• Welcome to Smashboards, the world's largest Super Smash Brothers community! Over 250,000 Smash Bros. fans from around the world have come to discuss these great games in over 19 million posts!

    You are currently viewing our boards as a visitor. Click here to sign up right now and start on your path in the Smash community!

Elo Ratings in Competitive Melee

Signia

Smash Lord
Joined
Feb 5, 2009
Messages
1,157
Yeah Ankoku, your variables are so objective and boundaries for the size of tournaments that count are impossible to make. There's no way it would work!

We all know players vary in skill level depending on the day or other factors. So I guess tournament results aren't accurate representations of skill level either!

And it's so much better if people with biases seed a tournament in 5 minutes (read: at least 20 min) than if a program does it instantly! God you're dumb.

no system is perfect so the elo rating system can't work LOL I'm sorry I can't do this anymore. God! are you that ********?
 

MTKO

Smash Journeyman
Joined
Feb 18, 2008
Messages
294
Location
Hampden, Maine
Yeah Ankoku, your variables are so objective and boundaries for the size of tournaments that count are impossible to make. There's no way it would work!

We all know players vary in skill level depending on the day or other factors. So I guess tournament results aren't accurate representations of skill level either!

And it's so much better if people with biases seed a tournament in 5 minutes (read: at least 20 min) than if a program does it instantly! God you're dumb.

no system is perfect so the elo rating system can't work LOL I'm sorry I can't do this anymore. God! are you that ********?
He's either really against this or he's trolling. I feel like a lot of the people that are determined to constantly bash this idea are doing it solely to disagree.

Sorry if this already got discussed, but I don't quite understand how people think the ratings will be very off and some people who aren't good will have super high ratings or something just from winning a few matches against better players in a tournament setting? If there are specific tournaments (large ones like a national) that are designated to try this system out and a player there beats a few high ranked people in a tournament match, then his rating is going to be good and I think that would reflect well on his level of skill, but it seems some people think that in that situation the person has not played enough matches to really determine their true rating? Is that what I'm getting from this? I think as long as you play a set number of rated matches before you get a rating, then that would eliminate that problem.

Edit: Sorry if that made no sense, I didn't explain it well.
 

Signia

Smash Lord
Joined
Feb 5, 2009
Messages
1,157
I think as long as you play a set number of rated matches before you get a rating, then that would eliminate that problem.
their plan is get data for a year then reveal the ratings.
 

MTKO

Smash Journeyman
Joined
Feb 18, 2008
Messages
294
Location
Hampden, Maine
their plan is get data for a year then reveal the ratings.
Ah I see I see. That would be good. It appeared to me like some people had this mentality like if you play x amount of matches (like a really small amount of matches, say 5 for example) and beat a bunch of higher rated players that are really good, then you don't deserve that rank because... well I don't really know what the reasoning for that would be. It seems maybe some of the people that are so against trying the system are confusing it too much with a ranking system, when it's supposed to just be a rating representing your average level of play.
 

GOD!

Smash Ace
Joined
Dec 30, 2008
Messages
778
Location
Rome, GA
their plan is get data for a year then reveal the ratings.
whats the point of doing a really really ****ty restatement of my argument and then trying to act like i think it

then mocking it. good waste of time noobbbb
 

AlphaZealot

Former Smashboards Owner
Administrator
Premium
BRoomer
Joined
Jul 6, 2003
Messages
12,731
Location
Bellevue, Washington
Good stuff Atlus if you just keep doing that you can get a decent database compiled.

The biggest problem are so many TO's are still lazy and don't post brackets, and even if they do, they do not post bracket files to tio (allisbrawl).

It looks like GOD! is just trolling at this point. You can scream to the high heavens that this won't work, but if someone steps up and compiles the tio rankings from 20-40 tournaments it will pretty quickly filter players based on how they have performed (and also based on what you claim are the already known levels of their skills).
 

GOD!

