That has merit Ferret!
...
...
That rhyme was not intentional!


Average Score Or Cherry Pick?
Started by Inyc, Apr 30 2013 09:24 PM
23 replies to this topic
#21
Posted 02 May 2013 - 06:11 AM
#22
Posted 02 May 2013 - 08:57 AM
FerretGR, on 02 May 2013 - 06:08 AM, said:
Why not an average of score/damage/whatever in winning games only? This eliminates the DCs/glitches/what have you, since your team likely loses when they're down one. Not perfect but perhaps workable. Set the low limit for average calculations at some value that'll give you reasonable results, ie. 15 or 20 wins.
I just don't know if there is a perfect system. Especially when PGI is attempting to have a solo tournament in a team based game, and it isn't really a tournament but a mini-leaderboard.
You suggestion may be open to even more exploitation. A Merc Company/House with 16 members arranges a sync drop and maniuplates the mech selection (and maybe team membership) to improve their odds of dropping together.
Team(A) goes AFK while Team(

Edited by Agent 0 Fortune, 02 May 2013 - 08:58 AM.
#23
Posted 02 May 2013 - 09:23 AM
Agent 0 Fortune, on 02 May 2013 - 08:57 AM, said:
Team(A) goes AFK while Team(
exploits whatever game mechanics are in place to get top score and win the game; next game reverse roles. Do that X time and now your whole Merc Corp is at top of the leaderboards.

Well, that can be done with the current system, no?

I'd be down with whatever system works, TBH. This tournament's scoring is certainly better than some of the previous ones. They're getting closer. I suspect nothing is exploit-proof.
#24
Posted 02 May 2013 - 07:09 PM
FerretGR, on 02 May 2013 - 06:08 AM, said:
Why not an average of score/damage/whatever in winning games only? This eliminates the DCs/glitches/what have you, since your team likely loses when they're down one. Not perfect but perhaps workable. Set the low limit for average calculations at some value that'll give you reasonable results, ie. 15 or 20 wins.
Yea that's more or less what I just suggested. Though I was wondering if it would be best to also have to way to "opt out" once you're happy with your average. Like opt in, play 20~30~40 games, feel like you've got a solid average with this, then opt out.
4 user(s) are reading this topic
0 members, 4 guests, 0 anonymous users