Jump to content

Rpi/elo Modifiers: Legging Win -2

Modifiers RPI ELO

28 replies to this topic

#21 Brauer

    Member

  • PipPipPipPipPipPipPipPip
  • 1,066 posts

Posted 25 May 2018 - 09:13 AM

View PostSurn, on 20 May 2018 - 01:34 PM, said:

I was discussing ranking systems with a mathematician and found a novel solution that would even work with the worthless Elo ranking in this game.

Have bonus/penalty modifiers to the ranking

For example:

Headshot +2
Destroy Both Legs on Opponent -2
Killed by legging +2
Overheat death -1
Win by opponent Overheat +1

This introduces an aspect to the game in which the community can reward good, honorable players without detracting from those who win at all costs, because sometimes.. you do have to win.


Part of skill in this game is, as I assume you value, aim and positioning. Part of the skill is choosing (or better yet designing from scratch) the right build. Another part of it is selecting your targets, and that includes deciding when it is better to go for legs (Bushies among others) or XL check an IS mech. ELO, in my opinion, should attempt to define the aggregate skill of a player in all of these aspects of the game (landing shots, positioning, selecting build, picking target). Choosing to weigh anything beyond the win in that context will just make ELO a less accurate statistic without adding any value.

#22 Surn

    Member

  • PipPipPipPipPipPipPipPip
  • The God of Death
  • The God of Death
  • 1,073 posts
  • Twitter: Link
  • LocationSan Diego

Posted 25 May 2018 - 12:50 PM

View PostBrauer, on 25 May 2018 - 09:13 AM, said:

Part of skill in this game is, as I assume you value, aim and positioning. Part of the skill is choosing (or better yet designing from scratch) the right build. Another part of it is selecting your targets, and that includes deciding when it is better to go for legs (Bushies among others) or XL check an IS mech. ELO, in my opinion, should attempt to define the aggregate skill of a player in all of these aspects of the game (landing shots, positioning, selecting build, picking target). Choosing to weigh anything beyond the win in that context will just make ELO a less accurate statistic without adding any value.


ELO in a game that is not win/lose in a fair scenario is a bad scoring system fundamentally.

I have an entire other post about scrapping it for RPI (Ranked Percentage Index) which would account for minor things like teammates, mech technology, mech tonnage, damage, kmd, assists, etc... ya know.. THIS game.

And yes, it is technically called Elo... but it is inappropriate in scored games.

Edited by Surn, 25 May 2018 - 12:51 PM.


#23 Naqser

    Member

  • PipPipPip
  • The Deadly
  • The Deadly
  • 60 posts

Posted 25 May 2018 - 01:19 PM

View PostSurn, on 25 May 2018 - 12:50 PM, said:

ELO in a game that is not win/lose in a fair scenario is a bad scoring system fundamentally.

I have an entire other post about scrapping it for RPI (Ranked Percentage Index) which would account for minor things like teammates, mech technology, mech tonnage, damage, kmd, assists, etc... ya know.. THIS game.

And yes, it is technically called Elo... but it is inappropriate in scored games.


But you don't win by score, you win by besting your opponent.
You want to put more emphasis on arbitrary things done in-game, chosen by you, weighted by you, (and added on top of that "Honor gaming" rules) to give players a skill rank.

#24 BTGbullseye

    Member

  • PipPipPipPipPipPipPipPip
  • The Solitary
  • The Solitary
  • 1,540 posts
  • LocationI'm still pissed about ATMs having a minimum range.

Posted 25 May 2018 - 01:42 PM

View PostNaqser, on 25 May 2018 - 01:19 PM, said:

(and added on top of that "Honor gaming" rules)

Those are more of a skill-based modifier to the existing Elo ranking system, and not a part of the RPI system. Because, yes it does take more skill to kill someone with a headshot than it does to leg them.

