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

KD
Kuski
Posts: 619
Joined: 14 Jul 2002, 11:48
Location: Szeged, Hungary

Post by KD »

Juski wrote:
KD wrote:and would practically limit the amount of RP you can have, so crazy höylas don't get a too big advantage.
Why shouldn't they?
Basically because this isn't scores but ranks, which is a battle-playing-skill order of people, and höyling skill is measured by other scores and stats good enough. :)

zworqy: these equations are pretty simple and nice, also have some good qualities like noone can go under zero in ranks, and noone can gain infinite points either.
If you keep going in last in battles you will eventually lose almost no RP (because you reached the actual level your skills represent :D ), but you can improve quite fast if you suddenly become a good player.

I'm against taking 0 apples into count, also not favour that same result means same place.
Thinking is not entertainment but an obligation! - Strugatsky bros
User avatar
wjelo
Kuski
Posts: 669
Joined: 28 Jan 2005, 19:45
Location: a random city called Radom

Post by wjelo »

KD wrote:I'm against taking 0 apples into count, also not favour that same result means same place.
0 apples should count in results, but all players on the same place, no point in punishing surviving longer (same with 2,3 etc apples)
with finishing times obviously it's different
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Post by Lukazz »

dunno if suggested yet, but i think statistics about different kind of levels would be VERY nice. that will only work of course when you set sort of level in startballe and i think nobody will do that :P like bounce level, pipe leve, hoyl level, uphill level, or something! then we can see who is the best piper and so on...
somehow very nice but i think ppl are too lazy. sometimes they are even too lazy to check "see others"
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
SveinR
Moporator
Posts: 5469
Joined: 21 May 2002, 08:05
Location: Oslo, Norway
Contact:

Post by SveinR »

Both letting 0 apples count and not letting them count has its problems. Say you have a lev, that is incredibly difficult, and doesn't contain any apples. Lots of people play this lev throughout the battle time, but only a few people finishes. If 0 apples doesn't count, then that would be unfair, since the few who have finished won't get the acknowledgement that they have beaten all those people.
But then there are also those battles where a lot of people just tries the lev once or twice then leaves (without getting an apple) or just spectates, and then these obviously shouldn't count in the results.
Since the last scenario occurs far more often than the first one, getting 0 apples should not count in the results.

If two or more people get the same result, whoever got it first should be placed higher. Sure it might feel stupid to be "punished" for surviving longer (if several people got the same amount of apples), but it would be far worse if someone were to get, early on, 50 apples in a pipe lev, and then 10 minutes later other people got the same, and they would be tied. Clearly the one who got it first should be placed higher.
Was it cast for the mass who burn and toil?
Or for the vultures who thirst for blood and oil?
Rules | FAQ
User avatar
Lukazz
36mins club
Posts: 5241
Joined: 4 Jul 2004, 12:10

Post by Lukazz »

SveinR wrote:Both letting 0 apples count and not letting them count has its problems. Say you have a lev, that is incredibly difficult, and doesn't contain any apples. Lots of people play this lev throughout the battle time, but only a few people finishes. If 0 apples doesn't count, then that would be unfair, since the few who have finished won't get the acknowledgement that they have beaten all those people.
But then there are also those battles where a lot of people just tries the lev once or twice then leaves (without getting an apple) or just spectates, and then these obviously shouldn't count in the results.
Since the last scenario occurs far more often than the first one, getting 0 apples should not count in the results.
Solution: Put some fucking apples in your levels!
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
wjelo
Kuski
Posts: 669
Joined: 28 Jan 2005, 19:45
Location: a random city called Radom

Post by wjelo »

SveinR wrote:If two or more people get the same result, whoever got it first should be placed higher. Sure it might feel stupid to be "punished" for surviving longer (if several people got the same amount of apples), but it would be far worse if someone were to get, early on, 50 apples in a pipe lev, and then 10 minutes later other people got the same, and they would be tied. Clearly the one who got it first should be placed higher.
ok i agree with that but first who gets those apples should be placed higher and not the one who gets those appple and dies first. see the difference? think about it
dunno if possible though, mila can tell us
teajay
Donator duck
Posts: 10043
Joined: 3 Apr 2003, 17:53

Post by teajay »

