Smash 64 ELO Rating

[clubbadubba]

As far as small tournaments diluting ratings, people should understand that you can't really improve your rating much unless you are beating players around or above you rating. Once you are stable (have played more than 13 games), a good player won't get much out of small tournaments. For example, Sensei's rating jumped about 30 points from the ENTIRETY of Broken Hit Box, where he went 5-0.

You are right I did not include any tiebreakers, they should definitely be included. Who beat who? Also weren't there tiebreakers for Apex 2012 as well?

 

[Kefit]

I beat Nangoku for 7th place. Jaime beat Sensei for 5th place. Both matches took place after top four had been determined, but before top four took place on Saturday.

I don't think Apex 2012 had tie breakers, but I could be wrong.

 

[SSB64-Jel]

when you start up the online and mixed ratings make sure to include the online ladder

 

[clubbadubba]

Updated Apex 2013 to include the tiebreaker games for 5th/7th place.

Jel, I will definitely do that. In fact, you could do me a favor and make me a text file for the ladder. After I get the file, I'll do all the updating. So the earlier you get it to me, the less you have to do

I will also be including Nova's tournament, as well as the league which will start pretty soon, I'll probably start it after Nova's tournament ends.

 

[MrMarbles]

cool that means i will be apart of the online ratings! yey

 

[jimmyjoe]

#140 (hotdiggitydog)
#85 (Zack) are the same person

 

[clubbadubba]

Oh ok thanks, I'll put him down as hotdiggitydog. Damnit now there are 199 players instead of 200 lol

 

[Olikus]

Im sure it stands in the section, but Im allergic to tons of wall with information. Why is like sensei lower than paralax on the ranking? I might be wrong but I think parallax placed 1st and 2nd on two small tourneys. Sensei might have some lower placing than that, but he has won at least 2 semi big and finishing 6th at apex. Shouldnt that weigh more? If the system is just percent og highest placing in every tourney no matter how small or big, it isnt very accurate.

Pls enlighten me with few sentences =)

 

[clubbadubba]

Sigh, okay I'll try to be concise JUST FOR YOU.

-Parallax is undefeated (I think I only have him in 1 tournament in which he got 1st)
-He has beaten pretty much only good players (~2000 rating) while remaining undefeated
-Parallax has only ever gained points, and gained significant points each game because he has only played good players
-Parallax has also only played 4 games, which means he doesn't have much information to base his rating off of. If he plays more games, his rating will become more accurate.

-Sensei has lost points 5 times, with one of those losses coming to kefit who is ranked lower than him.
-Many of his wins have come against players that do not have a great rating. His best local competition, kero, is plagued by his apex 2012 performance when he was a nub.
-He has played lots of games, which means his rating is pretty stable.

Your ranking has NOTHING to do with your placing, only who you've beaten and who's beaten you. Thus it isn't tourney size that matters, but who you play in a given tourney. There is no way to say that a player who won tourney A is better than a player who won tourney B, just basing off of the size of the tournament. That is a HUGE misconception. See Peru.

 

[Cobrevolution]

for when pools happen at zenith, is there going to be a different format to submit? not sure if the chronology matters in that case. i assume that's going to stabilize the rankings a bit, with the bracket then furthering them

 

[Olikus]

Sigh, okay I'll try to be concise JUST FOR YOU.

-Parallax is undefeated (I think I only have him in 1 tournament in which he got 1st)
-He has beaten pretty much only good players (~2000 rating) while remaining undefeated
-Parallax has only ever gained points, and gained significant points each game because he has only played good players
-Parallax has also only played 4 games, which means he doesn't have much information to base his rating off of. If he plays more games, his rating will become more accurate.

-Sensei has lost points 5 times, with one of those losses coming to kefit who is ranked lower than him.
-Many of his wins have come against players that do not have a great rating. His best local competition, kero, is plagued by his apex 2012 performance when he was a nub.
-He has played lots of games, which means his rating is pretty stable.

