Alpha wrote:

My first ALN.  Thanks, and by the way I support calling it Hugh.

But it's gotta be an acronym..

HUGH:  Highly Untested Gaming Heuristic

Works for me.  You know what they say.  One man's HUGH is another man's GAWP.

I added an acronym to my previous post before reading yours.

Hugh's Un-Gainly HUGH?

How about
Specially Calculated Hugh Win Average Rescaled To Zambezy
(a stretch, but z's are hard)
or
Wargear Earned Elimination Percentage

Actually, I am all for GAWP or GAVA as these actually say what the stat is.

kind of funny how quickly this thread devolved once the math problem was solved.

OK, let's get it back on track:

I figured I’d try my hand at how to put this on the help page.  Anyone please feel free to plagiarize any of my writing if they want to give it a go.

I worked with this formula, from 11’s post

 Sigma [ ((players per game - 1) * wins) / (((players per game - 1) * wins) + losses)]

 I think it can be simplified into something like this:

 (#Opps*#Wins)/(#Opps*W+Losses)

 Sample Blurb:

In a two player game, the probability of a given player winning can be expressed as a number that ranges from 0 to 1, where .5 means that player will win half of their two player games.  A probability of .625 means that player wins 5/8ths of their games. The HUGH/GAWP is an experimental statistic that attempts to express the probability that you will win any given game.  Obviously, the larger the number of opponents you face in a game, the lower your probability of winning.   The YAMS/GAWA formula takes this into account by normalizing or “Gamesize Adjusting” those larger games, and creates a number that still falls into the 0 to 1 range (as if all of your games were two player games), but yet is informed by all of the games that you have played, no matter what their size.

Here’s a list of the possible names, with a few I’ve added.

H-Rating:

 HUGH: Highly Untested Gaming Heuristic

HUGH: Heuristic Used for Gradation of Humans 

GAWP: Gamesize Adjusted Win Percentage

GAWA: Gamesize Adjusted Win Average

GAVA: Gamesize Adjusted Victory Average

YAMS: You Are Makingthis Shitup, aren’t you 11’s?

NWP: Normalized Win Percentage

HUWP: Hugh's Untested Win Percentage

HEWP: Hugh's Experimental Win Percentage

My favorites:  NWP, GAWP, and HEWP

 Vote early – Vote often.

It stands for Yet Another Meaningless Statistic. Oh, and you forgot:

Hugh's Un-Gainly HUGH

The point of that being that it is recursively recursive.

That being said, I vote in this order:

H-rating
GAVA
GAWP

GAVA gets the nod over GAWP only because it looks slightly cooler.

Since I have some experience writing things for challenged individuals I'll provide a potential blurb. I do not claim to fully understand the calculation of the statistic nor do I claim to be a very successful writer in general.


The H-rating attempts to equalize a player's performance for games of various sizes. It is expected that a player will win 1/2 of their 2 player games, 1/3 of their 3 player games, 1/4 of their 4 player games and so on. A player with an H-rating of .5 is doing as expected. [Could end here 1.] The closer an H-rating is to 1 the better a player is doing at winning games of various sizes. [Could end here 2.] For example it is better for your H-rating to win 30% of your 10-player games than to win 30% of your 3-player games since you are expected to win 10% of your 10-player games but 33% of your 3-player games.

I like H-Rating or H-Score because it's both a non-obvious nod to Hugh's role in creating the metric as well as 1 better than G.

If we go with GAWP or GAVA then presumably needs to be expressed as a Percentage.

Preferences there? Express as per the G-rating or as a %?

Also retain G-rating or blow it away?

I see no reason to keep G-rating.

Ditch the G-rating.  This replaces it.

You're right about the number..

Call it a GAWPercentage and the number must fall between 0 and 100

Call it a GAWAverage and it should be a decimal (as it is now)

Call it a HugHeuristic and it can be anything you want.

Call it a YAMS and I'll go to computer school and learn how to hack into your server.

BTW, shouldn't it be Game Adjusted Wins Average?

I agree that GAWA and GAVA beat out GAWP in the sounds cool dept so I was coming around to switching my votes to those, but then I took a quick look at the Rankings page and started to think in terms of what would make the most sense to the average viewer.

If, right after Won | Lost | Win% | you had Gamesize Adjusted % or Gamesize Adjusted Win% in the very next column, the number would make perfect sense.  They'd see a percentage in both columns and the one on the right would be a higher percentage, kind of like seeing a batting average followed by an on-base (including walks) average in baseball.  In fact, if you put it right after Win%, the number might speak for itself and you can use the acronym, but it really should be a GAWP.

You could make it  Gamesize Adj. % or possibly Normalized Win%

Thing is, the phrase "Gamesize Adjusted" says it all.

I say GAWP and combining M57 & Amidon's write-ups i got:


======================================

The GAWP is an experimental statistic that attempts to show a player's performance for games of various sizes by expressing the probability that you will win any given game. Obviously, the larger the number of opponents you face in a game, the lower your probability of winning. The GAWP formula takes this into account by normalizing, or “Gamesize Adjusting”, those larger games and creates a number that falls into the 0 to 1 range no matter what their size.

