Jump to content

Elo Vs Rpi Player Ratings

elo is bad mkay

46 replies to this topic

#41 BTGbullseye

    Member

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

Posted 27 May 2018 - 08:36 PM

View PostNaqser, on 27 May 2018 - 08:12 PM, said:

Only "exploitable" if you hold firm that choosing and equipping mechs aren't part of your skill.

Accounted for in the RPI.

View PostNaqser, on 27 May 2018 - 08:12 PM, said:

Doing more damage than your opponent yet still lose easily means you spread your damage, and couldn't torso twist good enough.

Also accounted for in the RPI.

#42 Naqser

    Member

  • PipPipPip
  • The Deadly
  • The Deadly
  • 60 posts

Posted 27 May 2018 - 09:49 PM

View PostBTGbullseye, on 27 May 2018 - 08:36 PM, said:

Accounted for in the RPI.

Also accounted for in the RPI.


RPI accounts for your opinion on wether or not choosing whatever mech and loadout to win, is part of your skill?

Feel free to elaborate how RPI differentiate between a high wide damage spread and bad torso twisting, to a low small damage spread with good torso twisting? OP's distribution rate damage quite high, which put quite some emphasis on it.

#43 Surn

    Member

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

Posted 03 June 2018 - 06:14 AM

View PostNaqser, on 27 May 2018 - 09:49 PM, said:

RPI account for your opinion on wether or not choosing whatever mech and loadout to win, is part of your skill?

Feel free to elaborate how RPI differentiate between a high wide damage spread and bad torso twisting, to a low small damage spread with good torso twisting? OP's distribution rate damage quite high, which put quite some emphasis on it.


Actually, damage taken would be a great stat, and I have been asking for it for years now.

With that, RPI would accurately take account of either a damage spread build on an opponent or twisting.

In Solaris this can be calculated

Edited by Surn, 03 June 2018 - 06:16 AM.


#44 Naqser

    Member

  • PipPipPip
  • The Deadly
  • The Deadly
  • 60 posts

Posted 03 June 2018 - 11:04 AM

View PostSurn, on 03 June 2018 - 06:14 AM, said:


Actually, damage taken would be a great stat, and I have been asking for it for years now.

With that, RPI would accurately take account of either a damage spread build on an opponent or twisting.

In Solaris this can be calculated


While it'd be an interesting stat to see.

How would you differentiate between these?

A; Opponent spreads damage on you with you not being good at torso twisting
B: Opponent spreads damage on you, while you are also good at torso twisting
C: Opponent is good at not spreading damage but you spread it out with good torso twisting?

Or are you suggesting the type of weapons on the one doing the damage spreading is taken into account?
At which point, you'd need to differentiate between A and B.

The biggest question I have though. How would you implement all of this into your algorithm? What does it look like now?

#45 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 03 June 2018 - 11:23 AM

View PostSurn, on 16 May 2018 - 08:33 PM, said:


The NAACP, lol you mean the NCAA. No where is RPI declared inferior to Elo, that is just not true. Nothing is perfect, but it is such a superior system, especially when applied to players on teams, that it is confounding that you don't understand it.

Further, because we have far more factors at our fingertips than just w/l and strength of schedule, our representation of w/l can be much more accurate than 75%, which is targeted at a 25 + game sample. Thus, the sos is relative between teams, so small differences still allow a ranking of sos in college basketball.

As to different pools, the flexibility to account for different environments is fundamental to a battle simulation.


Google RPI. Either look at the NCAAs statements on the RPI or even just the Wikipedia entry on it. They acknowledge that it's less accurate than Elo but that keeps it from being gamed for gambling purposes.

The purpose of a matchmaker is to predict who is going to win vs who and get as close to an even match as possible. A win is a win - how you win is irrelevant. The win is what matters.

Again, this isn't new or complex. A matchmaker for online gaming also isn't a new concept. The best is TrueSkill. Google it and it will go over in detail for you how it works both for 1 v 1 and team v team. Bow and why you're wrong has already been pointed out to you. If we spent 2 years sitting down and teaching you statistical analytics and operations research so you understand why you're wrong it's not changing the answer.

This subject has already been researched and answered repeatedly by much smarter folks than anyone here and multi-million dollar solutions created. An accurate MM would be based off win/loss because it's predicting win/loss. The K factor that then adjusts everyones value in the matchmaker (which is different from the MM itself) would look at the win/loss of who you played. That's it. No other metrics are relevant.

#46 Naqser

    Member

  • PipPipPip
  • The Deadly
  • The Deadly
  • 60 posts

Posted 03 June 2018 - 11:38 AM

View PostMischiefSC, on 03 June 2018 - 11:23 AM, said:

Again, this isn't new or complex. A matchmaker for online gaming also isn't a new concept. The best is TrueSkill. Google it and it will go over in detail for you how it works both for 1 v 1 and team v team. Bow and why you're wrong has already been pointed out to you. If we spent 2 years sitting down and teaching you statistical analytics and operations research so you understand why you're wrong it's not changing the answer.


Totally forgot about the TrueSkill system.
If I remember correctly, it's derived from the Elo system, no?

#47 MischiefSC

    Member

  • PipPipPipPipPipPipPipPipPipPipPipPip
  • The Benefactor
  • The Benefactor
  • 16,697 posts

Posted 03 June 2018 - 12:01 PM

View PostNaqser, on 03 June 2018 - 11:38 AM, said:


Totally forgot about the TrueSkill system.
If I remember correctly, it's derived from the Elo system, no?


Yes. Specifically because, not to belabor a point, win/loss is the only useful factor for creating a matchmaker for this kind of environment. Professional sports leagues have a number of factors like betting and divisions to juggle and so can end up using alternative, less precise systems but there's no argument or discussion about what actually works best to create matches based on equal chances to win.





1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users