Smash Ace
Joined
Dec 30, 2008
Messages
778
Location
Rome, GA
im not trolling lol

it comes to a point where ive said everything i can, and u cant say anything against my serious objections

like if srsly smashboards are ******** lol

definitely not troll at all. elo ratings are a ****ig impractical and plain bad idea w/e who cares chase your rating system noobs
 

Divinokage

Smash Legend
Joined
Aug 6, 2006
Messages
16,250
Location
Montreal, Quebec
im not trolling lol

it comes to a point where ive said everything i can, and u cant say anything against my serious objections

like if srsly smashboards are fuking ******** lol

definitely not troll at all. elo ratings are a ****ig impractical and plain bad idea w/e who cares chase your rating system noobs
It looks like you are talking to yourself when you say this.. the others are a reflection of yourself after all.

Ya I really see no problem with Elo, simply to help seed tournaments. It would save a lot of time for the TOs. Last tournament this saturday, Tect asked me to help me seed the tournament and it took some time to do it as sometimes we are unsure what people's skill level should be. I mean we obviously don't know every single person out there. For that idea alone, it should work extremely well. Now if we want to go further with Elo.. then idk.. it's more like speculation rather than experience after the idea of seeding.
 

GOD!

Smash Ace
Joined
Dec 30, 2008
Messages
778
Location
Rome, GA
wow it would be really great if u could use it for seeding. IT WOULD BE GREAT IF IT ACTUALLY WORKED

and so would money growing on trees and world peace but big problem, they dont work lol
 

Strong Badam

Super Elite
Administrator
Premium
BRoomer
Joined
Feb 27, 2008
Messages
26,545
you are so bad at posting oh my gosh, all of the points you made were either weak or awfully presented, and at this point you're just posting angrily without even making any points.
 

GOD!

Smash Ace
Joined
Dec 30, 2008
Messages
778
Location
Rome, GA
you are so bad at posting oh my gosh, all of the points you made were either weak or awfully presented, and at this point you're just posting angrily without even making any points.
i mean again

i dont care about the opninions of dumb people like u who contribute nothing so who cares loll

im not even mad haha u are just really dumb

all my points were fine. so u are just wrong and dumb. once again i dont expect a forum lurker like u to understand anything really. its ok to be dumb but please dont try and act like the actual reasonable people are anything like u. basically ur dumb

and i cant post on smashboards cause its like a huge herd of sheep trying to get notoriety, and most of them are ignorant as *** so who cares. have fun with ur system and stop replying to me
 

joeplicate

Smash Master
Joined
Nov 30, 2008
Messages
4,842
Location
alameda, ca
i'm really down to see what the elo rankings come up with


Especially if they're done well, I think it'd be really interesting. Plus, if they seem inaccurate, you can still just say "**** that" and move on....

Since the whole system is still really early, and everything's still being tested/worked out, we would basically be comparing the results of the Elo system to common sense, and not the other way around. (We wouldn't look at Atlus's post and think "Huh, Lovage ranked higher than Dr PeePee? Well, if Elo says it, it must be true!")
 

Zivilyn Bane

Smash Master
Joined
Nov 18, 2004
Messages
3,119
Location
Springfield, MO
This guy is totally outdoing me. Bet yall miss me now =D
Completely lol. At least you had some cute comics for us to laugh about. This God kid is just annoying.

GOD!, you do realize that Strongbad is one of the developers for Project Melee, right? I mean seriously if that's not contributing to the community then what is?

If Elo works it could be used as a great organizational tool for our community. I really could do no harm, so let's just try it and see how it goes. Shut up if you don't like it, we've read your redundant posts for long enough.
 

PEEF!

