EPTAligulac.com changelog and feedback thread - Page 19
| Forum Index > SC2 General |
|
Prillan
Sweden350 Posts
| ||
|
slowbacontron
United States7722 Posts
| ||
|
Prillan
Sweden350 Posts
| ||
|
slowbacontron
United States7722 Posts
Little note, why is NaDa said to have left AZUBU on July 16, 2013 on Aligulac? Liquipedia gives no indication that he's even left as of yet. | ||
|
slowbacontron
United States7722 Posts
| ||
|
Jensbond
Sweden54 Posts
![[image loading]](http://gyazo.com/94f7eb8b5aaf7e27fa2984794dcef16e.png) I know theres not much Ranom players around but still theres enough to make a list about them. Currently it kinda feels like ur discriminating agianst Random players. | ||
|
Grovbolle
Denmark3813 Posts
On February 14 2014 21:38 Jensbond wrote: Can you guys add Random to this list? I know theres not much Ranom players around but still theres enough to make a list about them. Currently it kinda feels like ur discriminating agianst Random players. Discriminating? Really? | ||
|
slowbacontron
United States7722 Posts
On February 11 2014 16:41 slowbacontron wrote: Also, the whole $N,one issue popped up in the earnings page again D: D: Why does this error happen, anyway? Aligulac should be able to convert from AUD to USD, right? | ||
|
slowbacontron
United States7722 Posts
For example: see Dear's page http://aligulac.com/players/1659-Dear/period/104/ As of now, 10:36 February 15 (the page will change soon), Dear's overall performance rating is 1881, which is an average of 1869 vP, 1688 vT, and 2086 vZ. However, I think it would be more accurate if his overall performance rating was 1852, which is (1869*5 games vP+1688*12 games vT+2086*8 games vZ)/(5+12+8). Now that I think about it, the practice of taking a straight average between vP, vT, and vZ ratings to form the overall rating has been a staple of Aligulac for a long time. I've generally thought this to be okay since SC2 is made to have 3 races equally represented, but now I'm wrestling with trying to figure out which way is a better representation of a player's overall skill. I mean, overall rating doesn't affect predictions which is a large part of Aligulac's appeal, but it does affect the ranking lists, which is another large part of Aligulac's appeal. Again, I don't even know if weighing the overall average is better or not, but I think it's worth discussion. Side note, jjakji's current overall rating would be 2090 either way :D | ||
|
Grovbolle
Denmark3813 Posts
On February 15 2014 12:03 slowbacontron wrote: D: Why does this error happen, anyway? Aligulac should be able to convert from AUD to USD, right? It happens when on a current day, the currency library doesn't have a conversion rate for the AUD to USD currency. | ||
|
slowbacontron
United States7722 Posts
On February 15 2014 20:36 Grovbolle wrote: It happens when on a current day, the currency library doesn't have a conversion rate for the AUD to USD currency. Interesting, thanks. So would the best solution be to pick the nearest day in which a conversion is recorded? | ||
|
TheBB
Switzerland5133 Posts
On February 15 2014 20:09 slowbacontron wrote: Hey guys, I'm wondering about another thing now. When a player has a "performance" rating for a period, his/her overall performance rating is a straight average of the vP, vT, and vZ ratings but in my opinion it should be a weighted average based on the number of games they played in vP, vT, and vZ each. For example: see Dear's page http://aligulac.com/players/1659-Dear/period/104/ As of now, 10:36 February 15 (the page will change soon), Dear's overall performance rating is 1881, which is an average of 1869 vP, 1688 vT, and 2086 vZ. However, I think it would be more accurate if his overall performance rating was 1852, which is (1869*5 games vP+1688*12 games vT+2086*8 games vZ)/(5+12+8). Now that I think about it, the practice of taking a straight average between vP, vT, and vZ ratings to form the overall rating has been a staple of Aligulac for a long time. I've generally thought this to be okay since SC2 is made to have 3 races equally represented, but now I'm wrestling with trying to figure out which way is a better representation of a player's overall skill. I mean, overall rating doesn't affect predictions which is a large part of Aligulac's appeal, but it does affect the ranking lists, which is another large part of Aligulac's appeal. Again, I don't even know if weighing the overall average is better or not, but I think it's worth discussion. Side note, jjakji's current overall rating would be 2090 either way :D You have a point. Basically only the race-specific ratings matter from a methodical point of view, and taking the average to give the “real” rating rests on the assumption that over a reasonable period of time most people will face a uniform distribution of opponents. For performance ratings, again, the overall one doesn't really matter, but the uniformity assumption can fail. I'll give it some thought. | ||
|
TheBB
Switzerland5133 Posts
On February 15 2014 20:38 slowbacontron wrote: Interesting, thanks. So would the best solution be to pick the nearest day in which a conversion is recorded? We're going to figure out how to fix this. Don't worry.  | ||
|
slowbacontron
United States7722 Posts
| ||
|
Prillan
Sweden350 Posts
On February 11 2014 16:41 slowbacontron wrote: Also, the whole $N,one issue popped up in the earnings page again D: FYI, I just fixed this. Sorry about the delay. | ||
|
Aeh
Germany31 Posts
| ||
|
slowbacontron
United States7722 Posts
As for this next topic I will mention, I apologize because I know it's been discussed many times and I might have even brought it up myself but forgotten the answer. What is going on in cases like this where Soulkey had a vastly higher performance rating (overall performance, not matchup specific) than his original rating but still ended up losing points? Does it have anything to do with online vs. offline games? | ||
|
Grovbolle
Denmark3813 Posts
On February 25 2014 16:56 slowbacontron wrote: Thanks for the earnings page fix! That's super great. One tiny thing I'd like, if possible, on that page is a page counter and turner on the bottom like there is in the ratings rankings. As for this next topic I will mention, I apologize because I know it's been discussed many times and I might have even brought it up myself but forgotten the answer. What is going on in cases like this where Soulkey had a vastly higher performance rating (overall performance, not matchup specific) than his original rating but still ended up losing points? Does it have anything to do with online vs. offline games? You can't really sum the performance like we do, you have to look at each specific case. Just looking at it, you see his ZvZ was so bad that even though his ZvP and ZvT were positive, he still lost points overall. Personally I think you could remove the Expected score vs. Actual score on the last line, since it leads to misunderstandings like this. | ||
|
Prillan
Sweden350 Posts
| ||
|
slowbacontron
United States7722 Posts
| ||
| ||
