AllegSkill Online

Correct · Post by **Correct** » Thu Jan 29, 2009 5:53 am

sgt_baker wrote:QUOTE (sgt_baker @ Jan 28 2009, 04:52 PM) Over the entire hitory of the project you've failed to make a single useful point.

History has a "s" in it. Happy?

Post by **Raveen** » Thu Jan 29, 2009 9:05 am

You even managed to get that wrong. It should read:

History has an "s" in it.

Post by **Raveen** » Thu Jan 29, 2009 9:07 am

BTW, any thoughts on my suggestion over on the last page?

Edit: I've made a spreadsheet of my ideas based on the current leaderboard.

Rav Rank uses 30/47.6 to give a rank out of 30

Rav Rank + Finki Rank uses Finki's rounding scheme (more or less)

Rav Rank 2 uses Weeds rank as the top and distributes the rest relative to Weed being 30.

http://spreadsheets.google.com/pub?key=pBQ...Mo0WohTzJgTOVUw

What do you think?

MrChaos · Post by **MrChaos** » Thu Jan 29, 2009 1:50 pm

Raveen wrote:QUOTE (Raveen @ Jan 29 2009, 04:05 AM) You even managed to get that wrong. It should read:

History has an "s" in it.

That is why I call him Awesomes One and why he kinda makes me mess in my shorts just a little bit

finki · Post by **finki** » Thu Jan 29, 2009 3:10 pm

I like Ravs+Finki rank, not only because of the name but it looks well distributed and it lets even weedman reach higher goals.
There is just the problem with minus-ranks at the bottom. What went wrong there? For example Donovon is a -5

EDIT: And the average lies at 8,34. That's Inter1 or? A nice name for the average player.

Post by **Raveen** » Thu Jan 29, 2009 3:26 pm

A newbie who loses a couple of games will lose mu quicker than sigma so their rank drops from 0 to -x. This happens under the current system as well so I guess they're normalised to 0 in some way.

sgt_baker · Post by **sgt_baker** » Thu Jan 29, 2009 3:30 pm

Yeah. With the mu - 3 * sigma setup there are quite a few negative ranks. They're simply being capped to 0 at the moment. As mentioned previously, I've already performed some additional research in this area. There are two things that potentially need solving:

1) Negative ranks - After much manipulation of the figures, using mu - 2 * sigma was the simpest and most elegant solution. This almost eliminated the problem entirely.

2) Rank Distribution - Lots of people are bunched up in the ~Vet1 area. Without revaling the inner workings yet, I came up with a system whereby the names ranks more closely match the obseved distribution of ranks.

More on this later.

Correct · Post by **Correct** » Thu Jan 29, 2009 8:26 pm

I want to say an, but it's not a vowel and a works.

Zapper · Post by **Zapper** » Fri Jan 30, 2009 12:16 pm

When can we have/see boot pr player?
I want more hate....

badpazzword · Post by **badpazzword** » Fri Jan 30, 2009 12:21 pm

No. To TAG, and therefore to the events db, and therefore to AS, there's no way to tell boots from quits because they have dropped from quits because of the hackbot from volountary quits from whatever else kind of quit you might have. All they know is a player left a team, for whatever reason.