View previous topic :: View next topic |
Author |
Message |
Evers17
Joined: 04 Aug 2007 Posts: 207
|
Posted: Tue Dec 04, 2007 2:08 am Post subject: |
|
|
Thats very interesting, thanks for the info and giving the formula also.
I = Math Nut _________________
GOALS
[X] 5* TTFAF
[X] 5* TWIE
[X] Get a 1st
[X] Top 15 GH:A - 15th |
|
Back to top |
|
|
ShinobiAC
Joined: 04 Dec 2007 Posts: 1065 Location: Raleigh, NC
|
Posted: Thu Dec 20, 2007 5:57 pm Post subject: |
|
|
It seems to me that a grading system based on score would be more fair than one based on placement.
You could have somebody blow away the competition in 3 songs, getting 20k over their nearest opponent, then get 2000 points less than the leader on the last song and end up losing? Doesn't make much sense to me...
Obviously points aren't a 1 to 1 mapping on each song. The solution is to have an exponential score mapping where 100 points would be the theoretical maximums for the songs (and still allow for scaling higher if somebody manages to squeeze more). All you would have to do is get the formula created and plug in the number you want to set as the max for that song and difficulty. I believe the effort required to set up a new GUI or modify what you have currently set up for inputting songs would be worth the trouble to make it a much fairer system.
I am borrowing this idea from the way scoring is done in decathlons. I think it's probably the closest comparison to what is being done in the leagues. _________________
|
|
Back to top |
|
|
moose526
Joined: 21 Jan 2008 Posts: 1
|
Posted: Mon Jan 28, 2008 1:44 am Post subject: |
|
|
i like this one more
it works better |
|
Back to top |
|
|
rafsta93
Joined: 24 Sep 2007 Posts: 250 Location: Manchester UK
|
Posted: Wed Feb 06, 2008 9:05 pm Post subject: |
|
|
i still dont really get it...... im sorry can sum1 plz put this in simpletr words |
|
Back to top |
|
|
yudlugar
Joined: 29 Oct 2007 Posts: 356
|
Posted: Wed Feb 06, 2008 9:33 pm Post subject: |
|
|
rafsta93 wrote: | i still dont really get it...... im sorry can sum1 plz put this in simpletr words |
The scoring system in the original post is, rank everyone in order according to scores from 1 to say 50. (50 being best, 1 worst)
Then multiply each position by a number, determined by the formula given. As the position goes up, the difference between the numbers increases.
A simplified example, 4 person tournament
last place = 1x1 = 1
3rd place = 2x2 = 4
2nd place = 3x4 = 12
1st place = 4x8 = 32
(I've simply doubled the multiplier for each improvement in rank, the actualy system used is more complex, but that should give you the general idea)
This means the people at the top of the league have a larger gap between the points gained than those at the bottom.
The scoring system suggested a few posts up is similar, but rather than multiplying the position by a number multiply the score divided by max score by a number to get the points.
For example, if the most points possible in a song is 100, then the points would be awarded:
4th place, 25 point, 1/4 x 40 = 10
3rd place,50 points, 1/2 x 50 = 25
2nd place 75 points, 3/4 x 70 = 52.5
1st place 100 points 1 x 100 = 100
Personally I feel giving points by rank is best, with the points based system songs with more points will give a greater deviation and therefore easier to make-up/extend a lead. Most of the songs with big scores are harder ones, so it makes the leagues biased towards players who do better at the harder songs and being able to squeeze and optimise lower songs becomes pretty much pointless, as you can just get within a few thousand points of the leader and concentrate on getting big scores in the other songs. |
|
Back to top |
|
|
rafsta93
Joined: 24 Sep 2007 Posts: 250 Location: Manchester UK
|
Posted: Thu Feb 07, 2008 1:34 pm Post subject: |
|
|
ah thx. I thought i was quite good at maths. Obviously not but that really helps. This is my first league i have entered because i now feel up to the standard. Ill go back to the original post and try to understand it........ again..... to see how it really works. |
|
Back to top |
|
|
K81o7
Joined: 06 Jan 2008 Posts: 446
|
Posted: Sun Feb 10, 2008 1:03 pm Post subject: |
|
|
Nice work, Barfo! However, is there some way to empirically calculate derivatives of that curve you're trying to approximate? If so, a not too complicated Taylor polynomial could get reaaaallly close.
