Forum
Rap1d and Drucifer85 are deadliest players for March
|
skrumgaer wrote
at 3:03 PM, Tuesday April 1, 2014 EDT
Here are the top 100 ranked by kills per root game.
KPRG Player 13.73 rap1d 13.65 Drucifer85 11.34 hahakaka6 11.34 mar101 11.07 barry gateaux 10.41 Maxin17 10.19 ProxyCheater 10.12 the die 9.39 panzer 9.15 Mork1 8.85 Rafkrik 8.78 michele 8.54 arkenkor 8.43 Holzgranate 8.39 BenJa LeZana 8.23 rocka09 7.57 MariquitaPower 7.56 Smoke Two Joints 7.52 Captain Ramius 7.26 Dengar2 7.19 The AntiTruce 7.19 evilsamurai 7.17 Seljacina 7.01 champion22 7.01 NogalMeaui 6.83 spanky6 6.78 Sabala 6.65 niczrobo 6.52 kakku man 6.50 no_name2 6.47 iDu 6.41 hcdug 6.31 prymus 6.31 djefferson 6.26 Kurwa_pendecho 6.20 1st.skyler 6.20 CumGetSome 6.19 Nerd Alert 6.05 sawwaveanalog 6.04 Yuck_Fou3 6.03 Dude111 5.80 IFIGENIUS 5.70 TParker 5.66 Sian85 5.64 Egazas 5.54 Rebel9 5.46 boysan 5.45 Honyo 5.43 Sensei_Mata 5.34 Hong Kong 5.23 Gauntlet15 5.14 moneymaker$$ 4.98 cicixbalka 4.94 tyla 4.86 Gurg1 4.78 B4M069 4.76 THRILLHO 4.72 wishbone 4.70 madlamb 4.68 deleuje 4.56 kostur 4.47 VinceMorris 4.40 fish28 4.37 moulue 4.35 seeb 4.33 N00bSCH001BU5 4.28 parsifal 4.26 prairiehobo 4.24 ImNotPga 4.16 Bart Lee 4.08 Greenhorn91 3.97 Monsanto 3.88 franklyghost 3.83 lipos 3.41 longpube 3.41 Help, I'm a rock 3.27 Soromon 3.16 elvissey 3.14 SHO-GUN 3.02 dr. zoidberg 69 3.00 Grux 2.97 leea1776 2.85 Melphis 2.67 ehervey 2.67 22 Apr 2.64 Ninjamonkey 2.57 Benjamin Videz 2.44 apignarb 2.39 rushwing 2.32 vino_en_carton 2.26 Zigua 2.24 Lil Johnson213 2.19 spman 2.12 craig blubber 2.03 Marcysch 1.96 PRESIDENT N° 1 1.89 shadolin 1.44 OneShot7 1.00 Voo 0.71 jurgen |
Replies 1 - 9 of 9
|
greeen wrote
at 8:07 AM, Tuesday April 15, 2014 EDT scary
|
|
Louis Cypher wrote
at 6:14 AM, Wednesday April 16, 2014 EDT What is a root game?
And how do you get more than 6 kills per game from a 7-table. Probably, if I knew what a root game was I knew how these stats were made up. If someone stupid like me just devided the number of kills by the number of games he'd come up with a still impessive 0.53 for drucifer. You don't have any kills finishing 7th and hardly any in 6th usually... I think anything above 0.5 is a great value. And since the stats still don't account for anything within tourneys, that's a quite representative value regarding how lethal a player is. |
|
jurgen wrote
at 11:43 AM, Wednesday April 16, 2014 EDT Let's say you got 16 kills in 16 games: your kills per root game (square root of 16 = 4) would be 4
Let's say you got 100 kills in 100 games: your kills per root game (square root of 100 = let me think...oh right...10) would be 10 to get a kill per root game of 10 with only 16 games played, you would have to get 40 kills in 16 games skrum can give you the stat-nerd and superaccurate explanation but it comes down to this: a kill/game avg gets more impressive with increasing number of games played. You can't just compare kills/game from someone who played 16 games and someone who played 100. So you use the square root of #games instead of dividing by #games |
|
|
Louis Cypher wrote
at 5:57 AM, Thursday April 24, 2014 EDT Hey Jurgen, I have to disssagree. I totally can compare kills/game from a player having played 100 games to a player that only played 16. There might be the need for a minimum number of games so you can't stop after 2 deadly ones, but then I can compare it. The problem is that kills are harder to get on higher tables as they are not given (loss of dom). But the table-level is not in that formula (nor would I know how it could).