Smash Hero
Joined
Jun 25, 2008
Messages
5,201
GOD!, you do realize that Strongbad is one of the developers for Project Melee, right? I mean seriously if that's not contributing to the community then what is?
In a way that might actually be taking away directly from the community haha
 

Juggleguy

Smash Grimer
Premium
Joined
Aug 16, 2005
Messages
9,354
Location
Ann Arbor, MI
Also, for the Name, Handle, and Elo fields, you should make it auto-highlight any values there if you tab and shift-tab between them. It's cumbersome to add a bunch of players through that because each time you tab to the next field you have to manually delete what's already there. You should also make it so we can press Enter and it'll act like pressing the "Add Player" button. Right now I have to click twice (once to exit the entry field, and once on the button) for it to add the new player.
Agreed x100, in case people missed this post
 

Zivilyn Bane

Smash Master
Joined
Nov 18, 2004
Messages
3,119
Location
Springfield, MO
Ok I know somebody before posted some ratings starting with a base of 1,400. I personally however believe this is a little high. At the same time, there is plenty of merit for those being higher because of the caliber of players being rated ie Mango, Lovage, S2J, MacD, DrPP, Fly, Lucky, ect. These are mostly players that would actually benefit the system by starting at a higher base. What I did was rate the last four local tournaments. During them we have had three out of state players and six out of community (spanning 4-5 cities) (including the oos) players. I used a base of 1,000. It's also important to note that all Winners Bracket matches should be calculated before Losers Brackets, because order of events can actually effect the outcome. Chronological order is important in other words. I will be posting a new list after each event that includes new players and updates on existing. I will be removing players after four consecutive months of inactivity. As of right now, our list is four months old thus nobody has been removed yet. The following is a table that details the ratings of each player that has participated in the Melee Singles brackets at Pathways since and including January. No low-tier, doubles, or brawl events are considered. Also, each match is calculated as 1-0 for the winner and 0-1 for the loser whether it be 2-1, 3-1, 3-2, ect. Note: The final (and current) rating would be listed under Elo-Apr. Chg is the change from the month before. In this column, new means it's the persons first tournament, na means they didn't participate in the event that month so there was no change. Record is recorded as: wins-losses-byes. Byes have no effect whatsoever on elo ratings. All ratings were calculated using the software provided in this topic.

Rank Gamertag Elo-Jan Chg Record Elo-Feb Chg Record Elo-Mar Chg Record Elo-Apr Chg Record
1 Zivilyn Bane 1120 new 5-0 1189 +69 9-0-1 1161 -28 12-2-1 1215 +54 16-3-2
2 Saberguy 1132 new 5-0 1132 na 5-0
3 Zantetsu 1049 new 4-2 1049 na 4-2 1031 -18 6-4 1079 +48 11-6
4 Sinister 1059 new 4-2 1059 na 4-2
5 VTS 1055 new 4-2 1055 na 4-2
6 BoP 1032 new 3-2-1 1041 +9 6-4-1 1041 na 6-4-1 1041 na 6-4-1
7 LinkMO 1038 new 3-2 1038 na 3-2
8 Kellen 1030 new 3-2 1030 na 3-2
9 DrewBear 980 new 1-2-1 1045 +65 5-4-1 1011 -34 6-6-1 1011 na 6-6-1
10 xzerolegend 1017 new 3-2 1037 +20 6-4 1037 na 6-4 1004 -29 7-6-1
11 Kodie 996 new 2-2 996 na 2-2 996 na 2-2 996 na 2-2
12 Linkz 956 new 1-2-1 942 -14 2-4-3 942 na 2-4-3 994 +54 5-6-4
13 Terrance 978 new 1-2-1 959 -19 2-4-1 987 +28 5-6-2
14 Terral 975 new 1-2 1011 +36 4-4-1 1007 -4 6-6-1 979 -28 7-8-3
15 Yoshiman 978 new 1-2 978 na 1-2
16 Male 975 new 0-2-1 975 na 0-2-1 975 na 0-2-1 975 na 0-2-1
17 BZA 1024 new 3-2-1 1024 na 3-2-1 1024 na 3-2-1 966 -58 3-4-2
18 an honest man 963 new 0-2 963 na 0-2
19 JeBus 953 new 0-2-1 953 na 0-2-1 953 na 0-2-1
20 Sam 953 new 0-2 953 na 0-2
21 ICE 974 new 1-2 951 -24 2-4-1 951 na 2-4-1 951 na 2-4-1
22 Kyle 950 new 0-2 950 na 0-2 950 na 0-2 950 na 0-2
23 Tempest 951 new 0-2-1 932 -19 1-4-1 932 na 1-4-1
24 YBMP 951 new 0-2 912 -39 0-4-2
25 Ugvts 952 new 0-2-1 908 -44 0-4-2 871 -37 0-6-2 847 -24 0-8-4