What wiel said I think we can all agree on, it's a good thing he stated it here. I was thinking exactly the same earlier today but figured it's impsy. I haven't got any clue on what the elma structure allows for modifications, and so far I've been surprised multiple times, but I don't think this is possible.
User avatar
berhabdul
Banned
Posts: 1125
Joined: 18 Jul 2005, 17:09
Location: Warsaw, Poland
Contact:

Post by berhabdul »

SveinR wrote:Both letting 0 apples count and not letting them count has its problems. Say you have a lev, that is incredibly difficult, and doesn't contain any apples. Lots of people play this lev throughout the battle time, but only a few people finishes. If 0 apples doesn't count, then that would be unfair, since the few who have finished won't get the acknowledgement that they have beaten all those people.
But then there are also those battles where a lot of people just tries the lev once or twice then leaves (without getting an apple) or just spectates, and then these obviously shouldn't count in the results.
Since the last scenario occurs far more often than the first one, getting 0 apples should not count in the results.

If two or more people get the same result, whoever got it first should be placed higher. Sure it might feel stupid to be "punished" for surviving longer (if several people got the same amount of apples), but it would be far worse if someone were to get, early on, 50 apples in a pipe lev, and then 10 minutes later other people got the same, and they would be tied. Clearly the one who got it first should be placed higher.
Kinda max, mega agreement:)
User avatar
milagros
Cheatless
Posts: 4560
Joined: 19 May 2002, 17:05

Post by milagros »

SveinR wrote:Since the last scenario occurs far more often than the first one, getting 0 apples should not count in the results.
why dont you just esc the level after the end of the balle? then result wont count
Last edited by milagros on 26 Jun 2007, 14:12, edited 1 time in total.
[carebox]
User avatar
milagros
Cheatless
Posts: 4560
Joined: 19 May 2002, 17:05

Post by milagros »

wiElOryB wrote:dunno if possible though, mila can tell us
no
[carebox]
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Post by 8-ball »

if 0 apples won't count, i'll put an apple on start position in all my future levels
39:37,91
User avatar
zworqy
Kuski
Posts: 3706
Joined: 19 May 2002, 23:17
Location: Lilla Edet, Sweden
Contact:

Post by zworqy »

8-ball wrote:if 0 apples won't count, i'll put an apple on start position in all my future levels
that only cures the symptoms, not the disease
<Fihlvein> another case of zworqy-is-always-right closed i guess
<yoosef> zworqy doesnt suck at anything
KD
Kuski
Posts: 619
Joined: 14 Jul 2002, 11:48
Location: Szeged, Hungary

Post by KD »

8-ball: maybe put an apple close to the start so people who actually want to play it will pick it up.
It's really not fair to put one at the very start.
At least give a chance to see the lev without any consequences.

mila: i often play some level then when i see there is a battle going on i check the lev, see its shit (not noni lev!), and go back to the lev i was playing. It is not that handy that i should press f12 all these times. (neither a nice thing to do)
Thinking is not entertainment but an obligation! - Strugatsky bros
User avatar
Juski
Kuski
Posts: 2200
Joined: 26 Dec 2003, 20:53
Location: irc://irc.ircnet.org/ranks

Post by Juski »

maybe a show a map button should be availible before palyig it so you can study the lev before playing teh, or you can open it in the editor...

most logical place tyo put show manp butotn would be in level menu, which is accesible after ylu play level :x

Also you should increase the amount of levels the editor can load, current max is 1000levs, i cant even view half of my levels :x
No regrets Image
Are you LOST?
User avatar
berhabdul
Banned
Posts: 1125
Joined: 18 Jul 2005, 17:09
Location: Warsaw, Poland
Contact:

Post by berhabdul »

hey isn't it wrong topic. hey, maybe The Server would make a map (jpg) while the level is uploaded, and Belma could download both the level (.lev) and map (.jpg) which could be open in a new Belma-submenu. i strongly guess that implementing a viewing-jpg option to Belma is very close to easy. ye0 that woulda be really elma2ish
User avatar
milagros
Cheatless
Posts: 4560
Joined: 19 May 2002, 17:05

Post by milagros »

berhabdul wrote:i strongly guess that implementing a viewing-jpg option to Belma is very close to easy
yes it's ez, go berh, you can do it!
[carebox]
User avatar
Zweq
34mins club
Posts: 4055
Joined: 28 Nov 2002, 15:54
Location: suo mesta

Post by Zweq »

or simply 1 testride before getting to results is possible. you know, for vsync holes etc
Image
User avatar
milagros
Cheatless
Posts: 4560
Joined: 19 May 2002, 17:05

