also how is skewing in favour of tons of games any better than skewing in favor of very few? Take your example of someone who has played 1 game with 6 kills. Clearly that person gets favoured since it is a 'perfect game'. Just pick a minimum number of games for a decent sample size to eliminate some favour of people who have had incredibly lucky kill streaks. 30-40 is pretty reasonable. It is highly unlikely that someone has had some sort of extra luck that allows him easy kills over that large a number of games.

xchloe:

Since you said you were too lazy to provide the numbers for the full seasons, I was too lazy to look at the actual numbers, so I didn't notice that William's number of games was significantly less than the others' because of the strike. You didn't sort the names, so the two lists were in the same order.

If you take out Williams, the other players line up in the same order in runs per game and runs per root game. Their numbers of games were bunched closer together. Which supports my point that if the number of games were exactly the same, no need to have three different stats.

Kills per root game promotes the playing of more games. If you are in the lead and about to be overtaken by someone who has played a lot more games, all you need is to make a few more kills. Your kills are worth more. Should be no problem to stay in the lead if you have true skill. Sitting on your laurels will cost you.

"Since you said you were too lazy to provide the numbers for the full seasons, I was too lazy to look at the actual numbers"
So did you not read my post at all? I pointed out more than once that Williams had far fewer games than the others, in both textual analysis and in the numbers.

"If you take out Williams, the other players line up in the same order in runs per game and runs per root game. Their numbers of games were bunched closer together. Which supports my point that if the number of games were exactly the same, no need to have three different stats. "
But the fact remains that certain players play far more games than other players. (So again, how is your baseball argument valid? Using home runs/root game favours more games, not less, so it puts the old greats at a huge disadvantage if they played fewer games as you stated.) Is it really fair that jona's kprg is more than twice sam's even though they have a similar kill rate? Okay, so you apparently want to favour people who have no lives, but that large a discrepancy is completely ridiculous. Even though Sam is just about as deadly as jona, he has no hope of ever overtaking jona because he's played a third the number of games jona has.
Also, I'm not calling for three different stats, I'm just calling for use of kpg. My analysis showed data and the two stats in question. I ask again: what's wrong with using kills per game? Kills per game favours those with very few games who have had extreme kill luck (just as tapl/tazd favours those with very few games who have had extremely good starts) so just require 30 or so games and those people are eliminated.

xchloe:

I must have burned out reading your post because I got bored. Why read a post about numbers that aren't any good?

But one thing I was wrong about. Maris did his big thing after the number of games was changed. According to my trusted source of wisdom in all things, Wikipedia, season changes since then have been in the calendar length of the season, not number of games, since fewer double-headers are being played. The old number of games was 154; the new number of games was 162; the square roots of these are 12.73 and 12.41, which will not make a difference unless the players are only one run apart or less.

The tapl/tazd does not require a minimum of 35; only that the Yate's correction be calculated for less than that number, which I have decided is too costly to do.

If I were to use KPG a better minimum would be 5 than 30.

Here are some stats I whipped up with the top 5 on the kill list included.

Games KPRG Player
0510 9.25 WoW 7
1038 8.38 konop
0547 7.95 jona_vicente
0884 7.20 Troy11
0746 6.15 Ogma
0660 5.45 weedalcoholstuff
0352 5.38 Noob Slayer

"I must have burned out reading your post because I got bored. Why read a post about numbers that aren't any good?"
Because I'm trying to prove a point and you're simply not listening? My numbers weren't perfect but they were damn near full season numbers so there's really no reason for you to

KPRG and KPG are not chi-square tests so why on earth would Yate's correction apply to begin with? It most certainly is valid in the case of tapl/tazd but would not be relevant to this statistic. Therefore a 'large sample' would be necessary- generally ~30.

The same people's kpg:
Games KPG Player
0510 0.410 WoW 7
0547 0.340 jona_vicente
0352 0.287 Noob Slayer
1038 0.261 konop
0884 0.242 Troy11
0746 0.229 Ogma
0660 0.212 weedalcoholstuff

Note the very different order. WoW 7 still leads due to a remarkably high kill rate (though the same kill rate with 'only' 250 games would have put him in dead last in kprg). However, konop falls to fourth. Note his ridiculously high number of games.

sorry about the incomplete sentence.
... damn near full season numbers so there's really no reason for you to automatically assume that the numbers are completely invalid. Especially given that Williams had so many fewer games.

u didnt answer my second question skrum

dnr:

Who are you?

xchloe:

KPRG is a non-central chi square distribution. Since kills, KPG, and KPRG result in the same ranking if the number of games is the same, any condition about the number of games for one can be applied to the other two.