Ranking Formula

[Furious_Angel]

ok so i got this formula that i think is right (wins +1/2(draws+losses))/sqrt(losses+1) and i think it might be the formula or close to it that seed uses and according to this forula i'm the best ranked on GL server (by far) :p if you think this is right or even close post and vote!

[Aro23r]

I believe it is something much like, or similiar to this.

(wins + draws) / (losses + draws)

But hey, its just a guess and one that should be utilized...

[Furious_Angel]

well so far my formula has been correct in every test o to the ranking and test it out itshould come out right if you can do math right

[Aro23r]

I haven't tested it out, either yours or mine, but does it account for a person with lets say 110 - 0 being better than a person with 109 - 0. Any formula should do this. I haven't tryed them, I'm just asking.

[Half-Mad]

Originally posted by Furious_Angel
ok so i got this formula that i think is right (wins +1/2(draws+losses))/sqrt(losses+1) and i think it might be the formula or close to it that seed uses and according to this forula i'm the best ranked on GL server (by far) :p if you think this is right or even close post and vote!

It does not work. I put your equation in to Excel, and...
Check ranks 79, 73, 71, etc.
(I started checking at 100, working backwards)

[Furious_Angel]

well like any formula theres gunna be decimals so it can round and even if it doesnlt come out right its pretty f**kin close either that or you put it in excel wrong

[Excession]

Since the ranking system doesn't take into account the relative skill/records of the players, does it really matter how it's calculated?

[Half-Mad]

Originally posted by Furious_Angel
well like any formula theres gunna be decimals so it can round and even if it doesnlt come out right its pretty f**kin close either that or you put it in excel wrong

I came up with a couple formulas that were just as good as the one you offered. None were accurate all the way down the list of the top 100.


Maybe it's wins/losses + x*games played.
x = ?
... just a quick thought. Haven't tried any calculations.

[Furious_Angel]

you have an aunknown in your formula does that mean your solving for the unknown and the unknown is youre score or your just dont know what it means?? lol

[AzN_GuY]

i believe the unknown is a factor that he just doesnt know...

[Snork]

Maybe to make it easier, if you come up with a formula, post some examples to show how it works right away, see if it's accurate.

[xyxaxyz2]

Ok everyone. A few points:

1. This is my formula.
2. It isn't exactly correct.
3. It is very, very, close, at least on the GL server. You can estimate very well what your ranking would be compared to the people already on the ranking using my formula.
4. I can't post examples because the calculations take too long. I will post point values in my next post.
5. Slightly more accurate results can be found with the formula wins+ ((.4)*loses) + ((.6)*draws) )/(sqrt(loses+1) on the GL server. Those are the numbers I'm going to post.

[xyxaxyz2]

Point values of ranks in Great Lakes:

1. 57.9589 4 0 128

2. 56.1783 13 0 205

3. 53.4442 6 0 139

4. 52.9998 9 0 164

5. 49.1354 6 1 127

6. 48 0 0 48
S
7. 46.1 3 0 91

8. 42.71 6 1 110

9. 39.7217 2 0 68

10. 39.1759 4 0 86

11. 37.3871 4 0 82

12. 37.2976 4 3 80

13. 37.0524 1 0 52

14. 34.4 8 0 100

15. 34.168 6 0 88

16. 34 0 0 34
S
17. 33.6788 26 1 164

18. 33.7333 8 0 98
[[[[ -0.0545676
19. 33.5169 1 0 47

20. 33.5169 1 0 47

21. 32.0062 12 1 110

22. 31.6627 4 2 68

23. 30.9006 7 1 84

24. 30.6789 4 0 67

25. 30.6 3 0 60

26. 29.6985 1 1 41

27. 29.6985 1 1 41

28. 29.5603 11 0 98

29. 29.6 8 1 85
[[[[ -0.0396662
30. 29.5571 17 6 115

31. 29.2742 1 0 41

32. 28.7333 8 0 83

33. 28.4 8 0 82

34. 27.9307 31 1 145

35. 27.5 15 0 104

36. 27.3935 23 0 125

37. 27.3526 5 0 65

38. 27.0691 9 0 82

39. 27.2 3 2 52
[[[[ -0.130903
40. 26.6539 4 0 58

41. 26.3044 17 3 103

42. 25.7595 4 0 56

43. 25.1724 6 2 63

44. 24.9031 5 0 59

45. 24.7 3 2 47

46. 24.4179 19 1 101

47. 24.3333 8 3 68

48. 24.1615 25 2 112

49. 24 0 0 24
S
50. 23.5559 2 0 40

51. 23.3052 13 0 82

52. 23.3445 4 1 50
[[[[ -0.0393694
53. 22.9333 8 1 65

54. 22.8316 9 1 68

55. 22.4927 10 1 70

56. 22.1818 4 0 48

57. 22.0103 10 0 69

58. 21.9333 8 1 62

59. 21.9219 6 1 55

60. 21.496 1 0 30

61. 21.544 6 1 54
[[[[ -0.0479288
62. 21.4086 39 3 118

63. 21.15 15 1 78

64. 21.0667 8 0 60

65. 20.7889 7 0 56

66. 20.7445 9 0 62

67. 20.6 0 1 20
S
68. 20.5613 6 0 52

69. 20.4667 8 2 57

70. 20.2944 14 0 73

71. 20.0818 7 0 54

72. 20 15 0 74

73. 20.0321 6 1 50
[[[[ -0.0321171
74. 19.9911 13 1 69

75. 19.8409 5 1 46

76. 19.7424 16 0 75

77. 19.799 1 1 27
[[[[ -0.05659
78. 19.7791 10 1 61

79. 19.7773 13 3 67

80. 19.6576 17 1 76

81. 19.6364 12 0 66

82. 19.5144 2 0 33

83. 19.5144 2 0 33

84. 19.4985 4 0 42

85. 19.4985 4 0 42

86. 19.4274 6 0 49

87. 19.4667 8 2 54
[[[[ -0.0392928
88. 19.4454 7 2 51

Explanation: first number for reference. Second is point value. Third is losses. Fourth Draws. Fifth wins.

