jarvis_pants
Joined: Oct 30, 2008
|
I like how i went from being the best Halfling Master to not being on this list at all... AND replaced by BillBrasky.
I can only assume therefore that this list is more accurate  |
_________________
"May Nuffle have mercy on your rolls." - St.Basher |
|
Igvy
Joined: Apr 29, 2007
|
 Posted:
Nov 10, 2013 - 12:45 |
 
|
| SzieberthAdam wrote: | | Igvy wrote: | And I've no idea why EvolveToAnarchism isn't on the ranked dark elf list.
Stats -
Record
141/17/28
Win Percentage
80% |
True. He was not on the site for half a year now and not included in coach rankings. Same for Csonti and Orc for example. In fact, this is a list of all-time performances of active coaches. Look forward to the calendar year version. Topic renamed. |
Got it. Good stuff.
Yea just make sure you include all games played in the calendar year version, regardless of activity. |
|
|
BillBrasky
Joined: Feb 15, 2005
|
| Posted:
Nov 10, 2013 - 12:46 |
|
Hmmmmm....
Maybe for my next project (after filling the grid) I should play 1000 fling games in the Box!
Then you would know to bow to your fling Sensai! |
|
|
DukeTyrion
Joined: Feb 18, 2004
|
| Posted:
Nov 10, 2013 - 12:49 |
|
Blimey, I am on more of those lists than I expected to be ...
... and I am not on the one I expected to be on  |
|
|
Overhamsteren
Joined: May 27, 2006
|
| Posted:
Nov 10, 2013 - 13:04 |
|
You are only masters of evil!!! |
_________________
Like a Tiger Defying the Laws of Gravity
Thanks to the BBRC for all the great work you did. |
|
uzkulak
Joined: Mar 30, 2004
|
| Posted:
Nov 10, 2013 - 13:12 |
|
When you do the annual list, it might be an idea to drop the games played requirement considerably.
50 games a year with one race is a huge number for most coaches. |
|
|
Igvy
Joined: Apr 29, 2007
|
| Posted:
Nov 10, 2013 - 13:22 |
|
There will be about 120,000 games played in the year.
50 should be good enough for the top to emerge.
I gotta say, if you haven't played 50 games with a race in a year, you probably don't qualify. You need at least 50 games to cut out the randomness of good runs too.
(For all time, i'd suggest upping that imo (80-100?), and too be fair it should probably cap at the most recent 100 games they played with that race) |
|
|
jdvmat
Joined: Nov 28, 2008
|
| Posted:
Nov 10, 2013 - 15:09 |
|
Good job, nice criteria
However it seems not many coaches pass the 50games requirement with goblins, halflings, underworld and vampires. |
|
|
BooAhl
Joined: Sep 02, 2004
|
| Posted:
Nov 10, 2013 - 16:16 |
|
Found 2 84s, 1 85 and 2 87s win %.
Endzone has two 85+ i Box which is impressive. With Orcs and WEs.
Then Vaclavs 84 % in Ranked with CD was also impressive since he plays many Smacks (just beat him in a final).
The other two, was DE with the Legend, 1200+spp, blitzer and some stay as low as possible Pact in box.
No 90+%? Maybe goal for my second account...
Edit: those Box flings and gobbos that are on the list... thats just amazing |
|
|
happygrue
Joined: Oct 15, 2010
|
| Posted:
Nov 10, 2013 - 16:41 |
|
| Igvy wrote: | There will be about 120,000 games played in the year.
50 should be good enough for the top to emerge.
I gotta say, if you haven't played 50 games with a race in a year, you probably don't qualify. You need at least 50 games to cut out the randomness of good runs too.
(For all time, i'd suggest upping that imo (80-100?), and too be fair it should probably cap at the most recent 100 games they played with that race) |
I would agree that at least 100 games is a nice round lower bound. What about taking into account the coach's total record, and giving the last 100 games double weight?
I think that an annual cap of 50 is a bad idea, as I don't see how it's related to the overall goal. If you activate 4 races per draw and play 3 games a week in the box you probably will miss this cap even with a 100% win rate (just to use a simple example). If such a "current status" is needed, perhaps just a check that 10(?) games with the race have been played in the last year. 50 is fine for folks who stick to just a few races, but is terrible for those who like to play a wide variety. |
|
|
the_Sage
Joined: Jan 13, 2011
|
| Posted:
Nov 10, 2013 - 16:48 |
|
#1 liz coach? Definitely picker list. =) |
_________________
Content: Twitch / Youtube ; Updates: Facebook / Twitter
(because big banners are compensating) |
|
SzieberthAdam
Joined: Aug 31, 2008
|
| Posted:
Nov 10, 2013 - 17:43 |
|
The next version will be based on match results instead of the coach rankings page. But that requires some work and time (not much though) and I am planning to do it after end of year.
I will make all-time CRP and annual CRP. Surely the latter will have lower match cap, perhaps 26 (played at least every two weeks). 100 for the all-time seems fair if we will have the other one too.
I am also thinking of weighting games somehow, especially to adjust higher weights for Tournaments. Lowering the weights of old matches is a good idea, however I would rather make it date based. |
_________________
   |
|
Igvy
Joined: Apr 29, 2007
|
| Posted:
Nov 10, 2013 - 22:51 |
|
| SzieberthAdam wrote: | The next version will be based on match results instead of the coach rankings page. But that requires some work and time (not much though) and I am planning to do it after end of year.
I will make all-time CRP and annual CRP. Surely the latter will have lower match cap, perhaps 26 (played at least every two weeks). 100 for the all-time seems fair if we will have the other one too.
I am also thinking of weighting games somehow, especially to adjust higher weights for Tournaments. Lowering the weights of old matches is a good idea, however I would rather make it date based. |
Sounds great.
Just be careful by lowering it too much. You will get coaches who have only played 26 games and had a massive luck streak. They will at the top of the list, anything under 35 is probably too few.
(Just as info my best luck streak lasted 20 games, and i've had that in my mere 500 games, then a limit of 26 feels like it will only show the lucky ones, not the skilled ones. |
|
|
OenarLod
Joined: Jan 26, 2011
|
| Posted:
Nov 11, 2013 - 12:57 |
|
I think you should find a way to put in soft constraints instead of hard criteria, otherwise results may be a lot volatile.
For example my case, humans in the Box. Not obsessed with it, using my case just to show an erratic behaviour in hard constraints.
Two days ago, I was 207/118/96 in the Box with humans, W% is 63.18%, so I'm out of the list because the hard criteria is W%>63.19%. But my score, given your method, would have been 2412, second place with 250 point lead on the third one.
After two wins, 207/118/96 and W% is 63.36%, so I'm in. And I jump in second place with 2427 points.
A calibrated logistic function may be better than an heaviside one. It may give more smooth results and easier interpretation for the ones who care about rankings. |
_________________
Join the Human League Premiership! |
|
SzieberthAdam
Joined: Aug 31, 2008
|
| Posted:
Nov 11, 2013 - 13:04 |
|
@OenarLod: Thanks for pointing on this. You mean THIS model? |
_________________
|
|
|
|
Finishing the 60 Gam...
Search
