AllegSkill Online

KofiMan · Post by **KofiMan** » Mon Jan 19, 2009 11:31 pm

What? It's always going to be long term. One game isn't enough to get a picture of how good a player is.

badpazzword · Post by **badpazzword** » Mon Jan 19, 2009 11:32 pm

Crossposting:

The point is, sigma and mu are not based on some kind of measurement of one player's awesomeness. Sigma and mu are the measurement themselves. The system works like this:

Let's assume everybody is an average player (mu, perceived skill, starts at 25 out of 50) even though I'm not so sure this is a good assumption (sigma, quality of perception, starts highest at 8.33). If we draw the bell you get on the x axis the measured rank and on the y axis how likely it is for that x to be the fairest measure of your skill.

Based on those assumptions, the system does some 'magic' to those bell shapes based on what actually happens -- the game. After each game those 'rank perceptions' are adjusted to best fit what happens -- not to fit them fairly good, actually to fit them BEST.

Since two highly significative numbers aren't that easy to track (the intended audience was xbox teenagers after all) you have to give just one watered down number that's easier to track and that people without a degree in stats can understand. That's the rank, and since the initial mu minus three times the intial rank is zero, you have the magic formula mu - 3sigma = rank.

Rank isn't really supposed to be very meaningful. Allegskill (or at least, the maths engine at its core) never works with the rank numbers. Don't look too hard into them.

As for the increasing/decreasing matter:

Based on a snapshot of data. Pseudorandom sample of the community.

At the beginning you lose sigma *very* fast. This makes it virtually impossible hard to lose rank with early games.

The more you play the less rank you gain from sigma reduction. This means that if you suck your rank WILL actually go down at some point and that if you rock your rank WILL continue going up.

Clay_Pigeon · Post by **Clay_Pigeon** » Mon Jan 19, 2009 11:42 pm

hotnoob,

frankly, your suggestions make no sense, and imply that you really haven't paid too much attention to the articles. You're also talking about things like replacing variance with absolute value of sigma and implying that sigma can be negative (it can't). This suggests that you haven't done a lot of work with statistics either. On top of everything, you're being condescending towards people who have spent a lot of time with this material, and frankly making an ass of yourself in the process. I recommend stopping while you're ahead.

-T

(who has taken plenty of math courses thankyouverymuch)

hunkyhoney · Post by **hunkyhoney** » Mon Jan 19, 2009 11:50 pm

Clay_Pigeon wrote:QUOTE (Clay_Pigeon @ Jan 19 2009, 04:42 PM) hotnoob,

frankly, your suggestions make no sense, and imply that you really haven't paid too much attention to the articles. You're also talking about things like replacing variance with absolute value of sigma and implying that sigma can be negative (it can't). This suggests that you haven't done a lot of work with statistics either. On top of everything, you're being condescending towards people who have spent a lot of time with this material, and frankly making an ass of yourself in the process. I recommend stopping while you're ahead.

-T

(who has taken plenty of math courses thankyouverymuch)

*sighs*
I said that sigma cannot be negative.
but with the help of additional programming you can make it NEGATIVE!

Equations are extrememly limited without the help of additional programming.

By using additional code, you can include a negative and positive value, rather than using an absolute value with only cares about the difference between two or more numbers.
---
look, i give up with trying to explain a more ideal ranking system... that is based less on experience.
but could you ATLEAST! change that dam 3 to a 1! so that my improved game play doesn't make me a freaking Intern 1!

... did u atleast take a look at the php code i posted?
i talk best in programming...

badpazzword · Post by **badpazzword** » Mon Jan 19, 2009 11:58 pm

"Equations are extrememly limited without the help of additional programming."

Whenever you see Alleg's autopilot failing at aligning you in one shot with the green door, you should really think about that line. When you're not trying to do anything "trivial", it's the equations you draw the code from. The course in Automatics (process theory and their control) I'm taking is really interesting on this aspect.

sgt_baker · Post by **sgt_baker** » Tue Jan 20, 2009 12:00 am

Hotnoob,

Plase calm down.

I'm here to entertain every possible grievance with AllegSkill. Unfortunately you're not making much sense at the moment. You're beset by exactly the same contraints as the person who just went from (22) to (14) - me! The rules are the same for everyone. Maths takes no hostages.

I'm quite prepared to discuss this with you once you stopped.... posting random stuff and started making sense.

That is all.

hunkyhoney · Post by **hunkyhoney** » Tue Jan 20, 2009 12:09 am

all rite, i lied... one more try... n then i mite stop playing alleg...

here are some numbers...

3 10 20 4 30
now this is how you calculate standard deviation... aka, sigma // i no the person who created the forumlua knows this, but its rediculous when it happens to a 3 year old account.
3+10+20+4+30 = 67/5 = 13.4 // this is your average.

13.4-3 = 10.4
13.4 - 10 = 3.4
13.4 - 20 = -6.6
13.4 - 4 = 9.4
13.4 - 30 = -16.6

k, at this step, in my forumlua, rather than having the standard deviation an absolute value, they are added.
thus you get, lol, ya get 0! therefore your rank would be based off of 13.4. and to figure out ur rank acording to the equation... 13.4*0.6= 8.04
however, with standard deviation... your sigma would be 10.2. thus your rank would be... 0.6(13.4 -3(10.2)) = -10.32 aka 0.

i belive that one of the main reasons why people do not stay playing this game is because their rank is too low for their skills. therefore, more people will leave the game!
because they will be 0's longer... rather than if got right into the game with a 8.04, because they were good enough!

Thus! by using my equation, RANK WILL MATTER! and there will be much much less stacking.

hunkyhoney · Post by **hunkyhoney** » Tue Jan 20, 2009 12:10 am

Badp wrote:QUOTE (Badp @ Jan 19 2009, 04:58 PM) "Equations are extrememly limited without the help of additional programming."

Whenever you see Alleg's autopilot failing at aligning you in one shot with the green door, you should really think about that line. When you're not trying to do anything "trivial", it's the equations you draw the code from. The course in Automatics (process theory and their control) I'm taking is really interesting on this aspect.

but its the if/else statements that really do the job.

hunkyhoney · Post by **hunkyhoney** » Tue Jan 20, 2009 12:14 am

sgt_baker wrote:QUOTE (sgt_baker @ Jan 19 2009, 05:00 PM) Hotnoob,

Plase calm down.

I'm here to entertain every possible grievance with AllegSkill. Unfortunately you're not making much sense at the moment. You're beset by exactly the same contraints as the person who just went from (22) to (14) - me! The rules are the same for everyone. Maths takes no hostages.

I'm quite prepared to discuss this with you once you stopped.... posting random stuff and started making sense.

That is all.

its not that random.... btw...

sgt_baker · Post by **sgt_baker** » Tue Jan 20, 2009 12:23 am

You're saying you *know* the person who discovered standard deviations?

Anyhow. You're not explaining yourself in a manner which is incompatible with my understanding. <---- READ THIS LINE FIVE TIMES.

Calm down (again). Don't post until tomorrow. Post with a definative and succinct suggestion.

As it stands, I have no idea what, exactly, you're suggesting needs fixing.

****AGAIN ----- READ THIS BIT*****
Tomorrow. Think. Coalesce. Present.
********************************