It is expected that a player will win 1/2 of their 2 player games, 1/3 of their 3 player games, 1/4 of their 4 player games and so on. In a two player game, the probability of a given player winning can be expressed as a number that ranges from 0 to 1, where .5 means that player will win half of their two player games. A probability of .625 means that player wins 5/8ths of their games. A player with an H-rating of .5 is doing as expected. The closer the GAWP is to 1 the better a player is doing at winning games of various sizes.

The GAWP is calculated with the following formula:

 (#Opps*#Wins)/(#Opps*W+Losses)

For yet another example: it is better for your H-rating to win 30% of your 10-player games than to win 30% of your 3-player games since you are expected to win 10% of your 10-player games but 33% of your 3-player games.

======================================

now you may need to adjust the "0 - 1" to "0 - 100" as it's a percentage but the essence is there for you smert guys to play with.

I think... H-Rating and keep it as 0-1 (or 0-2 with 1 being average) and get rid of G-Rating... but I'm also a bit unsure if my opinion matters to myself even :p

I think we need to take care not to be too lengthy or technical.  Just explain what it is and let the number speak for itself.  

======

The Gamesize Adjusted Win Percentage (GAWP) is a statistic that attempts to express the probability that you will win any given game with respect to the expected norm. Expressed as a percentage, a 50% GAWP means you are expected to win half of your two-player games and a GAWP of 62.5% means that you are expected to win 5/8ths of your two-player games.   Obviously, the larger the number of opponents you face in a game, the lower your probability of winning.  The GAWP takes this into account by normalizing or “Gamesize Adjusting” those larger games by taking the results of all of your games and applying the formula:

(#Opps*#Wins)/(#Opps*W+Losses)

This creates a number that is informed by all of the games that you have played, no matter what their size.  For example, if you play nothing but five-player games, and you win 1/5th of them (as would be normally expected), your GAWP would be 50%.

======

You could nix the last paragraph and this description might still hold water.

is this an expected or an actual? it reports how you've played and not what you're expected to get right?

weathertop wrote: is this an expected or an actual? it reports how you've played and not what you're expected to get right?

It could be interpreted as expressing a probability (expected) based on (actual) past performance*.  But I see what you are getting at.  Yes, technically it's just a stat.  We could get rid of the probability language if it's too leading.

*Past performance is no guarantee of future returns

k, how bout this:

=================

The Gamesize Adjusted Win Percentage (GAWP) is a statistic that attempts to show a player's performance for games of various sizes using the probability that you will win any given game. Obviously, the larger the number of opponents you face in a game, the lower your probability of winning. The GAWP formula takes this into account by normalizing, or “Gamesize Adjusting”, those larger games and creates a number that falls into the 0 to 1 range no matter what their size by applying the formula:

(#Opps*#Wins)/(#Opps*W+Losses)

Expressed as a percentage, a 50% GAWP means you are expected to win half of your two-player games and a GAWP of 62.5% means that you are expected to win 5/8ths of your two-player games. 

For example, if you play nothing but five-player games, and you win 1/5th of them (as would be normally expected), your GAWP would be 50%.

=================

tho i'm tempted to take out the "two-player" that i have striked out above...

I really think that opening sentence gets at the heart of the matter and is much better stated than the one in your example (no offense) and i really dislike the "This creates a number that is informed by all of the games..." so i took that out too.

weathertop wrote:

k, how bout this:

=================

The Gamesize Adjusted Win Percentage (GAWP) is a statistic that attempts to show a player's performance for games of various sizes using the probability that you will win any given game. Obviously, the larger the number of opponents you face in a game, the lower your probability of winning. The GAWP formula takes this into account by normalizing, or “Gamesize Adjusting”, those larger games and creates a number that falls into the 0 to 1 range no matter what their size by applying the formula:

(#Opps*#Wins)/(#Opps*W+Losses)

Expressed as a percentage, a 50% GAWP means you are expected to win half of your two-player games and a GAWP of 62.5% means that you are expected to win 5/8ths of your two-player games. 

For example, if you play nothing but five-player games, and you win 1/5th of them (as would be normally expected), your GAWP would be 50%.

=================

tho i'm tempted to take out the "two-player" that i have striked out above...

I really think that opening sentence gets at the heart of the matter and is much better stated than the one in your example (no offense) and i really dislike the "This creates a number that is informed by all of the games..." so i took that out too.

We're getting there.  However..

I see the P word.

"%" doesn't need to be defined.

GAWP doesn't "attempt" to show performance.  Past performance is a given.

===

The Gamesize Adjusted Win Percentage (GAWP) is a statistic that expresses your win percentage with respect to the expected norm. Obviously, the larger the number of opponents you face in a game, the lower your probability of winning. The GAWP takes this into account by normalizing or “Gamesize Adjusting” those larger games by taking the results of all of your games and applying the formula:

(#Opps*#Wins)/(#Opps*W+Losses)

Expressed as a percentage, a 50% GAWP means you are expected to win half of your two-player games and a GAWP of 62.5% means that you are expected to win 5/8ths of these* games. 

Because the stat is gamesize adjusted, if you play nothing but five-player games and you win 1/5th of them (as would be normally expected), your GAWP would be 50%.

===

* "these" are still 2-player games.