The used formula rewards large numbers of games. If you used kills/game the maximum value is 6, regardless of how many games youp played. With kills/sqrt(games)the maximum rises (almost 190 with 1000 games,112 for 350 games...). If you had 1 kill per game you'd have 18.71 as your value (9.75 with 0.5 kills/game). You'd be really good in this - but 0.5 kills/game is not deadly, is it? Basically, this is just another number-fiddling to "reward prestigious prices" to people lacking other things to do with their time but playing kdice. How about this one including a percentage of what could have been done? It lacks the minimum - maybe something like 10% of the games the guy with least real life played. Deadliness = (kills/sqrt(games))*100/(games/sqrt(games)) |
|
|
skrumgaer wrote
at 7:12 PM, Thursday April 24, 2014 EDT Louis-
I have made an attempt to adjust for level of table starting from an observation that the distribution of number of tables in play tend to follow the Golden Ratio (1.6something) in regard to the number of tables at each level, with 0 level having 1.6something times the number of tables at the 100 level and so on. Consider a skill X. If skill X promotes a player's likelihood of advancing to higher levels of table, the distribution of players with a particular level of skill X should be distributed among the levels of table at a higher ratio than the Golden Ratio. So correct each player's X by comparing that player's proportion of time spend at different levels of table with the Golden Ratio. For background, look at chapter three of my "collected works of skrumgaer" posted on my Wall. In regard to kills, kills at high tables are likely to occur less often per game, but on the other hand, the points for kills allows a player to ascend through the lower level tables faster, so in a given month a player with more kills may have more shots at being able to play at a 2k or 5k table. |
|
|
Louis Cypher wrote
at 1:17 AM, Monday April 28, 2014 EDT Skrum -
I can agree to all you wrote without problems. But, there is no means known to me to evaluate how many games at a higher table your player X advancing to 2k rapidly by vast amounts of kills on low tables really spend on higher tables. In other threads and on the table there is permanent complains about 20k sitting at 100 and so on. So we have the problem that on the one hand a player going for high ranks leads loads of points and thus, if we ignore the possiblility of making it through tourneys, has to play high tables resulting in less kills. The number of games possible at that level is lower than at 0 or 100 level. On the other hand a player aiming at kills will not go to those tables but stay at tables where kills are more likely. In addition he will have much more games. A player not doing 100+ games month might get to 1 or 2 kills per game which would be quite lethal imho but will never have the chance to reach the values published above. The bad thing in the used formula to me is, that the more games you play, the less kills per game you need to get to high values. That would be fine if players would go to high tables, which they don't. BTW, where do I opt in for that stat? In march I had 64 kills in 119 games - 5,87. |
|
|
skrumgaer wrote
at 5:49 PM, Monday April 28, 2014 EDT The kills per root game stat has a statistical bias because it is calculated only for the top 100 scorers, many of whom may have lots of tourney points. There may be players out there that have higher KPRG's than the leaders here.
On the other hand, kills seem to be as good a predictor of performance in regular games as wins or PPG, as can be seen by looking at the top five scorers in kills, PPG, and wins on the Leaderboard. Kills are a better predictor than dom. But performance in regular games is not a good predictor of total points, because only 15 of the top 100 show up in any of the top 5 groups on the Leaderboard, even if you count some more than once. The other 85 rise through the tourneys. The best I can say about your KPRG is that you seem to be in the middle of the pack. |
|
|
Louis Cypher wrote
at 11:22 AM, Tuesday April 29, 2014 EDT Well, it is good to know that only the top 100 are counted in. For those it probably is a good measure (note, I skip the bashing on too many games with all of them) ;-).
|
|
getting_revolt wrote
at 5:17 PM, Tuesday April 29, 2014 EDT "How about this one including a percentage of what could have been done? It lacks the minimum - maybe something like 10% of the games the guy with least real life played.
Deadliness = (kills/sqrt(games))*100/(games/sqrt(games))" Your formula, as stated here, is redundant and weird. As you probably know sqrt(x)*sqrt(x)=x since this is the very definition of a square root. Then, games/sqrt(games) is simply equal to sqrt(games) and your deadliness formula is equal to 100*kills/games which is just the kill average multiplied by an arbitrary scalar or number of kills per games played times 100. |