A few interesting things to note. For one we can see that player record does not directly correspond to rating, however they should be somewhat consistent. One of the major things one might notice about the March event is that Sinister came out rated higher than VTS. Both players were 4-2, yet VTS beat Sinister twice in the brackets! How does this happen? Well simple. While VTS was 2-0 vs Sinister, they were both unranked at the beginning of the event and started at the base of 1,000. VTS didn't beat anybody with a decent established rating, however Sinister beat me. When he beat me I was over 1,200 and lost over 40 points from the defeat (which he gained). So even though VTS beat Sinister twice, he's still rated lower. This is OK and even though elo can be a little tricky at first, it will absolutely balance itself out. In this case, maybe not since I doubt we'll see VTS back at one of these for a little while. But if we did another tournament with the same people I think we'd see them balance. Also note that in all reality, Sinister and VTS are both probably very close in skill and is entirely reflected in their respective ratings. Also note Zantetsu's change from April. He took a set off me in Grand Finals set one, however I still gained more points than him. Seems a little odd at first, however if you look closer the answer is pretty obvious. Since I was 2-1 vs him, the 1-1 almost offset (there might have been a mutual gain or loss of a very small amount), so it really only sees it as 1-0 for me, which makes sense if you think about it. And so as 1-0 for me, it's almost like a perfect record and Zantetsu having only one loss. It's also important to note that ratings change after each match, however if you're in round 5 of a tournament, your rating is still seen as what it was before the event until the event is over. Also notice only three players have participated in all four events, Terral, Ugvts, and myself.

Anyways, does anybody else have any opinions or thoughts on this? Please post up with comments and thoughts if you have any. Or any questions or things you notice.

I would personally be willing to start the database of players with elo ratings. However, I can not be the one to process each and every event until possibly there is a more automated system.

Brackets Used:
Pathways January
Pathways February
Pathways March
Pathways April
 

Windrose

Smash Lord
Joined
Mar 22, 2009
Messages
1,470
If ELO is used for seeding. Low seeded players would always get seeded against high level players wouldn't they? Wouldn't that mean most likely that the low ELO player would always be losing and thus dropping more and more (even if it's slowly) in ELO points?

Noobs like me bank on the chance that I'll get seeded into an easier pool/bracket so I can play more tournament matches
 

Orko

Smash Apprentice
Joined
Jan 28, 2011
Messages
121
Location
Sacramento
It's seeded, but it's Double Elim. You'll get dropped into loser's, where there will be other lower seeded Elo rated player.
 

Winston

Smash Master
Joined
Aug 13, 2006
Messages
3,562
Location
Seattle, WA (slightly north of U-District)
Noobs like me bank on the chance that I'll get seeded into an easier pool/bracket so I can play more tournament matches
But this isn't supposed to happen, ELO or not. Lower seeded players are always paired vs higher seeded players in brackets regardless of how the seeding is done.

And if it's pools, you'll get to play the same number of tournament matches anyway.
 

Cactuar

El Fuego
BRoomer
Joined
Mar 10, 2006
Messages
4,820
Location
Philadephia, PA
At a largescale, multiday tournament, it might be beneficial to play out a 32 man bracket. I've shown Nintendude a tournament format that plays out 1-32 and doesn't take an obscene amount of time. This raises the minimum number of matches that can be played in bracket from 2 to 5, and keeps the current maximum possible sets at 11.

