't change the ranking a lot.
veezay wrote:however i don't think it's a good idea to ignore the worst results after some magical number of battles played in a year or so. someone could just höyl like crazy so that he'd have at least some good results. then again if someone höyls like crazy, maybe he has deserved the ranking. or maybe not. he just plays a lot but might suck horribly. if the ranking list is an attempt to define one's skill level and represent it as a number, then höyling shouldn't affect it too much per se. of course if you höyl a lot but still suck horribly, after removing the worst results you'd still end up with having very sucky results. oh well.
But if we don't do that, someone who hoyls a lot will still get a lot more points than someone who only plays maybe a couple battles a day.

Actually made this system with the combination last night, but some rounding problem or sach, so can't show just yet.
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

I realised that this would never work well with the current formula, because your own ranking at the time of batle matters, so you can't really compare the amount of points in one battle, with the amount you get in another. So I tried making a different formula, based on the same, but instead of counting the difference between your own ranking and the person you beat's ranking, I only use the person you beat's ranking, and makes this relative to the ranking points of number 1.

I do this by dividing number 1's ranking with 1000, and then divide the person you beat's ranking with that result. This means you get same amount as if the difference were 1000 points if you beat number 1, and beating others is relative to this. And at the end, instead of multiplying the results with your ranking, I simply add it. I think this should mean a more stable ranking system, in current you can jump from 50th to 2nd in day or so. And also the fading and that will work well with it.

You can see the results here http://elmaonline.moposite.com/?s=ranki ... ts=rankinc

So it would be nice if some ppl try to track their ranking on that link, play some battle and see what happens :)

Edit: Forgot to mention, I've set the limit to 100 battles a year for now, so we can see the effect.
User avatar
Ali
39mins club
Posts: 98
Joined: 23 Oct 2004, 19:21
Location: Czech repablick
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Ali »

I won't pretend I understand your formula but imo these standings look quite realistic considering how much ppl play and how good they are. Also good that starting point is 0. I will try to track this on some battles...
Czech Across Racing Team member --> http://www.homer.kbx.cz/cart
</signature>
Ves
Kuski
Posts: 232
Joined: 13 Mar 2005, 03:15
Location: Kiev, Ukraine

Re: Belma site and statistics: ideas and suggestions needed

Post by Ves »

shoutbox messaging has problems with encoding (check my private message to Elfy). if it designed to not work correctly with non-english characters just put some error message.

also would be nice to see some notification when the message arrives.
Through the externals to the skill.
Through the skill to the internals.
Through the internals to the divinity.
User avatar
Mawane
Kuski
Posts: 3299
Joined: 15 Apr 2007, 01:05
Team: SV
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Mawane »

Ves wrote:also would be nice to see some notification when the message arrives.
there is in top left box
Website || TT:41:45:64 || Team Image
Image[url=steam://friends/add/76561198025490048]Image[/url]
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Belma site and statistics: ideas and suggestions needed

Post by Lukazz »

hm, i still wonder about this new test-ranking.
i have 36 wins and i'm 11th in Battle Experience Poinkts Ranking
jon has 17 wins and is 14th in BEP-ranking.
and still i am 18th in the new ranking and jon is 14th.
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

Lukazz wrote:hm, i still wonder about this new test-ranking.
i have 36 wins and i'm 11th in Battle Experience Poinkts Ranking
jon has 17 wins and is 14th in BEP-ranking.
and still i am 18th in the new ranking and jon is 14th.
What you should remember is that the test ranking is counts the 100 best results, and wins and BEP is all battles
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Belma site and statistics: ideas and suggestions needed

Post by Lukazz »

oh i see.
TT: 36:59:53 || Avg TT: 38:09:65
VT
Kuski
Posts: 8
Joined: 18 Jun 2007, 20:15

Re: Belma site and statistics: ideas and suggestions needed

Post by VT »

some data get lost i think. i have less wins now und almost all my times for thor-levs get lost.
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

not really lost, it's just some time ago that I copied the data to the new server
Ves
Kuski
Posts: 232
Joined: 13 Mar 2005, 03:15
Location: Kiev, Ukraine

Re: Belma site and statistics: ideas and suggestions needed

Post by Ves »

is the eol site concept "every finished time appears on the site immediately" or there may be some variations?

for example, optional delayed records publication to not spoil internal world records and have "verified: no cheating, droven online" flag at the same time.
Through the externals to the skill.
Through the skill to the internals.
Through the internals to the divinity.
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

There's has been some talk about some option on site you can set, like "don't show my internal times newer than x days" or something like that. But no one has really said they wanted it, so I haven't made it a priority.
User avatar
Mawane
Kuski
Posts: 3299
Joined: 15 Apr 2007, 01:05
Team: SV
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Mawane »

Kopaka wrote:There's has been some talk about some option on site you can set, like "don't show my internal times newer than x days" or something like that. But no one has really said they wanted it, so I haven't made it a priority.
would be useful for cups too :)
Website || TT:41:45:64 || Team Image
Image[url=steam://friends/add/76561198025490048]Image[/url]
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

new ranking?

