64 replies to this topic

[Sparks Murphey]

Lostdragon, on 06 May 2014 - 07:50 AM, said:

Your grasp of statistics is highly dubious. The math you describe does not work with ELO. The only way to get a truly accurate picture of what you are describing would require a more complex model that we probably don't have sufficient data to produce.

Well, yeah, it doesn't work with an Elo rating. I wasn't talking about Elo. I was talking about Win/Loss ratios, and pointing out that the sample size, at 149 games, isn't (how did you put it?) sufficient data to conclude an individual's contribution to ratio.

[Creovex]

Flaming oblivion, on 05 May 2014 - 11:43 PM, said:

Ok so I've found my spot which is a assault missile platform it seems to be where I perform best for now , But I appear to be doing something wrong......

My stats aren't great but judging by the fact I have only played a moth I don't see them as terrible.

So my blr-1 as it currently stands statistically

149 matches played
68 wins 78 losses-0.87
111 kills 101 deaths - 1.10
damage done - 41,084

I've definitely improved my performance compared to how I was doing with my orion , But still i'm finding my w/l and k/d telling two different stories, I always try to do the best I can to contribute as much as I can without putting my self in a position where I die in a blaze of epic failure. Don't know what else I can do.

Forget all of these stats, and do assist/death ratio... then we have some GOOD metrics to work with!

[Adiuvo]

Creovex, on 06 May 2014 - 06:29 PM, said:

Forget all of these stats, and do assist/death ratio... then we have some GOOD metrics to work with!

Since even 1 point of damage counts as an assist, they literally mean nothing.

[Lostdragon]

Sparks Murphey, on 06 May 2014 - 06:20 PM, said:

Well, yeah, it doesn't work with an Elo rating. I wasn't talking about Elo. I was talking about Win/Loss ratios, and pointing out that the sample size, at 149 games, isn't (how did you put it?) sufficient data to conclude an individual's contribution to ratio.

Your idea is correct, your math is not. If it were that simple there would not be companies, books and blogs dedicated to creating mathematical models to do what you are trying to do by adding the number of players in 1 out of 149 games to both sides of the W/L equation. One of my professors in college was Ken Massey, who is known for his Massey Rating project which rates and ranks teams and athletes in sports. His ranking is part of the formula the NCAA uses to determine the rankings of Division I college football teams. He taught me calculus and statistics. I learned more than you ever want to know about how to rank and rate individuals and teams based on metrics like W/L.

Edited by Lostdragon, 06 May 2014 - 07:50 PM.

[Joseph Mallan]

Latest Alt.
100% Fatality
55% Victory

Dying seems to be good for W/L!