[[[[ means error. Next to this is the difference between the previous number and the number before the previous number. All others are in correct rank.

[Half-Mad]

Originally posted by Furious_Angel
ok so i got this formula that i think is right (wins +1/2(draws+losses))/sqrt(losses+1) and i think it might be the formula or close to it that seed uses and according to this forula i'm the best ranked on GL server (by far) :p if you think this is right or even close post and vote!

Been thinking about this, and you, or Xyxaxyz2 may be correct. Xyxaxyz2 has private messaged me a slight alteration to your calculation. Xyxaxyz2's equations is:
wins+ ((.4)*loses) + ((.6)*draws) )/(sqrt(loses+1)
Only a slight modification to yours, Furious.

I imagine that the discrepancies in the calculation could be from people playing the game while the calculations are happening. First, the system calculates the top 100 list, while people are playing, then after getting that list, reads the stats for each of the people in the list, and assigns those stats to the appropriate ranks on the list.
One of the admins mentioned before that calculating the top 100 list can take an hour, with the huge number of players on the system. With newbies ditching games, it's possible to get several games in during that time.

[xyxaxyz2]

Actually, both are my calculations. :D

[AzN_GuY]

He's telling the truth.

[Furious_Angel]

yea xyx how long that take you to calculate all that n stuff?? yo use excel? or some kingd of program?

[xyxaxyz2]

I made my own program to calculate relationships between wins, losses, and draws. But a lot of it was tedious, non-inspiring guess and check.

[Izzard]

Originally posted by Aro23r
I believe it is something much like, or similiar to this.

(wins + draws) / (losses + draws)

But hey, its just a guess and one that should be utilized...

you cant use that equation because everyone with zero losses and draws would have undefined rankings

[Aro23r]

Sorry then, i edit it to this

(wins + draws) / 1+(draws + losses)

And yet again, i really have no idea. Its just an educated guess. Rank is then determined by whoever has the highest ratio.

[MasterGolem]

maybe its whoever they like best:D

[Furious_Angel]

i still think that the formula that i posted is pretty close...

and i noticed that the rankings STILL havn't been updated i mean wtf is going on?

[ArcPaladinZero]

I kinda stopped worrying about the ranking seeing as how some people just cheat the system and end up with 100 wins and no losses... because it is possible

[Furious_Angel]

number 1 number 1 number 1!!!!!

[xyxaxyz2]

Anyone want to check my formula and see if it still works? I'll do it myself this weekend.

[Furious_Angel]

O_o

[xyxaxyz2]

It isn't over yet.

[Furious_Angel]

its the end of the beginning and the beginning of the end!!!

WOO JUST GOT BACK FROM RETURN OF THE KING

SO BAD ASS!!!

[A.F.I.]

mayhap

but aro23r is definatley wrong

if it was yours then why would sum1 that is

500-1-12
be above 60-0-0

[Crypt-Lord]

Try the new JOKE thread it will realy make you feel good !!
read a good one then leave a good one !!

we realy like blond jokes !!

[S_K_O_F]

ok...all of you guys finding these formulas...I have 2 things to say...first off WOW!!! you must be really bored
And b. everybody suck up to these guys cuz they are probably gonna be our bosses someday!!!

[xyxaxyz2]

Bored? Not really...Its fascinating (sort of. At the best of times). Especially since the last part of the formula still eludes me.

I tested the old formula on the new banff scores (not just for its own sake. I has a higher purpose) and it still works well, but still imperfectly.

[CthulhuXX]

How do you account for lower win percentages being ranked above higher win percentages if not by wins counting for more than losses?

[xyxaxyz2]

How do you account for lower win percentages being ranked above higher win percentages if not by wins counting for more than losses?

This is accomplished by taking the square root of the losses. That means, the more losses you have, the less they matter.

Actually, thats the basis of my formula: wins/square root losses. I just don't know what specifically is neccesary to get it perfect.

[Killer_kev]

Originally posted by CthulhuXX
How do you account for lower win percentages being ranked above higher win percentages if not by wins counting for more than losses?
draws

[xyxaxyz2]

I don't think draws matter so much. I've tried the formula on series of rankings without draws.

[CthulhuXX]

It can't be draws, i think xyxax's square root theory makes more sense because players with no losses or draws can still be ranked lower.

Which is good because obviously a 10-0-0 player shouldn't be ranked number one.

[Furious_Angel]

when will TAO tell us how it is done?!

[Omega_Paladin]

When I change my name to James Bond and I actually feel a chill up my spine when I look at your posts.

[Furious_Angel]

then look at killer_kev lol

[Utamaro]

Haha this is a really really old thread so i declare myself the ghost of old threads. :)

[xyxaxyz2]

Its a useless thread too.

[ArcPaladinZero]

Well yeah, now it is.