Post by Kopaka »

Igge wrote:I want a new (improved) ladder with cool arrows so you can see who's climbing up and who's falling behind.

Seriously though. A nice ladder that actually works would be coal (and it doesn't require new eol coding).
Since most ranking discussion has been going on in this thread I thought we should continue it here, so the other topic doesn't get cluttered with that stuff.

I agree that the current ranking system doesn't work as well as I had hoped it would. In short it works like this: It takes your 100 best results from every half year, last two years. So there's four batches of 100, = 400. But this means that if you play more than 100 battles every half year some/many won't count in any way, and even if you beat some good player in a battle you might not get points from that, if there was few players and therefor few points to gain from that specific battle.
(to be more detailed it also matter the current position in ranking of the ppl you beat, with no relation to your own position)

The reasoning behind the system has been things like these:
1. Shouldn't be possible to lose points as that would discurage to play.
2. It should be possible to be competitive even if you don't play 24/7.
3. It should matter how many and who you beat in a battle.

There's ofcourse also the relative ranking, which is working pretty well. You can pretty quickly see yourself improving ect. But ofcourse it doesn't live up to first reasoning, and it's pretty short term (all played battles counts, but you can quickly lose all your points or regain lost ones).

So yeah I don't know.
User avatar
Mats
39mins club
Posts: 1183
Joined: 5 Apr 2007, 12:30
Location: Norway, Sandnes

Re: Belma site and statistics: ideas and suggestions needed

Post by Mats »

Just a question: Does all kind of battles count?
TT:39.59.86|| AvgTT:41.49.24 || Multi TT:27:43:82 || Team [TR]
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Kopaka »

Each type has their own ranking (only normal shown in site atm though) and any sort of crippled does not count anywhere.
User avatar
analcactus
Kuski
Posts: 421
Joined: 7 Dec 2010, 12:54
Location: St. Petersburg, Russia
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by analcactus »

Kopaka wrote:Each type has their own ranking (only normal shown in site atm though) and any sort of crippled does not count anywhere.
that's what i was about to talk about, ok. it sucks cause anpalcactus has zero wins in statistics but ACTUALLY he does have 1 balle win ( in ultra-pro-one-turn-one-life-pob-balle )
http://elmaonline.net/?s=battles&b=4153

gimme my win in stats so my mum doesn't think i'm nab anymore
Image Image
Image
User avatar
Mats
39mins club
Posts: 1183
Joined: 5 Apr 2007, 12:30
Location: Norway, Sandnes

Re: Belma site and statistics: ideas and suggestions needed

Post by Mats »

Hahaha Now that's a quote XD
TT:39.59.86|| AvgTT:41.49.24 || Multi TT:27:43:82 || Team [TR]
User avatar
Bludek
38mins club
Posts: 1725
Joined: 23 Sep 2009, 10:56
Team: CART
Location: Some pub in Prague

Re: new ranking?

Post by Bludek »

Kopaka wrote:...nice stuff about ranking...
but I still dunno, how is relative ranking calculated. Can you pls explain it?
Image
EOL top 10 kuski of 2014 and 2015.
21:03:48 <umiz> i like 99% of bludek levels
User avatar
Kopaka
39mins club
Posts: 6610
Joined: 23 May 2002, 13:59
Team: LAME
Location: In a northern danish city beating YOUR record.
Contact:

Re: new ranking?

Post by Kopaka »

Bludek wrote:
Kopaka wrote:...nice stuff about ranking...
but I still dunno, how is relative ranking calculated. Can you pls explain it?
In short you gain points for those you beat in a battle and lose points for those that do better than you in a battle. The amount you gain or lose for each player depends on his and yours ranking, so you get a bit more if he has better and a bit less if he has a worse ranking. So depending on how many and who you beat or lose to in a battle you might gain or lose points from that battle.
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Belma site and statistics: ideas and suggestions needed

Post by Lukazz »

team rating should be points of 3 best players of the team together, and not average.
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
Lousku
Kuski
Posts: 2925
Joined: 5 Feb 2010, 00:25
Team: BAP
Location: expensive land of dads

Re: Belma site and statistics: ideas and suggestions needed

Post by Lousku »

Lukazz wrote:team rating should be points of 3 best players of the team together, and not average.
That would be perfect, but another solution is to make sensible teams with less than a thousand members. I'm guilty.
then again i don't know anything
maybe easier not to think abouut alöl things thought than not things thought ... or something..=?
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Re: Belma site and statistics: ideas and suggestions needed

Post by Lukazz »

i have the feeling that the points of the whole top 10 hasn't changed for a couple of days now.
TT: 36:59:53 || Avg TT: 38:09:65
Post Reply