Staffa Staffa

Metaverse scoring weirdness

▼

Metaverse scoring weirdness

from Stardock Forums
Something appears to be wrong, either we are no longer
informed on the math behind the metaverse or its very
buggy.
I calculated the top 3 players scores using excel
not counting any decay

Staffa Actual 82479,by math 49559.15
Ray... Actual 74665 by math 42075.42
Ponti. Actual 70495 by math 45902.79

1,291 views 69 replies
Reply #51 from Stardock Forums Top
Looks like I didn't truncate for the last block of scores before adding.

I'll have to take a look at another sample set of scores. The way C and C++ programming works makes truncating really easy but it doesn't always have to happen.
Reply #52 from Stardock Forums Top
IceMole,

Strictly 30 days? I come up with different blocks if it's strictly 30 days.

Block 3 - 4/3 to 4/28 - 9 games
total - 7982
divided - 2660
player - 1862
meta - 2261

Block 2 - 4/29 to 5/28 - 12 games
total - 25654
divided - 7405
player - 5924
meta - 6665

Block 1 - 5/29 to 6/27 - 12 games
total - 118843
divided - 34307
player - 27445
meta - 32591

Block 0 - 100% - 6/28 to 7/27 - 15 games
total - 233560
divided - 60304
player - 60304
meta - 60304

Player Total - 95535 (instead of 97265)
Meta Total - 101821 (instead of 99970)

(no edit: dang it! i didn't get to this page!)

[Message Edited]
Offworld Trading Company
Reply #53 from Stardock Forums Top
Icemole:
A block for July is 31 days. A block for June is 30 days, A block for May is 31 days, etc.

(i.e. Ray's blocks are 27/07-27/06, 26/06-28/5, 27/05-27/04, and 26/04-28/03)?
Reply #54 from Stardock Forums Top
Ok, I'm taking a look at ricbayer.

2 blocks
8893(1)
102440(11)

raw square roots
8893
30886.82

truncations
8893
30886

degradations
8448.35
-

sum without last block truncation
39335.17 -> 39335

sum with last block truncation
339334.35 -> 39334 (right)

So there is definitely truncation of the last block. Ice Mole could you check your calculations? Only yours seem to break the mold.
Reply #55 from Stardock Forums Top
IceMole and Gengsta,

I'm going to end up pulling my hair out if I try to calculate scores while this block question is unresolved. Which one do y'all think it is?

- always 30 days
4/3 to 4/28 - 9 games
4/29 to 5/28 - 12 games
5/29 to 6/27 - 12 games
6/28 to 7/27 - 15 games

- 30 or 31 days (depends on month of submission)
4/3 to 4/25 - 9 games
4/26 to 5/26 - 11 games
5/27 to 6/26 - 12 games
6/27 to 7/27 - 16 games

- 30 and 31 days (depends on month of submission and adjusted per last submission in individual blocks... well Ray's submissions don't have a distinction here)
4/3 to 4/26 - 9 games
4/27 to 5/27 - 11 games
5/28 to 6/26 - 12 games
6/27 to 7/27 - 16 games

And I don't think Ray's Metaverse score is the result of any of these versions. I stand by my "Frogboy doesn't like him" theory. ;)
Reply #57 from Stardock Forums Top
- 30 and 31 days (depends on month of submission and adjusted per last submission in individual blocks... well Ray's submissions don't have a distinction here)
4/3 to 4/26 - 9 games
4/27 to 5/27 - 11 games
5/28 to 6/26 - 12 games
6/27 to 7/27 - 16 games

this one is it, check it.

