May 12, 2008
12:57 PM | |  ANNOUNCEMENT:
;
|
If you are a coach: login, click "Teams" tab, then "rankings", click the link to vote, read the info and then go ahead and vote.
I need some votes in there before I can complete the display on the main rankings page.
Site should recognize the level your team plays in and present you only with a ranking chart for that level. It will also recognize and adjust for double voting (voting for a team in more than one slot), and short voting (not filling out the entire card). You can short vote, there is no problem with that, but the overall weight of your vote card diminishes.
Once the coaches poll is set and working as intended, I will enable the players poll.
|
| |
|
May 12, 2008
3:01 PM | | | ; |
Okay, players poll is up now.
If you are a player with an account (set up by your coach rostering you): login, click "Teams" tab, then "rankings", click the link to vote, read the info and then go ahead and vote.
| |
|
Practice Squad
asst coach | 
|
May 13, 2008
4:28 PM | | | ; |
So the numbers are basically based off of where a coach player ranks you out of X number of teams? So a first place vote in a 13 team league is 13 points? And then that number is applied to the percentage of voting players/coaches?
| |
|
May 14, 2008
9:57 AM | | | ; |
So a first place vote in a 13 team league is 13 points?
Only if the coach or player registered 13 votes. If he only voted for five teams (only entered a top 5), then his first place vote is worth 5 points.
And then that number is applied to the percentage of voting players/coaches?
Ranking Poll Average (rpa), is the average of the weighted rankings of all three polls. If there is 50% participation in any given voting poll, then that poll carrys equal weight to the computer poll. More or less than 50% and the weight is proportional.
For example:
Team America is #1 in the Novice computer rankings, so they start with "1". They are #2 in the coaches poll. Coaches poll has 38% participation. We double that number to get weight - 76%. (remember that at 50% poll carries equal weight to computer, so if participation was at 50% poll weight would be 100%). Back to 76%, multiply this by their #2 ranking and you get "1.52". Add to the "1" they get from computer rankings and the sum is "2.52". In the players poll they are ranked #1. This poll currently has 7% participation - or a weight of 14%. Apply 14% to #2 rank and you get ".28". Add to "2.52" and we`re at "2.80".
The maximum rank possible:
(#1 * 76% for coach poll) = .76
+(#1 * 14% for players poll) = .14
+(#1 * 100% for computer) = 1
-------------------------
="1.9"
Divide the Team America sum by the max possible rank:
(2.8)/(1.9)
=1.47, which is their rpa.
=
| |
|
May 14, 2008
10:00 AM | | | ; |
I should note that I just finalized the rpa calculation routine this morning. Before that, it was still showing the raw sum of all three polls, not the actual rpa.
| |
|
Practice Squad
asst coach |
|
May 14, 2008
10:14 AM | | | ; |
Not to be stupid but...
More or less than 50% and the weight is proportional. (does html work?)
Proportional how? So if there`s only 1% participation, it is doubled and a #1 rank would be worth .02?
| |
|
Practice Squad
asst coach |
|
May 14, 2008
10:23 AM | | | ; |
Ok, I get it now. Basically, if the participation != 50%, then you double the % and multiply by the rank for coach and player polls. Then add that to the computer rank and divide by the max rpa.
| |
|
Practice Squad
asst coach |
|
May 14, 2008
10:33 AM | | | ; |
You did make a mistake in our example though. If TA is #1 in the Players Poll (which they are) then their number form there is .14 not .28, which gives us 2.66 as a total, divide by 1.9 and you get 1.4 is their correct RPA.
| |
|
Practice Squad
asst coach |
|
May 14, 2008
10:47 AM | | | ; |
Oh, and I know this is annoying, but how many significant digits? Just one after the decimal? Because I calculate the Wolves` RPA to be 1.025 or 1.03, but it is listed as 1.1? Even with one decimal place that`s a rounding error, right?
Max RPA = (.26*2) + (.01*2) + 1 = .52 + .02 +1 = 1.54
Wolves` = ((.26*2)*1) + ((.01*2)*3) + 1 = (.52*1) + (.02*3) + 1 = .52 + .06 + 1 = 1.58/1.54 = 1.0259 round to 1.03
| |
|
DFA
Chuck Norris #21
May 14, 2008
11:37 AM | |
May 14, 2008
11:39 AM | | | ; |
Yes, that was a mistake above I made as Team America should get .14 from the players poll.
