Forum
Anyone wanna do an alternative skill ranking yearly competition?
|
superxchloe wrote
at 11:54 PM, Thursday August 5, 2010 EDT
using a slightly modified version of moondust's alternative skill ranking formula:
((games*PPG)/1000)(4(% 1sts) + 3(% 2nds) + 2(% 3rds) + 1(% 4ths) + 0(% 5ths) - 1(% 6ths) - 2(% 7ths)) Basically, this alternative ranking is a weighted average. In a sense, you get 4 points for every percentage point you have in first, 3 for second, 2 for third, 1 for fourth, none for fifth, and you lose 1 for 6th, and lose 2 for seventh. Then the sum of your points is multiplied by the points you earn at regular tables, and divided by a thousand. The difference between this formula and the original is division by 100000 overall instead of 10000 (because I convert the percentages to decimals), since regular table scores are much larger yearly than monthly. I might also use a different weighting... the current 4/3/2/1/0/-1/-2 I think doesn't take enough for seventh. I propose 5/3/1/0/-1/-2/-3, keeping in mind that bad starts often cause 7ths. Would anyone be interested? I could do top 100 monthly if anyone would want to see that also. |
|
jurgen wrote
at 4:42 AM, Saturday August 7, 2010 EDT This system would totally not work but I sent a mail once to Ryan explaining a system to get rid of many kinds of cheating and with a similar approach to ranks based on skill
In "short": -only 0 tables and tables for people having more than 500 points, you can always play 0 tables without affecting your rank -only "500"tables qualify for rank -minimum number of 50 games required to get ranked (with a coefficent for reaching 50 100 150 200 games - lower game number = lower coëffciënt) -rank based on a comparable formula as (4(% 1sts) + 3(% 2nds) + 2(% 3rds) + 1(% 4ths) + 0(% 5ths) - 1(% 6ths) - 2(% 7ths)) multiplied by the coëfficiënt -random seating: depending on the number of online people, 14,21 or 28 people sit and then get randomly divided over 2,3 or 4 tables --> harder to pga |
|
|
superxchloe wrote
at 9:01 AM, Saturday August 7, 2010 EDT thanks for the plus skrum!
Yeah jurgen, this is meant to be more like the tazd competition than a proposal to change the current point system. A multiplier for more games is an interesting idea though- perhaps since the numbers are relatively small a point boost for each 50 games is more appropriate? I don't mind incentives to play more games, but I don't like weighting the number of games you play heavily either. |
|
|
jurgen wrote
at 9:17 AM, Saturday August 7, 2010 EDT actually it was not ment to be a motivation to play more but more a correction for the fact that the same good 1st and 2nd percentages are more impressive for 200 games than for say a lucky 50 streak
|
|
|
superxchloe wrote
at 9:30 AM, Saturday August 7, 2010 EDT statistically speaking, it's highly unlikely that someone would have a 50 game luck streak. I do agree that maintaining good percentages is far less impressive with a small number of games, but after about 50 games, your percentages are your percentages and they aren't likely to change significantly if you play more because you play pretty much the same way most of the time. A multiplier for more games would seem more like incentive to play more than a statistical correction to me.
|
|
skrumgaer wrote
at 10:51 AM, Saturday August 7, 2010 EDT Chloe:
You're welcome. To all: I would like to go back to the original Elo system, with the modification that rankings are based not on your current Elo score but a moving average of your last 50 or 50 games or so (which requires that you have at least 50 or 60 games to qualify). Also, instead of resetting all the Elo's to 1500 periodically, void all games that are older than, say, three months. Meaning, that after three months, the Elo's of the seven players of that particular game are readjusted as if that game had never been played. In that way, players who have played more games will have a smaller reset penalty. |
|
|
skrumgaer wrote
at 11:26 AM, Saturday August 7, 2010 EDT Correction to my last:
When I said players who play more games will have a smaller reset penalty, what I meant was players who stop playing games will lose a bigger percentage of their games each month than those who keep playing. For players who play roughly the same number of games each month, the penalty would be 1/n of the games each month, sort of like a flat rate tax. Higher scorers would have a higher score penalty but the same percentage. Losers would get an antipenalty. |
|
the full monte wrote
at 1:29 PM, Saturday August 7, 2010 EDT most of us love elo too skrum. sadly i dont think Ryan will ever revert back to it.
perhaps saving the last 3 month's worth of finises for a few thousand people would eat up too much memory...? |
|
|
skrumgaer wrote
at 6:13 PM, Saturday August 7, 2010 EDT monte:
I thought the same thing. It would likely be impossible to create an alternate universe each time a game is voided, because the changed elo's after game 1 is voided would affect seven games that each of those players would participate in next, even if those games weren't up to be voided. There would be a chain reaction of changed elo's throughout the whole population, even before the second game was voided. A simpler system would be to change everyone's elo at once: on the first day of month m, adjust each elo according to the difference of that player's elo in month m - n from 1500. Then you would need to store only n prior elo's for each player. |
|
EInegro wrote
at 6:36 PM, Thursday August 19, 2010 EDT Nerds without life
|
|
|
superxchloe wrote
at 8:23 PM, Thursday August 19, 2010 EDT says he who must have looked through a few pages of kdice forum posts before reading this one, as it is old.
|