Post by milagros »

Zweq wrote:or simply 1 testride before getting to results is possible. you know, for vsync holes etc
ez for skint0r to remove in stats but many times you manage to do only one try (like in that 44mins balle where jappe was supposed to pwn)
[carebox]
teajay
Donator duck
Posts: 10043
Joined: 3 Apr 2003, 17:53

Post by teajay »

milagros wrote:ez for skint0r to remove in stats but many times you manage to do only one try (like in that 44mins balle where jappe was supposed to pwn)
Hihi, that was fun =)

but yeah, sometimes a battle is like 5 mins and you do only one finish on first try or one that is better than everything you drive after that. I get that sometimes.
User avatar
dz
first 39tt
Posts: 3749
Joined: 19 May 2002, 15:16
Team: FM
Location: Finland
Contact:

Post by dz »

milagros wrote:
SveinR wrote:Since the last scenario occurs far more often than the first one, getting 0 apples should not count in the results.
why dont you just esc the level after the end of the balle? then result wont count
Not very easy on levels where you need to find a safe place to chat, most often you die before you manage that. :)
User avatar
niN
Kuski
Posts: 2631
Joined: 22 Aug 2005, 12:23
Team: HoHo
Location: Sweden, Gothemburg
Contact:

Post by niN »

I think it would be cool with some Oldies goodies days maybe once every month or so, when on that day we only play levels which have been battled the previous month with the most 'plays' (amount of kuskis participating). Maybe this is stupid but I sure as heckens name like it :D
Team HotHorses, and I'm converting to Icelandian now...
User avatar
berhabdul
Banned
Posts: 1125
Joined: 18 Jul 2005, 17:09
Location: Warsaw, Poland
Contact:

Post by berhabdul »

omg hurry up never mind what they would look like, where are they?

omg om i know it needs time
User avatar
Xiphias
39mins club
Posts: 4098
Joined: 23 Nov 2004, 23:05

Post by Xiphias »

It registeres everything right?... new times on internals and externals ?
then I'd like a simple Date of improvement list for the internals instead of checking all of the replays =/ which I don't even have atm.

Code: Select all

Date of last improvement

01 Warm Up     10. Jan 2005
02 Flat Track  25. Jun 2007
03 Twin Peaks  30. Aug 2006
...
Not that I know much about how many things you have to do n' stuff.. but this shouldn't be hard.
Thorze wrote:I just wanted to make a cool topic like Juish have cool topics..
User avatar
zebra
Kuski
Posts: 1009
Joined: 23 Sep 2003, 15:35
Team: TAP
Location: Finland
Contact:

Post by zebra »

milagros wrote:it was like this
lets say players on 1.-4. position had ranks A,B,C,D before
after balles ranks will be
A(n+1) = A(n) * (1 + k*exp(q*(B(n)-A(n)))) * (1 + k*exp(q*(C(n)-A(n)))) * (1 + k*exp(q*(D(n)-A(n))))
B(n+1) = B(n) / (1 + k*exp(q*(B(n)-A(n)))) * (1 + k*exp(q*(C(n)-B(n)))) * (1 + k*exp(q*(D(n)-B(n))))
C(n+1) = C(n) / (1 + k*exp(q*(C(n)-A(n)))) / (1 + k*exp(q*(C(n)-B(n)))) * (1 + k*exp(q*(D(n)-C(n))))
D(n+1) = D(n) / (1 + k*exp(q*(D(n)-A(n)))) / (1 + k*exp(q*(D(n)-B(n)))) / (1 + k*exp(q*(D(n)-C(n))))
there will be also some max/min value of exp(..) so it wont be any unstable
these rules would keep A(n)*B(n)*C(n)*D(n) constant
if someone much better beats someone who sax, it will increase his coefficient by some 1.000001 and decrease the others one byt same 1.000001, if someone much worse beats someone, ez increase weight a bit more, constants q,k will be some 0.01 or smth, simply somehow set (it means the speed of ranks changing), starting value will be some 1000 or 1.000 or smth, goodplayers will have >1, bad players <1
its not decided if 0 apples results are taken, if same times means same position or not and if same apples means same result (definitely for 0 apples)
After thinking this system some more i understood that it sucks.