Not that what you have here isn't close enough though. Nice job. |
|
Back to top |
|
|
zztwig12
Joined: 21 May 2007 Posts: 12
|
Posted: Sat Mar 08, 2008 3:09 am Post subject: |
|
|
confusing, but pretty cool |
|
Back to top |
|
|
GuitarGeek08
Joined: 19 Jan 2008 Posts: 3213
|
Posted: Sun Apr 27, 2008 5:03 pm Post subject: |
|
|
so the formula would look something like this (simplified and easier to understand)?
F(x)= 62.5 * (x+7.5) * (x^4 +30) * x^7 |
|
Back to top |
|
|
CaesarsSalad
Joined: 18 Sep 2007 Posts: 25
|
Posted: Sat May 10, 2008 12:29 am Post subject: |
|
|
@ Barfo:
I didn't read the whole thread, but why didn't you just use an exponential function, like y=(2^x)-1 instead of your polynomial? First player would get 1 point multiplied with whatever you see fit, the last would get zilch (if there are infinite participants) or close to nothing and the tangent for x=0 would be ln(2), which seems alright. |
|
Back to top |
|
|
MindControlFun
Joined: 28 May 2007 Posts: 78 Location: The 'burbs.
|
Posted: Sat May 10, 2008 4:07 pm Post subject: |
|
|
this may be a really stupid question, and if it was answered already somewhere else I'm sorry:
Why don't you just take an average of the scores and turn them all into a percent of the best? That way it would be normed, and wayyy simpler.
Example:
Say we have someone who got:
-126938, 340498, 213462, 116104
-111873, 326938, 212678, 113103
-109004, 299328, 190473, 100024
First place would be (126938+340498+213462+116104)/4 = 199250.5
2nd: (111873+326938+212678+113103)/4 = 191148
3rd: (109004+299328+190473+100024)/4 = 174707.25
The points would be distributed as so:
1st: 100 (or 1, whichever)
2nd: 95.93 (or .9593)
3rd: 87.68 (or .8768) _________________
Knee deep in the belly of the beast-- offering us another cut-rate version of your dream come true.
The road to 6*ing expert:
GH1: 32/47
GH2: 59/64
GH80: 7/8 played
GH3: 27/44 played
GHWT: Not sure? |
|
Back to top |
|
|
CaesarsSalad
Joined: 18 Sep 2007 Posts: 25
|
Posted: Sun May 11, 2008 12:54 am Post subject: |
|
|
@mindcontrolfun:
Because it is very hard to turn a near-optimum score into an optimum score, though the difference in points can be less than ten. Barfo wants to emphasize the difference at the very top. |
|
Back to top |
|
|
atruebassist
Joined: 04 Aug 2007 Posts: 152
|
Posted: Sun May 11, 2008 3:28 pm Post subject: |
|
|
I'm sure this has been asked before so I'm sorry. When will everything show up? (places, points [however you guys do it]) I heard 24 hrs. but then i also heard 48 hrs. Just wanted to clear it up. _________________
3 finger player of GH and PROUD OF IT!
Season 9 GH2X Div. 3 Proud runner-up! |
|
Back to top |
|
|
MindControlFun
Joined: 28 May 2007 Posts: 78 Location: The 'burbs.
|
Posted: Sun May 11, 2008 11:23 pm Post subject: |
|
|
@caesarsalad:
Thanks, I was wondering. I guess now that someone said so it makes sense. Wouldn't a logrithmic function be much easier than what he has now? _________________
Knee deep in the belly of the beast-- offering us another cut-rate version of your dream come true.
The road to 6*ing expert:
GH1: 32/47
GH2: 59/64
GH80: 7/8 played
GH3: 27/44 played
GHWT: Not sure? |
|
Back to top |
|
|
CaesarsSalad
Joined: 18 Sep 2007 Posts: 25
|
Posted: Mon May 12, 2008 12:18 pm Post subject: |
|
|
@mindcontrolfun:
Logarithmical? I think you mean exponential. I already suggested that a few posts earlier and yeah, it would be easier... |
|
Back to top |
|
|
|