Meta Melee State Power Rankings

Kaiza

Smash Journeyman
Joined
Dec 21, 2013
Messages
227
Location
South Australia, Australia
#82
the thing about it is that you'd have to keep a PR panel approved rankings and a points/elo/garpr ranking seperate imo. They're fine to have to just look at numbers but they need to be exclusive to each other
 

OninO

Smash Journeyman
Joined
May 19, 2014
Messages
289
#85
Yeah double elim definitely means you need to run more tournaments to get concrete numbers, which is kind of made worse by seeding which results in very similar brackets each time. Our initial foray into it was an experiment. It turned out to work surprisingly well with the exception that in the very first rankings Spud was below DZ, but we already had the feeling that he was going to or already had surpassed him. After integrating 1 more major tournament, and 2-3 ranbats the rankings have settled to something very close to what I think the scene would settle on in a voted power ranking.

If anyone in the Aus scene has a working knowledge of Python, I'd be happy to walk them through setting up the database and importing results. Though my experience is all challonge and I believe you guys use Tio? Either way, would be a cool experiment to see how the numbers compare to your voted PRs.
 

Splice

Smash Hero
Joined
Mar 1, 2009
Messages
5,125
Location
AUS
#88
We do have some Elo ratings here.
The algorithm on this is something like
Scrubs beaten - Amount that you live in Victoria = ELO

secretly pleased that someone actually realised I've got placings with not just Falco, but also GnW and Puff from this year : )
 
Last edited:
Joined
May 25, 2010
Messages
595
Location
Melbourne, Aus
#89
The more data you collect, the more your top players are going to have inflated ratings since you'll have scrubs come and lose and donate a few ELO points to the pool and then never show up again. It's not a functional way to rank players when there's so little overlap between scenes. Smopup is not the second best player in the country. Soz Gerd. This is a problem in chess too, the ratings of top players keep going up and up.
TLDR ELO isn't well suited to smash.
 
Last edited:

OninO

Smash Journeyman
Joined
May 19, 2014
Messages
289
#90
I think the Microsoft trueskill rating sitting behind garpr doesn't work in the same way that Chess type ELO does. I'm not 100% sure but I think if a high skill player (with correspondingly low variance) plays and beats a newcomer with default skill and high variance, nothing much happens to the old players rating except that perhaps their variance shrinks a bit.

Here is a write up http://research.microsoft.com/en-us/projects/trueskill/details.aspx

Also you can play around with this calculator: http://boson.research.microsoft.com/trueskill/rankcalculator.aspx to experiment with what happens when you introduce a fresh player to a stable ranking. Run the default tournament 3-4 times you'll see skill estimates asymptotically approach some values. Introducing a new player and having them get bodied (set last place to 25 +- 8.33 i.e. fresh) doesn't change the asymptotic approach of the higher tier players.
 
Last edited:

Russ The Love Muss

I just want to love you
Joined
Jun 17, 2014
Messages
249
Location
Brisbane
NNID
RussellTLM
#91
The problem is that those systems are designed for match making players of the same skill level so it doesn't work well for a double elimination bracket where players are constantly playing against people way better or way worse than them
 
Joined
May 25, 2010
Messages
595
Location
Melbourne, Aus
#93
I think the Microsoft trueskill rating sitting behind garpr doesn't work in the same way that Chess type ELO does. I'm not 100% sure but I think if a high skill player (with correspondingly low variance) plays and beats a newcomer with default skill and high variance, nothing much happens to the old players rating except that perhaps their variance shrinks a bit.

Here is a write up http://research.microsoft.com/en-us/projects/trueskill/details.aspx

Also you can play around with this calculator: http://boson.research.microsoft.com/trueskill/rankcalculator.aspx to experiment with what happens when you introduce a fresh player to a stable ranking. Run the default tournament 3-4 times you'll see skill estimates asymptotically approach some values. Introducing a new player and having them get bodied (set last place to 25 +- 8.33 i.e. fresh) doesn't change the asymptotic approach of the higher tier players.
Didn't realise garpr used trueskill. I did read a pretty interesting write up on some of the maths behind trueskill, it's pretty cool. Forgot most of it though lol. Wonder how good trueskill is at dealing with the disparate nature of our scenes though. Would want to be able to have SD stay above Gerd say even if they only play once a year.
 
Last edited:

Atticus

Smash Journeyman
Joined
Nov 2, 2006
Messages
295
Location
Melbourne
#96
Should DQs affect ELO rating? I notice I lost a lot of points for sets I didn't actually play at the March CW (I left before I lost any PM sets and before Melee bracket even began).
 