Many times I (or some other) come to battle when there is 1 minute left. I get 1 apple and will be the last in the list. With this system, I would lose some points, without any reason. It's pretty ok if the noobs ahead of me get some points but I shouldn't lose any points.

So, the pointing system has to be such that you can't lose any points/rank by playing (you can only lose rank by not playing when other players pass you (when they get more points than you have)).
A winner of 4 GAA's (mc2 included), winner of mkup206, and a proud member of team TAP.
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
User avatar
skint0r
39mins club
Posts: 768
Joined: 16 Jun 2002, 07:36
Location: Oslo, Norway

Post by skint0r »

Well unless you have a better suggestion (in the sense of an actual algorithm), this seems to be pretty damn good. If you're so worried about losing a few ranking points, then don't play. It's not like you would be the only one joining a battle late, I'm sure it would even out amongst players.

Again, feel free to make something better. Sure it's nice to hear suggestions and such, but we really need something... "real" -- a formula, and not just an idea.
Prestigious member of 14.6x Tutor14 club
User avatar
8-ball
39mins club
Posts: 4496
Joined: 9 May 2003, 13:30
Team: MiE
Location: Riga, Latvia

Post by 8-ball »

that seemed a good formula. if you don't want to lose points, don't join at 1 min left.
39:37,91
User avatar
zebra
Kuski
Posts: 1009
Joined: 23 Sep 2003, 15:35
Team: TAP
Location: Finland
Contact:

Post by zebra »

ok, i was thinking about something like this:

at the beginning everybody has 0 points.

let p(x) be the points of the player x.
let d be the duration of the battle in minutes.
let b(x) be the amount of players the player x beat in the battle.
let k be a tiny constant, let's say 0.01.
let m be another constant, let's say 0.1.
let n be another little bigger constant, let's say 20.

players a,b and c play a 10-min-battle and a wins, b is second and c last.
before the battle, a had 300 points, b had 500 points and c had 200 points.
now, if p(b) - p(a) is positive (i.e. player b has more points than a), player a gets (p(b) - p(a)) * (d + n) * k points because he beated b. Likewise, if p(c) - p(a) is positive, player a gets (p(c) - p(a)) * (d + n) * k points because he beated c. Additionally player a gets b(a) * (d + n) * m points ((d + n) * m points for each player which he beated). These 'additional' points have to be there to get the points to rise from 0 to some bigger numbers.

In this case player a would get (500 - 300) * (10 + 20) * 0.01 + 2 * (10 + 20) * 0.1 = 66 points. That means that he gets 60 points for winning player b who had more points than him, and 3 points for each player which he beated.
Player b gets 1 * (10 + 20) * 0.1 = 3 points because he beated c.
Player c gets 0 points (the last player would never get points).

Now, player a has 366 points, player b 503 points and c has 200 points, so their ranks are 2, 1 and 3rd (assuming that there aren't any other players in belma at all).

Now, you wouldn't lose any points by playing, which make sense.
You could only get higher in ranks by playing, which make sense.
You wouldn't be punished from playing battles.
Good players would get many points at the beginning. When they reach their level, they wouldn't get any points so easily anymore.
Bad players would only get points from players which they beated, which is not much, but is still something.
You would get more points from winning a longer battle. By adjusting the constant n the effect can be made smaller or bigger.

About the problem that some players would get points easily by just playing against each other while no-one other is around: yes, they would get some points but not much, and the best player of those would get only the 'additional' points (which i mentioned above) which would be very low. Also, there won't be much times when such situation occurs, because everybody is welcome to join at any time.

The values of the constants k and m has to be chosen relatively small to avoid player points to go too big. And player points should be floats (or doubles) so that the pointing would be accurate. Ranks (positions) could be integers.

About the 0-apple problem, i think it's better to leave it out of the point system. Simply take away the 0-apple-players and the designer of the level from the results before making any kind of calculations, so they wouldn't get any points (and other ppl wouldn't get any points for beating them).
If more than one ppl gets for example 1 apple, those shouldn't be tied, but to give points in the order in which they got the apple (like it is now in the results). More generally, with this pointing system, the time/apple amount which you get in the level would not be relevant, only your position in the battle results.
A winner of 4 GAA's (mc2 included), winner of mkup206, and a proud member of team TAP.
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
DaFred
39mins club
Posts: 417
Joined: 17 Jan 2007, 21:00

Post by DaFred »

Ah heres right place for this ^^ ..