Your ranking has NOTHING to do with your placing, only who you've beaten and who's beaten you. Thus it isn't tourney size that matters, but who you play in a given tourney. There is no way to say that a player who won tourney A is better than a player who won tourney B, just basing off of the size of the tournament. That is a HUGE misconception. See Peru.

Thanks I see. Makes more sence now.

 

[Yobolight]

You should include Sweet XII

 

[Cobrevolution]

follow the format as listed on the previous page and write the textfile for it, shouldn't take more than ten minutes

 

[clubbadubba]

You should include Sweet XII

yea someone could make it and give it to me. I might get around to it eventually, but I would prefer if TO's or some other people help out so I don't have to do them all.

As far as order goes, it is best to keep things in chronological order. For example, lets say there are 4 players in a pool, A B C and D, all starting at 1500. Assuming a KFactor of 32 (standard, not what this uses though), A beats D, B beats C, so A and B have a rating of 1516. Then A beats B while C beats D, and A has a rating of 1532, B has a rating of 1500.

Now say instead, A played B first, so A = 1516, B = 1484, D = 1484, C = 1516. Then A beats D and B beats C, so A = 1532, B = 1501.47. This is a small difference, but order definitely matters even in pools. Differences could be magnified depending on circumstances. Order should be preserved when possible.

 

[Cobrevolution]

hmm, i feel like keeping track of that would be a bit more difficult though. i'll try to keep a piece of paper and record who played who in what order at zenith but it might prove problematic, especially if i'm tryna get my friendlies in.

 

[clubbadubba]

well i think I'll try to talk to the TO, and just ask that whenever a match is put into the computer, that it also be written down on a piece of paper or a word doc

 

[Battlecow]

The simple answer to the parallax question is that we don't have enough data to make these ratings relevant or useful yet. With more tourneys and more data you should see people's rankings beginning to make more and more sense. Of course, we're gonna need a number of tourneys before we get there--the 64 community in general has little enough data that it'll be a while.

 

[Olikus]

Maybe we can say you have to attend at least 3 tourneyS to get ranking? I think at least 2.

 

[Sangoku]

The parallax paradox!

 

[clubbadubba]

not setting a minimum for # games, because I think its actually been very accurate so far, plus very few people have played enough games or been to enough tournaments. 1 thing I just did is up the minimum KFactor from 64 to 80, which means that your 10th game will make you reach a "stable" rating. This "stable" rating will be less stable than it was before, but it will be reached faster. This will make players with more than 10 games less stagnant, and thus give less of an edge to players who may have done well early on in their records. Also, I will put an asterik next to players who have not yet played 10 games, so you know who still has a very volatile rating.


Oh and hey bcow/kero, should I use your money match from Apex last year? You guys did both go try hard right?

 

[Cobrevolution]

kero/yobo/whoever reads this first: link me to the thread of results from that tournament in NY and i'll do the textfile if everyone is lazy

same for sweet i guess

 

[Sensei]

I suppose the match set count is not calculated in this, since I don't think this has been kept on record for most tournaments and it would make the ratings fairly more complex but at the same time more accurate. Winning against someone 3-0 should carry more weight than winning 3-2. I think the newer versions of TO have this feature to record set counts, but not sure. This is just something to consider.

 

[clubbadubba]

I'd agree if data was consistently kept AND people took games 100% serious when they are up against someone they know they are better than. 1 high ranked person blowing a game against a low ranked person because they were messing around has way too much potential to make this less accurate.

cobr, i'll do sweet tonight. thanks for being a beast and making text files though.

 

[Battlecow]

There are a lot of factors that go into set counts. Say I have a strategy that leads to me winning every tournament but dropping a lot of matches--I'm still the best player in the world because winning is all that matters IN TOURNAMENT. Elo's shouldn't take set counts into account because players don't necessarily aim to win every match, they aim to win the set.

they went through this in the SSB(M)PD discussion, came to the correct consensus

Yeah count me v kero from Apex

I'd argue against counting on this for anything serious until we have a lot more data. Most people have at most 2-3 tourneys, that's 2-3 points of data.

