Forum
Wertrew is deadliest player for August.
|
skrumgaer wrote
at 3:37 PM, Sunday September 1, 2013 EDT
The best indicator for skill in killing is kills per root game. Here are the KPRG stats for August.
14.52 wertrew 12.57 rednttautology 10.29 michele 10.21 diceptr 9.85 shirahoshi 9.79 mrlee 9.47 Berna 8.99 Noobie 8.63 the die 8.21 jsonproxy 8.18 stakaboo 8.11 marcb 8.10 Beybi_Sistr_Hors 8.07 kdiceplaya! 7.97 kdice 123 7.74 BigDaddyKane 7.74 TriG8 7.70 Rebel9 7.65 greekboi 7.63 beyazguvercinus 7.52 integraI 7.50 OneShot7 7.49 0123 7.43 bcmatteagles 7.40 morrinson 7.24 svran 7.06 ugolino 7.06 gogi 6.84 lipos 6.62 Manfred Singh 6.49 alex_sfb 6.47 twinky 6.39 dr. zoidberg 69 6.36 jf220 6.25 moulue 6.04 musetel 5.90 22 Apr 5.86 Honyo 5.86 rap1d 5.82 joe2me 5.68 Lex Lauher 5.46 tajmtoedaj 5.41 Whitehawk 5.39 ikoko 5.36 Gonna4v4U 5.33 6v16to6111111to6 5.25 Jan Paul Kraemer 5.17 siccannibal1 5.11 javanse 5.09 negramarta 5.06 farq 5.02 bivo 4.87 Kakkap 4.84 Monsanto 4.73 danceswithdices 4.63 peter luftig 4.62 Ninjamonkey 4.59 spman 4.56 masticore 4.56 shatteros 4.55 Chris_Fun 4.51 inipi 4.50 Soromon 4.49 dalius1 4.36 spanky6 4.34 1a2b 4.31 fuzzymcfuzz 4.29 Slinus 4.26 rocka09 4.23 W00PW00PW00P 4.17 PureOdyssey 4.12 {A}Monkey SLayer 4.11 themall 3.92 63 belly 3.85 gontis 3.81 Ledexo 3.78 Nokia3310 3.60 sablja 3.50 ji-jo 3.48 SprintTx 3.42 Aken 3.39 vino_en_carton 3.20 Carloos 2.96 franklyghost 2.79 special k dice 2.70 @jeremywright 2.53 Lil Johnson213 2.50 mattz1 2.36 Voo 2.33 Vermont 2.32 no_name2 2.31 Who'sNutz? 2.17 LemonSong 1.97 TheKnightWalker 1.80 Beer Me! 1.71 Gurgi 1.69 caesar-blue 1.41 jurgen 1.21 GreGGwar 1.00 parsifal |
|
jurgen wrote
at 8:54 AM, Saturday September 7, 2013 EDT I like getting_revolt
|
|
|
jurgen wrote
at 9:07 AM, Saturday September 7, 2013 EDT I wish it was possible to ame a good kill skill stat but it won't be easy
there are so many variables and it's hard to squeeze results from low table players and high table players into one fair evaluation. some people play more high point games. These will more likely be 5 player tables so only 4 shots at a kill, compared to 6 in 7 player games. Also, the really good players usually play less games than robot-like, average players who can easily gather a lot of kills in 400+ lower table games. I'm sure there are at least 5 players who didn't medal last month who will still beat wertrew's "1st". So I'm going to repeat my initial problem with this stat: idc if I'm at the bottom myself but a stat where lots of average players end up on top and lots of skilled killers end up in the middle of the ranks at best can't be a good measure of any Kdice skill. |
|
|
jurgen wrote
at 9:08 AM, Saturday September 7, 2013 EDT ame = made
|
|
|
jurgen wrote
at 9:08 AM, Saturday September 7, 2013 EDT made = make etc etc
|
|
|
skrumgaer wrote
at 11:11 AM, Saturday September 7, 2013 EDT The "average" players in the ranks are drawn from the top 100 point scorers, as are all the other players shown. I did not cull the top 100 of those with less than 35 games like I do for the TAPL because this is not a chi square.
The stat is based on the performance of skilled players against randomized robots. Randomized robots would encounter the same phenomenon about 5k tables being more likely to be 5 player games, etc. (This particular month's stats has one nonrandom bias--the two player tables. Caesar-blue's stat is so low because the average possible kill per player per game in a two player game is 1/2, as opposed to 4/5, 5/6, or 6/7 on the regular tables. Future month's stats will not show this bias.) Groups of robots who have no preference as to whether they sit at 5, 6, or 7 person games at any level of table will exhibit mean number of kills that go as the number of games and variance as the number of games and it is against this number that the actual players are compared. |
|
|
skrumgaer wrote
at 11:15 AM, Saturday September 7, 2013 EDT Also, Caesar-blue did not claim a kill for all his wins. He claimed them only to get to the higher level tables faster and after that he did not need them.
|
|
getting_revolt wrote
at 11:47 AM, Saturday September 7, 2013 EDT "If you encounter one of these players, their threat to you is directly correlated with their position on this list." "This particular month's stats has one nonrandom bias--the two player tables"
I beg your pardon but I think that should be at least two: the two player tables _and_ the number of games. Your chance of getting killed by jurgen and GreGG that month was low because they did not play many games. But considering they killed 4 guys in 8 matches and 5 in 17 the chance of getting killed by them PROVIDED you play at the same table (<-conditional probability) could not have been that low. Suppose I write a script for a robot that achieves 1 kill in 30 matches. That doesn't sound too deadly, does it? But being a robot, soulless and dumb as he is, he could play 10000 games in a month and average around 300 kills in a month. His KPRG would have been around 3. Sure, a lot of players on the lower tables could be killed by the script a few times cuz it plays so many matches. But does it mean it would be a more serious kill threat than jurgen and GreGG if you encounter them @ a regular table? |
|
|
skrumgaer wrote
at 11:55 AM, Saturday September 7, 2013 EDT In regard to the John and Jane Doe competition, it was fair because John paid the penalty for not exploiting the advantage that he had. Why should quitters win?
|
|
|
skrumgaer wrote
at 12:04 PM, Saturday September 7, 2013 EDT One kill in 30 matches is not a serious kill threat because even at the two player tables the average expected number of kills in 30 matches would be 15.
|
|
|
getting_revolt wrote
at 12:13 PM, Saturday September 7, 2013 EDT Your point about John is valid, a good stat should reward a diligent player for their higher than average skill in something, cause that way it can make sure that they had a good kill ratio _because_ of their hunger for blood that is intrinsic to them, and not because they got lucky.
What was not OK with the stat is that it rewarded Jane who did not have any extraordinary talent or advantage. IMHO a good stat should reward you for a consistent _good_ (better than average) performance, but it should probably penalize a consistent bad performance. (E.g. my soulless robot should have gotten a negative score.) |