I think its not good that you can dl a level from b.attle.info before battle starts. There is no link but for example if Bjorn make a few levels. First one Bjorn101, sekond Bjorn102 and I see that theres another lev by Bjorn in queu its not difficult to guess whats the name of next lev.
User avatar
milagros
Cheatless
Posts: 4560
Joined: 19 May 2002, 17:05

Post by milagros »

that zebra's idea sax, because the one who plays most will have the most points because there are always n00bs around
[carebox]
KD
Kuski
Posts: 619
Joined: 14 Jul 2002, 11:48
Location: Szeged, Hungary

Post by KD »

Yeah, zebra keeps mixing up the points system with the ranks system, those are two different thing.
Of course you loose ranks if you play and finish at a bad place, that is how your "strength" is decided.
Thinking is not entertainment but an obligation! - Strugatsky bros
User avatar
zebra
Kuski
Posts: 1009
Joined: 23 Sep 2003, 15:35
Team: TAP
Location: Finland
Contact:

Post by zebra »

milagros wrote:that zebra's idea sax, because the one who plays most will have the most points because there are always n00bs around
What's the problem with that?
anyway, the effect wouldn't be so big as it was in #battle when practically only the amount of battles was counted.
Remember that if one player would be very high in points, you would get much points if you beat him and catch him pretty easily.
Furthermore, he wouldn't get many points from beating noobs, only the 'additional points' which i explained in my post, which would be rather small.
KD wrote:Yeah, zebra keeps mixing up the points system with the ranks system, those are two different thing.
Of course you loose ranks if you play and finish at a bad place, that is how your "strength" is decided.
why they should be two different things?
I don't like the idea that i'm being punished of playing battles.
There are so many things why i could accidentally end up being the last one in battle: maybe i entered the battle when 1 min was late, or the lev was bad and i made only 1 try, my connection broke up etc.

That's why I would like to maintain those points which i have already gathered. Of course those who beated me in that battle where i was last, will get more points than me and maybe pass me in ranks, but i wouldn't lose my rank and get passed by a players that didn't play the battle. There's a slight difference.

There's also one weak point in the system which milagros explained: if my ranking goes below 1.0, what keeps me from making a new nick (if the starting value would be 1.0)? That's how i could easily get up in ranks. I think that would be widely used among noobs who don't have a known elma nick yet and who could easily change nick.
A winner of 4 GAA's (mc2 included), winner of mkup206, and a proud member of team TAP.
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
User avatar
niN
Kuski
Posts: 2631
Joined: 22 Aug 2005, 12:23
Team: HoHo
Location: Sweden, Gothemburg
Contact:

Post by niN »

I haven't read anyones pointing systems, but I just want to say this:

Points should be given like such: The higher rank the person you beat has got, the more points you get. You will be ranked by your amount of points.

Quite simple huh?
Team HotHorses, and I'm converting to Icelandian now...
User avatar
Xiphias
39mins club
Posts: 4098
Joined: 23 Nov 2004, 23:05

Post by Xiphias »

niN wrote:I haven't read anyones pointing systems, but I just want to say this:

Points should be given like such: The higher rank the person you beat has got, the more points you get. You will be ranked by your amount of points.

Quite simple huh?
haha.. now come with a formula for that system =D
Thorze wrote:I just wanted to make a cool topic like Juish have cool topics..
KD
Kuski
Posts: 619
Joined: 14 Jul 2002, 11:48
Location: Szeged, Hungary

Post by KD »

niN: it's exactly like that already.
And it has every other feature zebra mentioned (you get more points if you beat better players, etc.).

zebra: in long-term those things won't matter at all, everyone else does them too. You are not punished by playing, it just gives a chance for the system to evaluate your skill more and more precisely. Remember, everyone else can join late in battles, and stop playing when meets a bad level. I think if you are really that obsessed with your ranks you can eliminate those with a little attention, like not joining battles in the last minute, or making a testride on the level and trying to avoid apples (watch your minimap instead) - so you can decide if you want to play it.
Anyway your best option would be to ignore your ranking completely and only watch your points if you are that crazy about this.
Thinking is not entertainment but an obligation! - Strugatsky bros
User avatar
niN
Kuski
Posts: 2631
Joined: 22 Aug 2005, 12:23
Team: HoHo
Location: Sweden, Gothemburg
Contact:

Post by niN »

then wth is all this fuzz about? A system like that should be veery easy to create... I've got it in my mind already.
Team HotHorses, and I'm converting to Icelandian now...
User avatar
Palme
Kuski
Posts: 375
Joined: 14 Oct 2006, 10:49
Location: Berlin
Contact:

Post by Palme »

Well Zebras new System is good imo, but it don t fixes the Problem, that you count in stats when u only play 1 min out of 20.