I emailed you results for the bracket of FC legacy, clubba. I did the bracket from memory (8 guys so not that hard, 99% sure it's completely correct) but the file I sent clubba if anything below contradicts what people remember) and I don't have access to pools results so sadly we don't have that data.

battlecow
moogle
han solo
knitephox
jace
rumbly
zduck
vudujin
End Of Entrants List
battlecowvudujin
mooglezduck
knitephoxrumbly
han solojace
battlecowknitephox
mooglehan solo
battlecowmoogle
rumblyvudujin
jacezduck
knitephoxjace
han solorumbly
han soloknitephox
mooglehan solo
battlecowmoogle

 

[Cobrevolution]

hit tab in between the names, cowboy

unless smashboards is like [[midfing]]

 

[clubbadubba]

sick tabs lol. thanks man. cobr just sent me some new stuff too, ggsmash3 from last december. I just added sweet XII as well.

bcow, try .txt instead of .rtf next time.

Almost forgot to add the money match! Everything is in now.

 

[Cobrevolution]

mm clubba did you add the one with you nintendude and sk? or is six people NOT CUTTING IT

 

[clubbadubba]

oh yeah well i asked if nintendude could do it. not sure where the data is. I'll look out for it.

Also, at Sweet XII, is jerry == valoem?

 

[Cobrevolution]

yeah, jerry's been entering as jerry instead of valoem for some unknown reason

 

[clubbadubba]

haha, well he's going to permanently be valoem. Maybe I'll come up with a list of alternate names people use so i don't have to keep changing things. ****ing cmu6eh/karajan, diablo/skyfire, jerry/valoem, comboblaze/combo blaze (seriously wtf)

 

[Battlecow]

it was tabbed in the rtf

 

[clubbadubba]

and that's why you use the txt bro

 

[thegreginator]

Who are the active mods for these boards? Clubba you should message them about getting this topic and the player map stickied.

 

[Nintendude]

not setting a minimum for # games, because I think its actually been very accurate so far, plus very few people have played enough games or been to enough tournaments. 1 thing I just did is up the minimum KFactor from 64 to 80, which means that your 10th game will make you reach a "stable" rating. This "stable" rating will be less stable than it was before, but it will be reached faster. This will make players with more than 10 games less stagnant, and thus give less of an edge to players who may have done well early on in their records. Also, I will put an asterik next to players who have not yet played 10 games, so you know who still has a very volatile rating.


Oh and hey bcow/kero, should I use your money match from Apex last year? You guys did both go try hard right?

There's formulas for making some approximations to account for round robins. There shouldn't be a need to keep track of game order.

 

[clubbadubba]

I'll look into it.

Having difficulty, if you could point me in the right direction?

 

[clubbadubba]

Double post cuz its my thread and I DO WHAT I WANT! Also no mods anyway lol.

Playing *** Vwls today and he told me he was SA-in-NE at Apex 2013 (which I think I knew at one point), so I changed his name in the ratings. While I was at it, I made it so that ties are noted by the ranking number.

 

[clubbadubba]

TRIPLE POST CUZ I DON'T GIVE A MOTHER ****!

Its finally here. Online ratings as well as a joint console/online rating has been added!!!! So far only the ladder has been added

 

[Kimimaru]

Hey clubba, I actually have 3 wins in the ladder rather than 2.

 

[clubbadubba]

Hey clubba, I actually have 3 wins in the ladder rather than 2.

Are you sure that all 3 wins were real wins, or was one of them potential a forfeit win? I am not counting forfeits for the ratings. Jel sent me a list of ladder games, but it seemed to be only recent ladder games and seemed to be missing some, so I went back through the thread myself and made my own file. Both me and Jel only found 2 games you won.

So far, I only have you beating battlecow and loto. If there is another win you have, just tell me who you beat and link me to the post where it was reported so I can put it in.

 

[Kimimaru]

Yeah it was a forfeit win. Never mind in that case.