Kaiza

Smash Journeyman
Joined
Dec 21, 2013
Messages
227
Location
South Australia, Australia
#99
~ new SA PR comin atcha ~
November 2015:

1. Nick (Marth) - ▬ 0
------------
2. Muscat (Peach) - ▲ 2
3. Kaiza (Falco) - ▲ 4
4. Flow Rats (Sheik) - ▲ 2
5. Blackshadow (C. Falcon) - ▼ 2
6. Oli (Fox) - ▼ 4
Grim (Fox/Falco/Ice Climbers/Marth)
Ghost (Jigglypuff)

------------
7. Scraddles (Fox) - ▲ 2
8. Ace (Marth) - ▲ 2
------------
9. Jeremy (Marth) - NEW
10. Nehruvia (Samus) - NEW

HONOURABLE MENTION: Edsta (Fox)

Panelists:
Kaiza, Nick, Flow Rats, Blackshadow

Quick wording about the 2nd tier of players:
Pretty much this whole tier is basically even in skill, rankings were decided on more recent placings + attendance to get a better gauge of skill since last PR

Based on tournament performances & attendance since last PR in state and interstate and Nick's opinion
 
Last edited:

Marcusee

Smash Rookie
Joined
Mar 14, 2016
Messages
1
NSW, Sydney Melee Power Rankings as of March 2016.

Ordering within tiers is alphabetical.



TEXT VERSION:
S+: RJ :peachmelee:, Zxv :sheikmelee::marthmelee:
S-: Cunchy :falconmelee:, MC :falcomelee:
A: Birdies :jigglypuffmelee:, Phez :pikachumelee:, RCJay :marthmelee:
B: Sora :foxmelee:, Zorsy :yoshimelee:
C: Cal :marthmelee:

HONORABLE MENTIONS: Ted, Cao, Gabe, Unreon
 
Last edited:

Kaiza

Smash Journeyman
Joined
Dec 21, 2013
Messages
227
Location
South Australia, Australia

Just in time for BAM 8 we have a slightly updated PR for SA

1. Nick (Marth/Falco) - ▬ 0
------------
2. Flow Rats (Sheik) - ▲ 2
3. Kaiza (Falco) - ▬ 0
4. Muscat (Peach/Sheik) - ▼ 2
5. Blackshadow (C. Falcon) - ▬ 0
6. Oli (Fox/Sheik/Falco) - ▬ 0
------------
7. Scraddles (Fox) - ▬ 0
8. Ace (Marth) - ▬ 0
9. Nehruvia (Samus) - ▲ 1
10. Edsta (Fox) - NEW

Previously Ranked:
Jeremy
Grim
Ghost

Panelists:
Kaiza, Nick, Flow Rats, Blackshadow, Muscat

Quick wording about the 2nd tier of players:
Pretty much this whole tier is basically even in skill, rankings were decided on more recent placings, set counts over other players in same tier, and attendance to get a better gauge of skill since last PR.

Players are ranked on tournament performance, slightly skewed towards more recent results since last PR, set counts over other players, & attendance in state and interstate.
 
Last edited:

Kaiza

Smash Journeyman
Joined
Dec 21, 2013
Messages
227
Location
South Australia, Australia


Pre-OHN SA SSBM PR comin at 'cha

1. Jungle | GorillaBoy (SA Nick) (Marth/Falco) - ▬ 0
------------
2. Kaiza (Falco) - ▲ 1
3. Muscat - ▲ 1
4. Flow Rats - ▼ 2
5. Blackshadow (C. Falcon) - ▬ 0
6. Oli (Fox/Sheik/Falco) - ▬ 0
------------
7. Scraddles (Fox) - ▬ 0
8. Nehruvia - ▲ 1
9. Ace (Marth) - ▼ 1
10. Edsta (Fox) - ▬ 0

Previously Ranked:
Jeremy
Grim
Ghost

Panelists:
Kaiza, Nick, Flow Rats, Blackshadow, Muscat

Players are ranked on tournament performance, & recent results since last PR (in state and interstate), set counts over other players, & attendance.
 

Kaiza

Smash Journeyman
Joined
Dec 21, 2013
Messages
227
Location
South Australia, Australia
End of year SSBM rankings have been unofficially out for a while but here's the official announcement:

1. Jungle | GorillaBoy (SA Nick) (Marth/Falco/Fox) - ▬ 0
------------
2. Kaiza (Falco) - ▬ 0
3. Flow Rats (Sheik/Luigi) - ▬ 0
4. Muscat (Peach/Sheik) - ▬ 0
------------
5. Nehruvia (Samus) - ▲ 1
6. Mylo (Jigglypuff/Fox) - NEW
7. Ace (Marth) - ▲ 2
8. Scraddles (Fox) - ▼ 1
9. Edsta (Fox) - ▲ 1

Previously Ranked:
Oli
Blackshadow

Panelists:
Kaiza, Nick, Flow Rats, Muscat
Players are ranked on tournament performance, & recent results since last PR (in state and interstate), set counts over other players, & attendance.
 
Last edited:

Splice

Smash Hero
Joined
Mar 1, 2009
Messages
5,125
Location
AUS
The official Australia Wide Power Rankings for the end of 2019 have been finalised. Heavily prioritizes majors results!

Also, don't believe those who say otherwise. Sundowns really is better than PROLIFIC!

1576764725861.png
 
Top Bottom