Lets say Zweq is first in Ranking with lot s of points, he joins in a really fucked up lev, get s one apple and then quits playing this lev cuz it sucks. Then many player would get lot s of points, just because zweq got one apple, not because they were really better.

Is there a way to recognize the time players spent in a specifiv lev? Then we could say you only count to the ranking if you ve played 1/10 of the battle duration or sach.

Or maybe make a key wich states that u don t want to count in stats and u only can press the key when u died less than 5 times...

I don t know what is possible for mila to find out in some prog about one player related to one lev - but unfortunatly i don t think that there s much he can do.

I just recognized this thread, so i ll think of some solution myself :)
Apoku RuLeZ! X) http://teamapoku.funpic.de
Image

- me signature is good signature -
User avatar
zebra
Kuski
Posts: 1009
Joined: 23 Sep 2003, 15:35
Team: TAP
Location: Finland
Contact:

Post by zebra »

Palme wrote:Well Zebras new System is good imo, but it don t fixes the Problem, that you count in stats when u only play 1 min out of 20.

Lets say Zweq is first in Ranking with lot s of points, he joins in a really fucked up lev, get s one apple and then quits playing this lev cuz it sucks. Then many player would get lot s of points, just because zweq got one apple, not because they were really better.

Is there a way to recognize the time players spent in a specifiv lev? Then we could say you only count to the ranking if you ve played 1/10 of the battle duration or sach.
Yeah the time spent on a lev could be counted too but i'm afraid that the system goes too complicated that way.
KD wrote:in long-term those things won't matter at all, everyone else does them too. You are not punished by playing, it just gives a chance for the system to evaluate your skill more and more precisely. Remember, everyone else can join late in battles, and stop playing when meets a bad level. I think if you are really that obsessed with your ranks you can eliminate those with a little attention, like not joining battles in the last minute, or making a testride on the level and trying to avoid apples (watch your minimap instead) - so you can decide if you want to play it.
Anyway your best option would be to ignore your ranking completely and only watch your points if you are that crazy about this.
You really think you could be without playing a battle level when you enter belma and see that it's 3 minutes left? ;)

I explained my system because it feels more intuitive than the milagros's one. I think it's pretty much alike the one in trackmania nations. The other thing is to bring up some conversation about which is really the best solution.

I'm ok with milagros's system too, and when i'm blaming it i'm not very concerned about my own rank (that sux anyway), but the system in general.
A winner of 4 GAA's (mc2 included), winner of mkup206, and a proud member of team TAP.
Play uni levels: http://koti.mbnet.fi/zebra/uni.html
Homepage: http://koti.mbnet.fi/zebra/elma.html
User avatar
skint0r
39mins club
Posts: 768
Joined: 16 Jun 2002, 07:36
Location: Oslo, Norway

Post by skint0r »

Not sure what the problem is, there will be both a ranking system and old point system independant of each other, so you can choose to just höyl points if you want. Still have not tested this formula yet, but it seems to be pretty solid I think.

As for people re-registering, that would be a bad idea on their part: a nice way to get yourself permbanned.

One idea however is this (just randomness, haven't thought if it's good or not): that if you're inactive, your rank will slowly decrease if your rank is above 1000 (or whatever value), and increase if you're under (increasing maybe a bit faster than the decreasing?). That way, 1000 is kind of the "bottom" value, and going under it will only show you as being pwned in battles for that week or month or whatever but you won't be stuck there forever.

Also it will keep someone from reaching top ranking, then quitting and just staying there and seem like being unreachable (though maybe this is the point if getting a good rank, dunoz).

Anyway, just little idea.
Prestigious member of 14.6x Tutor14 club
User avatar
Palme
Kuski
Posts: 375
Joined: 14 Oct 2006, 10:49
Location: Berlin
Contact:

Post by Palme »

You should be able to see the rank of "last month" "last three month" "last year" and "oberall".
When you become inactive your overall rank will slowly decrease as skint said.
With this new kuskis could just check the "last month" because they would probably be very bad in "overall". Just an idea though.