Ice Mole: Maybe last step is a rounding step?
Reply #58 from Stardock Forums Top
#55 by Citizen KitWarrior - 7/27/2003 9:16:21 PM IceMole and Gengsta,I'm going to end up pulling my hair out if I try to calculate scores while this block question is unresolved. Which one do y'all think it is?
End of quote


Sorry, KitWarrior! Seems my research was directed only toward disproving a direct calendar relationship. It's taking time for me to work with Gengsta's stuff. Hopefully we will get this worked out soon.
Reply #59 from Stardock Forums Top
- 30 and 31 days (depends on month of submission and adjusted per last submission in individual blocks... well Ray's submissions don't have a distinction here)4/3 to 4/26 - 9 games4/27 to 5/27 - 11 games5/28 to 6/26 - 12 games6/27 to 7/27 - 16 games
End of quote


I agree with the game distribution, but I would say 4/28-5/27.
Reply #60 from Stardock Forums Top
Ok rapid calculations
Otakun:

Two Blocks
34200 (2)
84338 (8)

Sqrt and Truncation
24183
29817

Degradation
22973.85
-

Total (Rounding)
52791
Total (Truncating)
52790 - right

Definitely truncating!


Reply #61 from Stardock Forums Top
Have you run my score yourself yet? If I have made a mistake I am more likely to make it again. I agree its not rounding...I'm thinking there is a combination or something like round one step truncate the other.
Reply #62 from Stardock Forums Top
Your calculations check out, but the truncating has worked on every single instance except for yours.

I'll have to check on truncating after the degradation rather than the sqrts.
Reply #63 from Stardock Forums Top
This new method works well for all the numbers I have crunched, Staffa's, Rays, yours, and Otakun's. No point with ricbayers.

Make blocks
Take sqrt (no rounding no truncating)
Degrade (truncation)
Sum

List score should pop out. Up to you to double check, and determine the exactly how blocks are organized.
Reply #65 from Stardock Forums Top
Right now blocks are my primary concern.

Torchmaster @ 16 (player page score):

70%
4/15 4
4/16 2
4/19 2
4/24 3
4/25 159
Sum:170
Sqrt:76
Degrade:53.2

80%
4/27 210
5/21 77
5/21 9
5/21 4
5/23 16
5/23 4
5/23 5
Sum:325
Sqrt:123
Degrade:98.4

90%
6/8 929
6/13 5180
6/18 233
Sum:6342
Sqrt:3662
Degrade:3295.8

100%
7/25 117
Sum:117
Sqrt:117
Degrade:117

Ok, if I understand your block method, it predicts that the 4/25 score should go in the 80% block, right?
However, it seems to fit in the 70% block. Have I understood your method correctly here?
Reply #66 from Stardock Forums Top
hmmm, based on that, there are many possibilities:

Ray's latest block definitely stretches into 6/27 which is the 31st day
Torchmaster's oldest block definitely includes the score on 4/25

The best explanation is that there are more detailed time stamps hidden in the metaverse database. If that is so, the data only rules out my theory and fixed 31-day blocks.
[Message Edited]
Reply #67 from Stardock Forums Top
The best explanation is that there are more detailed time stamps hidden in the metaverse database. If that is so, the data only rules out my theory and fixed 31-day blocks.
End of quote


I have yet to see anything that rules out my theory:
Blocks are 30 days not counting the day of submission. In other words: 0-30 days old, 31-60 days old, 61-90 days old, 91-120 days old, etc.

This meshes well with T-Man's explanation from the score submission problems thread:

The aging works pretty much like this

Scores older than 30 days take 90% of the calculated value.

Scores older than 60 days take 80% of the calculated value

Scores older than 90 days take 70% of the calculated value.

Scores older than 120 days take 60% of the calculated value.

Scores older than 150 days take 50% of the calculated value.

Scores older than 220 days take 40% of the calculated value.

Scores older than 290 days take 20% of the calculated value.
End of quote

Reply #69 from Stardock Forums Top
#63 by Citizen Gengsta - 7/27/2003 9:55:18 PM This new method works well for all the numbers I have crunched, Staffa's, Rays, yours, and Otakun's. No point with ricbayers.Make blocksTake sqrt (no rounding no truncating)Degrade (truncation)Sum
End of quote


Also works for the ones I've crunched today: Code Monkey, Dave Javan, m_rock01, and TorchMaster. This looks good.