For small scale (32ish entrants) tournaments, a possible idea is to have pools run, then complete a 16 man bracket with full placement, and have a second set of elo matches for the 17-32 range of players eliminated in pools.

Edit: Or a 17-32 bracket. Not that I wanted to bring up the forbidden amateur bracket topic.

The objective is to increase the minimum number of matches that can be played to normalize the amount of possible elo swing (inputs) that can be expected at a tournament.
 

Divinokage

Smash Legend
Joined
Aug 6, 2006
Messages
16,250
Location
Montreal, Quebec
At a largescale, multiday tournament, it might be beneficial to play out a 32 man bracket. I've shown Nintendude a tournament format that plays out 1-32 and doesn't take an obscene amount of time. This raises the minimum number of matches that can be played in bracket from 2 to 5, and keeps the current maximum possible sets at 11.

For small scale (32ish entrants) tournaments, a possible idea is to have pools run, then complete a 16 man bracket with full placement, and have a second set of elo matches for the 17-32 range of players eliminated in pools.

Edit: Or a 17-32 bracket. Not that I wanted to bring up the forbidden amateur bracket topic.

The objective is to increase the minimum number of matches that can be played to normalize the amount of possible elo swing (inputs) that can be expected at a tournament.
I actually have no idea what you just said. :troll:
 

Zivilyn Bane

Smash Master
Joined
Nov 18, 2004
Messages
3,119
Location
Springfield, MO
Good thoughts Cactuar. I wondered how Melee could implement this with it's love for the double elimination bracket. As I'm sure you are aware, chess uses a swiss system where there are no eliminations. There are a set number of rounds which are determined primarily ahead of time by the expected number of entrants. So if there's a tournament expecting 30 people, they'll probably do 5 rounds. Seeds 1-15 play 2-30 in round one. After that, players with approximately the same number of wins and losses play in each round and no two players can play each other twice in the same tourney. It actually works very well and I love the swiss system personally. However the few times it's been tried in Melee it's been a disaster (mostly caused by gimmicky adaptations implemented by TOs inexperienced with how a swiss format should be ran). I've often considered running a regional with a swiss system to show the community how awesome it really can be. I may someday still try it.

Kage, what Cactuar is talking about is basically adapting a system that would allow players likely to go 0-2 in a tournament to get more rated matches in.

I doubt there will be any movement amongst the community to switch from Double Elim, so there's always the option of just rating pools matches also. I'm all for that anyways and really see no downside. But then again like with our current system small-scale tournies rarely if ever do pools.

Your idea is feasible but again it's just a huge matter of getting TO's and players to care enough to make these changes to begin with. Somebody like me, while being involved in melee for a long time, couldn't possibly convince TO's to do something like this because I'm still pretty much a nobody. However Cactuar on the other hand....you could probably get TO's to listen to you :)

EDIT: Also Cactuar can you also show me what tournament format you are referring to that plays out 1-32?
 

DippnDots

Feral Youth
Joined
Sep 27, 2006
Messages
2,149
Location
Cbus, Ohio
Just a heads up with rating pool matches, there are a lot of people who use pools as warm ups. Not to say this is a reason to not rate pool matches, just that the results won't necessarily reflect the truth, that is until like, the pros take pools seriously again.
 

Nintendude

Smash Hero
Joined
Feb 23, 2006
Messages
5,024
Location
San Francisco
Just a heads up with rating pool matches, there are a lot of people who use pools as warm ups. Not to say this is a reason to not rate pool matches, just that the results won't necessarily reflect the truth, that is until like, the pros take pools seriously again.
Rated pool matches should encourage top players to take pools more seriously and warm up for them / not sandbag them.
 

DippnDots

Feral Youth
Joined
Sep 27, 2006
Messages
2,149
Location
Cbus, Ohio
right, but that doesn't mean they will.

edit: Don't get me wrong, I would like pool matches to be rated, that way you'd see less bracket manipulation with pool seeds, and people who don't even make it to bracket won't be rated at all if we don't?
 
Top Bottom