I m not clear if i understood zebra right:
Yeah the time spent on a lev could be counted too but i'm afraid that the system goes too complicated that way.
Does that mean mila can get these times or were you talkin about assumed case that he can get the times :P
If he can, the system wouldn t be complicated at all - there would be only one more calulation to get in stats or nat (but you should be able to choose)

Then for mila's ranking algorithm i don t get why "e" is needed in this. Seems way to complicated imo.
Maybe i don t understand it right, so this is what i think it does:

In my example, when I m talin bout C(n)-D(n) or sach i m talkin bout the whole "(exp* xxx)" thing
A(n+1):
Lets say q is like 0,1, then exp(q*(B(n)-A(n)) would be between0,09999 and nearly 0 and when k is like 0,01 it s between 0,00099999 and nearly 0
So the same with C(n)-A(n) is a bit smaller and D(n)-A(n) even smaller.

A(n+1)=Old Rank * 1,0005 *1,0003 *1,0001 =Old Rank * 1,0009...
Sounds nice here.

B(n+1):
B(n)-A(n) as before --> like 1,0005
C(n)-B(n) could be the same as B(n)-A(n) but also greater or lesser
D(n)-B(n) is smaller than C(n)-B(n)

B(n+1) = (OLD Rank / 1,0005) * 1,0005 * 1,0003 = Old Rank * 1,0003
I Wasn t sure about the "()" but i think the way i put them in my example are quite logical
Sounds nice here too.


C(n+1):
C(n)-A(n) is like small
C(n)-B(n) is bigger than (C(n)-A(n)
D(n)-C(n) can be greater lesser or same as C(n)-B(n)

C(n+1) = Old rank / (1,0004 *1,0005) * 1,0005 = Old Rank /1,0004
okay

D(n+) Would then be something like Dn /XXX


So when i got this right, in general the upper part of the battle will slightly increase (or when u are very good but only become 10th out of 20 decrease) and all others will slightly decrease.

So when i got this right, this seems quite max system.
Apoku RuLeZ! X) http://teamapoku.funpic.de
Image

- me signature is good signature -
KD
Kuski
Posts: 619
Joined: 14 Jul 2002, 11:48
Location: Szeged, Hungary

Post by KD »

Palme: i'm not sure which "e" you are referring to.

I think you got the formula right, the example seems ok.
Thinking is not entertainment but an obligation! - Strugatsky bros
User avatar
Palme
Kuski
Posts: 375
Joined: 14 Oct 2006, 10:49
Location: Berlin
Contact:

Post by Palme »

e = exp
Apoku RuLeZ! X) http://teamapoku.funpic.de
Image

- me signature is good signature -
KD
Kuski
Posts: 619
Joined: 14 Jul 2002, 11:48
Location: Szeged, Hungary

Post by KD »

In that case it is needed to make a nice curve, so when you are highly ranked you won't gain that much points, neither you loose when low ranked.
Thinking is not entertainment but an obligation! - Strugatsky bros
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 »

So if anyone doesn't know I've taken over making the site. I'll make some topic later with planned features and such, but for now I just need to ask a couple of things.

For the main kinglist/ranking thing no sort of crippled battles will count and only normal battles, I think we agree on this already. But there's a couple of options that I would like your opinions on.

See Other on/off
See Times on/off
Allow Starter on/off
Accept Bugs on/off

In my opinion see others, see times and accept bugs should count both on and off, but allow starter only off. Does people agree on this?


Second thing is, did we agree on the ranking system that milagros posted?
(Plan is to make both this and a "normal" kinglist like old zebra stats system)
milagros wrote:it was like this
lets say players on 1.-4. position had ranks A,B,C,D before
after balles ranks will be
A(n+1) = A(n) * (1 + k*exp(q*(B(n)-A(n)))) * (1 + k*exp(q*(C(n)-A(n)))) * (1 + k*exp(q*(D(n)-A(n))))
B(n+1) = B(n) / (1 + k*exp(q*(B(n)-A(n)))) * (1 + k*exp(q*(C(n)-B(n)))) * (1 + k*exp(q*(D(n)-B(n))))
C(n+1) = C(n) / (1 + k*exp(q*(C(n)-A(n)))) / (1 + k*exp(q*(C(n)-B(n)))) * (1 + k*exp(q*(D(n)-C(n))))
D(n+1) = D(n) / (1 + k*exp(q*(D(n)-A(n)))) / (1 + k*exp(q*(D(n)-B(n)))) / (1 + k*exp(q*(D(n)-C(n))))
there will be also some max/min value of exp(..) so it wont be any unstable
these rules would keep A(n)*B(n)*C(n)*D(n) constant
if someone much better beats someone who sax, it will increase his coefficient by some 1.000001 and decrease the others one byt same 1.000001, if someone much worse beats someone, ez increase weight a bit more, constants q,k will be some 0.01 or smth, simply somehow set (it means the speed of ranks changing), starting value will be some 1000 or 1.000 or smth, goodplayers will have >1, bad players <1
its not decided if 0 apples results are taken, if same times means same position or not and if same apples means same result (definitely for 0 apples)
User avatar
niN
Kuski
Posts: 2631
Joined: 22 Aug 2005, 12:23
Team: HoHo
Location: Sweden, Gothemburg
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by niN »