#25 Brauer

    Member

  • PipPipPipPipPipPipPipPip
  • 1,066 posts

Posted 25 May 2018 - 02:58 PM

View PostSurn, on 25 May 2018 - 12:50 PM, said:


ELO in a game that is not win/lose in a fair scenario is a bad scoring system fundamentally.

I have an entire other post about scrapping it for RPI (Ranked Percentage Index) which would account for minor things like teammates, mech technology, mech tonnage, damage, kmd, assists, etc... ya know.. THIS game.

And yes, it is technically called Elo... but it is inappropriate in scored games.


Thank you for correcting me on Elo.

Since we are talking about Solaris it is more of a binary win/lose situation. In a 1v1 all that matters is if you get the kill first. In a 2v2 there is more room for confounding factors, but they should be less significant than in any game mode outsude of Solaris. As far as I know Elo is only used in Solaris (tiers are used elsewhere I think) so thinking about Elo in context of FW and QP seems irrelevant. I am not an expert on Elo, but your arguments against Elo don't seem relevant to Solaris. I may be misinterpreting here, but RPI seems more relevant for all non-Solaris modes.

#26 BTGbullseye

    Member

  • PipPipPipPipPipPipPipPip
  • The Solitary
  • The Solitary
  • 1,540 posts
  • LocationI'm still pissed about ATMs having a minimum range.

Posted 25 May 2018 - 05:07 PM

Indeed it is more relevant for non-Solaris, but it is applicable in Solaris as well since the rating is used to try and match players together for a fair fight, not just in an "I've won the same number of times as you" way.

#27 Brauer

    Member

  • PipPipPipPipPipPipPipPip
  • 1,066 posts

Posted 25 May 2018 - 06:36 PM

View PostBTGbullseye, on 25 May 2018 - 05:07 PM, said:

Indeed it is more relevant for non-Solaris, but it is applicable in Solaris as well since the rating is used to try and match players together for a fair fight, not just in an "I've won the same number of times as you" way.


I get that, I do think Elo is workable for 1v1 at the very least since it at least is supposed to adjust the rise/fall of your score based on your opponent's Elo relative to yours. To me in a 1v1 all that matters is if I win or lose, getting more components in some contexts can mean I played poorly, so I wouldn't want those types of factors weighted in the formula for 1v1s. I do see the argument for using RPI in QP, at the very least to smooth out the very rough edges of tier based matchmaking.

#28 Naqser

    Member

  • PipPipPip
  • The Deadly
  • The Deadly
  • 60 posts

Posted 26 May 2018 - 02:16 AM

View PostBTGbullseye, on 25 May 2018 - 01:42 PM, said:

Those are more of a skill-based modifier to the existing Elo ranking system, and not a part of the RPI system. Because, yes it does take more skill to kill someone with a headshot than it does to leg them.


And pulling consostent kill headshots will regardless increase your skill rank. You're already getting rewarded for it as long as you win with it, you do not need to add a flat modifier to the gain on your skill rank for that.
Seeing as the last paragraph of the OP was about "honorable players", it is about implementing flat number modifiers on certain win / loss conditions based on "honor gaming rules". I mean, it even has "consolation reductions" for losing both legs.

Doing sick stuff netting you plenty of wins are already accounted for in a high skill rank.

#29 Thrudvangar

    Member

  • PipPipPipPipPipPipPip
  • The Hammer
  • The Hammer
  • 646 posts

Posted 26 May 2018 - 03:30 AM

Legging anything lighter than assaults is no skill in a game where ppl can alphastrike so much and that hight.
And legging assaults is just dumb.

That means imho (and only for solaris DUELS): legging is awfull and requires no skill but guarantees a fast win way too often.

I played 11, yes eleven games the first week solaris was up. I lost the first three for some reasons
then i started going for the legs only and i've won the following 8 games in a row... then i stopped
playing this boring crap.

Edit: i think that gamemode would have much more ppl playing it if legging would give no benefit.
Oh and stock mode eventually :)

Edited by Thrudvangar, 26 May 2018 - 03:33 AM.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users