Thanks for the other heads-up (fixed now), I had it adding .05 and then truncating to one decimal place, which is okay for getting the % but not for the rpa calc. I cannot round in the coding but SQL rounds automatically when you insert.
| |
|
Al Bundy
Practice Squad #72
May 14, 2008
11:50 AM | | | ; |
haha bored Bundy?
No, I just have serious case of ADHD which means I get easily......what were we talking about?
| |
|
Question?
May 14, 2008
2:23 PM | | | ; |
Should teams that are tied in a poll receive the average of the tied ranks (i.e. two teams tied for first place receive "1.5" rather than "1")?
Just a curious question ...
| |
|
Practice Squad
asst coach |
|
May 15, 2008
10:15 AM | | | ; |
You mean if they`re both ranked first overall, they each get half a point instead of a whole point? Or for third place they`d get 1.5 instead of 3?
That wouldn`t work in this system, at least not all of the time. The issue becomes, for example, if you use 1.5 for a third place team, that`s less than the 2 points you would use for a second place team.
For example, assuming 50% participation across the board (so no weighting essentially, so MaxRPA = 3)
Team A is ranked second in all the polls, so their RPA is 2.
Team A RPA =(2+2+2)/3 = 2
Team B is tied for 3rd in the Coach`s poll, third in the player`s poll and first in the computer rankings. Because of the tie, we use the 1.5 instead of the 3 they would usually get.
Team B RPA = (1.5 + 3 + 1)/3 = 1.83
Which means they`ll be ranked overall ahead of Team A, whereas if they have the 3 points vs the 1.5 their RPA becomes 2.33.
Team B RPA(norm) = (3 + 3 + 1)/3 = 2.33
So, in that situation Team A is clearly the second place team, but Team B would get the nod over them because of the points split.
| |
|
Follow up
May 15, 2008
10:45 AM | | | ; |
To state things more clearly, I was thinking it may make sense to say if two teams tie for 5th (Current C Coach`s poll, for example) they currently both get 5 factored into their rank.
However, they are really tied for 5th and 6th because teams after them start at 7th. Maybe they should both have the average of 5 and 6 factored into their rank (5.5 each).
If 3 teams were tied, theoretically, for 5th, they would truly represent 5th - 7th in the standings. Maybe the average, or 6, should be factored into their rankings.
This is only a minor nit pick that I bring up because I thought the overall rank should average out to the same amount as 1 + 2 + 3 + ... + 13. The numbers come up a bit short because teams that tie currently get a slight boost.
| |
|
Al Bundy
Practice Squad #72
May 15, 2008
1:01 PM | | | ; |
I understand your question better now, but I don`t see the advantage of that versus the way it`s handled now.
| |
|
Numbers
May 15, 2008
1:53 PM | | | ; |
The overall rank would be slightly more accurate/representative of where teams stand in comparison to others.
Based on the current C rankings with 5 coaches and 7 players voting, the teams that are tied in a poll would have their overall numbers move up slightly if the suggestion is implemented. That would mean TA and CN since they are tied in the player poll, and WP and TC since they tie in the coaches poll.
Overall Rank:
#1 1.44 TA (slight increase)
#2 1.56 CN (slight increase)
#3 3.47 QH
#4 3.53 Midgets
#5 5.20 WP (slight increase)
#6 6.39 TC (slight increase)
#7 6.69 RB
#8 8.26 USIS
#9 8.60 Collision
#10 9.86 PS
#11 11.07 WTF
#12 11.93 Prowlers
#13 13.00 DW
| |
|
Question
May 15, 2008
2:55 PM | | | ; |
Would you like a cup of shut the f up?
| |
|
bundy
May 15, 2008
8:26 PM | | | ; |
you need to find other things to do.
| |
|
Al Bundy
Practice Squad #72
May 15, 2008
10:56 PM | | | ; |
I have plenty to do, thanks :)
| |
|
bundy
May 20, 2008
10:31 PM | |
Al Bundy
Practice Squad #72
May 20, 2008
11:45 PM | | | ; |
Sorry, is the math still confusing you? :)
| |
|
|