Vote Kopaka for president!
Team HotHorses, and I'm converting to Icelandian now...
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 don't really understand the ranking system but whater.
in general i think you shouldn't lose points if you have some bad position because of shit level or whatever. anything else doesn't really matter for me.

and i agree that battle starter shouldn't count.
TT: 36:59:53 || Avg TT: 38:09:65
User avatar
Antz
Donator
Posts: 1168
Joined: 14 Sep 2002, 16:52
Team: Quack
Location: Espoo, Finland
Contact:

Re: Belma site and statistics: ideas and suggestions needed

Post by Antz »

it should take in account the time spent playing. i wouldn't want my rank to drop because i tried the level once and felt it was shit so didn't wanna play for another 30 minutes.
Image
User avatar
FinMan
36mins club
Posts: 2038
Joined: 13 Feb 2007, 11:14
Team: dat
Location: Jyväskylä, Finland

Re: Belma site and statistics: ideas and suggestions needed

Post by FinMan »

What about best winning rows or sumthing else... Or whole weeks average, of all played battles, and counts only if played (some ok number) in that weeks battles. :o

Btw. if player does not play the level at all, like got 0 apples and only looked level and quitted there, count?
Image
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 »

i know that i'm off-topic but i suggest to edit one life balle for ## life(s) balle where ## can be chosen by the designer
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 »

Mawane wrote:i know that i'm off-topic but i suggest to edit one life balle for ## life(s) balle where ## can be chosen by the designer
ye that would be nice. to chose at least from 1-10 lifes :)
TT: 36:59:53 || Avg TT: 38:09:65
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 »

Lukazz wrote:
Mawane wrote:i know that i'm off-topic but i suggest to edit one life balle for ## life(s) balle where ## can be chosen by the designer
ye that would be nice. to chose at least from 1-10 lifes :)

Code: Select all

[16:58] <Mawane_TR> i suggest also to edit one life balle for ## life(s) balle where ## can be chosen by the designer
[16:58] <teh_mila> that woudl be ez xiit by disconnecting, testing and then driving
[16:59] <veezay> oh my god thanks mila luv you forever <3
[16:59] <Mawane_TR> but can can do same for one life
[16:59] <teh_mila> for onelife you dont have really time
[16:59] <veezay> dame i wanna try it
[16:59] <teh_mila> but if many tries => you dont have to be there at start => ez xiit
[17:00] <Mawane_TR> so, edit for: you HAVE to be there at start or you're not in results, no?
[17:00] <teh_mila> also you would be able to disconnect during run so it may not count and all that shit
[17:01] <teh_mila> but then you wold be there at start.. do one try.. disconnect every next try 1 sec before end or what and then do again if sucekd
[17:01] <Mawane_TR> if you leave lev, u lose results
[17:01] <teh_mila> you can always use different elma
[17:01] <Mawane_TR> damn, u really dont want this
[17:01] <teh_mila> yes i dont:)
[17:01] <Mawane_TR> xD
[17:02] <teh_mila> anyway, its not really big difference to normal battle
[17:02] <teh_mila> like if you have 30 sec lev and want 4 tries, its almost the same as setting 2 mins time
[17:02] <teh_mila> its not really but kinda
lol vz was like ImagerandomImage

Code: Select all

[16:42] <teh_mila> polygons to 1500 or 1200 (forgot), objects to 252, vertices to 20k
Website || TT:41:45:64 || Team Image
Image[url=steam://friends/add/76561198025490048]Image[/url]
Post Reply