So I had a thought. If you took the current h-rating "h" and looked at 2h-1 as a percent, then you get the percentage a player is away from being the "average" player. That is with the adjustment above (an h-rating of a .6 is a 20%), a 20% would correspond to winning 20% more of the games you are in then is expected based on probability. Similarily with negative probabilities. This would be easier to read than the current stat and easier to explain.
Name away or ignore and keep it as is.
The transformation is linear, so it is mostly cosmetic, similar to multiplying by two to make it so that 1 is the middle. I do think there is something psychologically nice about making the midpoint 0 - the wording might be more palatable. However, the +20% still has to be explained as a "gamesize adjusted 20%" - it is not as though you would be winning an additional 20% of your 4-player games with the +20% score. You'd be winning an additional "adjusted 20%", which is fine, but it seems like you'd be trading an edge in one aspect of the wording for difficulties in another aspect of the wording. A sample of how the wording would go might convince me that it is an improvement :)
I like. The numbers look better, and I accept your challenge..
An alternative way to express it that I think better captures its meaning is.. (currentHrating - 1/2)*2.
AND ..if I'm not mistaken, is actually MUCH easier to describe.
=====
The H-score is a statistic that describes the percentage of games you have won above or below the expected norm. It is calculated as follows:
[{Sum (W * O)} / {Sum (L + (W * O))} - 1/2] * 2, where:
W# = Number of wins in #-player games.
L# = Number of losses in #-player games.
O# = Number of opponents you would play if you played one game in that category.
=====
Can it be this simple? If so, this is MUCH better!
I left out all that Gamesize adjusted jargon because anyone that has the ability to question that aspect of it should be able to interpret the calculations well enough to see that it's included.
How about a new setting which allows you to choose what the rating is called?
tom wrote: How about a new setting which allows you to choose what the rating is called?
Great Idea! What should we call it?
M57 wrote:tom wrote: How about a new setting which allows you to choose what the rating is called?Great Idea! What should we call it?
How about Generic Auxiliary Web Personalization?
I was thinking Godlike Access to Wargear Alteration, or General Alternative to Website Appearance
Come on! Who would wanna say "I GAWPed my stats?
The Rankings page has been updated so the G-Rating has been dropped and replaced by H-Rating. I adopted the (Hrating-1/2) x 2 formula as I think that works well. Rating score ties will now be settled by H-Rating, not G-Rating.
Minimum games played has been updated to 10 to get on the Global Ranking scoreboard and you can change this minimum if you wish.
Awesome!
The help file needs to be updated.
Can we include it in our player-stat page too?
I was just noticing that an H-Rating is 0 if you haven't won any games.. Technically it should be -100% (there should be an positive integer in the denominator because of any losses), unless perhaps you are using the old formula.
I don't see how it wouldn't come out as -100 with (oldhrating-0.5)*2
Of course, you could have 0 wins = NA, because -100% is so harsh.
More on the 0 wins Flaw:
Consider a player goes 2/4 in a pair of 2-player games and 0/1 in an 8 player game.
Our oldhscore equation would give us (2+0)/(4+7) = 2/11which translates to a -77% H-Rating after subtracting ½ and multiplying by 2, which hardly seems fair. When the numerator of one of the partials is 0 in a heavily populated game, the denominator kicks in hard.
I'm thinking the oldhscore should probably be something along the lines of (2+3)/(4+7) where 3/7 is the penalty for losing that 8 player game. This would yield a much more accurate -9% H-rating.
On the other hand, the easiest fix is to not include categories with no wins accumulated, but of course there then comes a problem when the number of losses in that category exceeds the expected norm.
I don’t know if this can be done, but perhaps an exception could be coded:
IF #W = 0, THEN the numerator defaults to (#O/2) – #L, and the denominator defaults to #0
Employing this exception to our above player we get a denominator of 7/2 – 1 = 2.5 for the 8 player category
Now we can add things up: (2+2.5)/4+7 and do the conversion dance to get -18%
The only flaw I see with this fix is that it is possible to accumulate an H-Score of less than -100%
For instance if you lost the first 10 of your 2-player games, your H-Score would be -290%, but then we shouldn't have a problem with that.. According to the definition, you have lost 290% of the amount of games you were expected to lose, right?
So 50% is average expected win percentage?
If so how do I have a H-Rating of 37% when my stats ( http://www.wargear.net/players/info/Yertle/Player%20Stats ) display that I am above the average expected win percentage in all #-player game categories?
Same for asm with his 35%.
Now I'm all confused at the H-Rating again.
Updated the Help page, although still really confused about my H-Rating now.
Then how does this H-Rating come out to 72% for a 1vs1 board? http://www.wargear.net/boards/view/Gear%20Wars:%20Episode%20I/Rankings
Thanks for watching my back, Yertle.
Having looked at my stats and Yertle's, I am also confused about how this is supposed to work.
Also G-rating should be replaced everywhere, not just the main rankings page.
Your rating is 0% if you win exactly the expected number of games. This is after I added the 2 x (HRating - 1/2) modifier suggested above. So a rating of 35% means you are winning 35% more games than expected.
At least that is my understanding...
You won 43 out of 50 games = 86%
You were expected to win 25 of them. Let's forget about those = 0 H-rating
You were expected to lose 25 of them.
You won 18 of 25 games you were expected to lose.
You won 18/25 = 72% more games than expected.
I meant to start the debate on this before any sudden and quick changes were made. First, I want to emphasize, that debating between the original formula and transforming it by 2h-1 is not a mathematical one. Mathematically, the same thing is being captured. Do we express the number between 0 and 1, or between -1 and 1 is the question. I'm not convinced the psychology of -1 to 1 is superior.
Yertle's example gives a good illustration of that:
Yertle won 86% of his (two-player) GearWars games. Before the 2h-1 transformation, the H-rating would have been 86%. After, it is 72%, meaning that he won 72% more than the norm. I prefer the 86%, because it is direct, whereas "72% better than average", while understandable, does not express the ratio directly. To think about what it is to perform 72% better than the norm, my mind wants to convert it to "ah, that means I win 86% of the time".