Official: SSBPD unsupported; source code released.

[LLDL]

Sure. chances are it won't be duplicate work, because I'm looking up tourneys that I have been to that didn't have results threads, back when tournaments were in regional zones. Also, apparently they didn't use tio at all this no johns LOL, they did it by hand on paper. Idk the full story though.

 

[Bones0]

Can anyone tell me if my Graphix tourneys are being processed? What exactly does processing entail because if it's just fact checking and handling duplicate names I'm entirely willing to help out. I've probably flagged like 50 different people for wrong/duplicate names so if I could just do the dirty work myself it'd probably help free up more capable individuals to do more specific work.

 

[FoxLisk]

Can anyone tell me if my Graphix tourneys are being processed? What exactly does processing entail because if it's just fact checking and handling duplicate names I'm entirely willing to help out. I've probably flagged like 50 different people for wrong/duplicate names so if I could just do the dirty work myself it'd probably help free up more capable individuals to do more specific work.

hey, sorry, we've got kind of a backlog right now. we're trying to get through a bunch of results tonight so we can rerun the ratings and get things a little more accurate.


In that vein, though, if you or anyone else is interested in being a mod and helping out with this stuff, shoot me a PM. we're looking to find another person or two to help out so let us know!

 

[Nintendude]

Yeah, if you uploaded results but don't see them on the database yet then don't worry. We just have a long list of stuff waiting to be processed and we haven't had time to get through it all.

 

[Max?]

Until the No Johns and other Tristate tournies are put in here, this list will never truly be accurate

 

[LLDL]

I'm workin on it otg, I've been salvaging a lot of old NJ tournaments and Nintendude working on NY. Alukard kind skimped out on posting a lot of brackets tho =(

 

[Max?]

I'm aware, Alu's a bad TO.

I think the best thing would be to start taking tourney results from a certain point and time and going from there as opposed to using data from YEARS ago.

 

[Strong Badam]

after recent tourneys start getting uploaded and the project goes for about a year or so, tristate will be fine. Chu said he'd upload his bi-weeklies and we can probably talk to Alukard about his no-johns series.

 

[Nintendude]

Until the No Johns and other Tristate tournies are put in here, this list will never truly be accurate

I managed to recover brackets from 17 No Johns tournaments, starting from the very first one back in 2008 or so. That's still only about 2/3 of them, but it'll help a lot. There's also 10 Mass Madnesses that have yet to be put in the system, in addition to the ESTICLE series (we have about half of them), Cataclysm series, SPOC series, etc. Tristate actually has a fair amount of data waiting to be processed.

Things will never be perfect, obviously, but I think we can get pretty damn close. If we use ratings to seed, TOs can make exceptions for players like you who are screwed by not having much data in the system. Things will get a lot more accurate over time too, but people really have to get on Alukard's case about documenting results.

It's also important that we do use data from years ago. Otherwise people get screwed by, for example, losing to PC when he's rated 1300 when he should actually be 1700+ based on his MLG performances.

 

[Max?]

It's also important that we do use data from years ago. Otherwise people get screwed by, for example, losing to PC when he's rated 1300 when he should actually be 1700+ based on his MLG performances.

Makes sense.

 

[Bones0]

You can convince tristate players of anything as long as you somehow link it with the threat of PC Chris's skill being marginalized.

 

[kalamazhu]

This data base is cool and all but is there any way to make it so that major tournaments count MORE than locals and regionals. The problem here is that a lot of players who don't travel outside of their region just farm points everyweekend against the same locals and are never at a risk of losing any points by playing better people

 

[Zivilyn Bane]

This data base is cool and all but is there any way to make it so that major tournaments count MORE than locals and regionals. The problem here is that a lot of players who don't travel outside of their region just farm points everyweekend against the same locals and are never at a risk of losing any points by playing better people

Can you show an example of some players that are doing this and have much higher elo's than they should?

Nationals and regionals have more players, thus larger pools, more rounds of pools, and bigger brackets. The sheer size of opportunity makes large tournaments weight a LOT more than the 4-5 games you get in at a local.

 

[kalamazhu]

Can you show an example of some players that are doing this and have much higher elo's than they should?

Nationals and regionals have more players, thus larger pools, more rounds of pools, and bigger brackets. The sheer size of opportunity makes large tournaments weight a LOT more than the 4-5 games you get in at a local.

Kels and dart
Edit: also if you go to a local every weekend and win them all thats like 10 sets you win that tourney with no losses every weekend and you can just rank up by going to locals. i think the ssbpd should idealy only count like top 5 or 7 tournaments so you can only rank so high from going to small tournaments, which i believe is more fair because you have to prove urself by going to nationals and stuff. I realize this may be hard to program and stuff but that would be the best idealy

 

[Prince_Abu]

yeah only count the top 5-7 tourneys so that people who do well at big tourneys inherently get better ELO so this way the rankings will be really accurate
thats what USTA tennis does and it works well

 

[kalamazhu]

Also maybe like a point decay system where matches don't count after like 2 yrs would be nice. Ken is ranked so high, but hasn't played a match since 2007 and hes not losing points because hes not playing more tournaments

 

[Kal]

Elo only factors in who you play against. It doesn't make sense to factor in the size of the event because, as far as the Elo is concerned, there is no difference between beating M2K at a local and beating him at APEX.

The "time" problem has been addressed. Here. Hopefully, Ken does come back, beats everyone, then doesn't show up for another couple years. So that, two years from now, I can reiterate this point.

 

[Strong Badam]

There's only so many points you can gain by beating low rated players. It eventually ends up only netting you 1-4 points per win, which makes "farming" locals pretty silly.
Kels and Dart's ratings seem accurate to me. Need moar data tho

 

[kalamazhu]

okay ken would get wrecked now
It would be nice if Elo factors in the size of the event combined with only taking into account top 7 tournaments.
I think those would give more accurate rankings

 

[Strong Badam]

Okay buddy you already said that and we already answered that. Have a nice day!

 

[kalamazhu]

There's only so many points you can gain by beating low rated players. It eventually ends up only netting you 1-4 points per win, which makes "farming" locals pretty silly.
Kels and Dart's ratings seem accurate to me. Need moar data tho

But they will remain high by "not losing" at locals and are not at risk of losing points. players like Kirby Kaze should be higher by placing well at several bigger tournaments rather than winning lots of locals

 

[DoH]

http://ssbpd.com/tournament/Test Your Skills

This tournament, the oscar is actually mojo, http://ssbpd.com/player/Mojo

 

[Strong Badam]

The problem, kalamazhu, isn't that players are getting points from locals, but rather we don't have Canadian local data from which players like KirbyKaze and unknown522 would get points. Local data is very important for making national tournaments where regions mesh and play each other matter, otherwise it'd be a bunch of 1200s against each other and it'd be like a local with more potential for playing more 1200s.

DoH: Done! Thanks!

 

[FoxLisk]

Kalamazhu: your points have already been well addresses by strong bad, but in case you'd like it ex cathedra, decay and tournament size factors aren't happening.

And for an extra angle: who cares if Kels' rating is too high? Hes not hurting anyone, and if you think his rating is too high, just be less scared of him when you meet him in bracket.

 

[Prince_Abu]

idk i thin unless u take some other factors into account the rankings r gonna b more inaccurate than they wud otherwise b

 

[Fortress | Sveet]

I think the rankings are pretty accurate

 

[Mahie]

Kels isn't too high, the others aren't high enough, because data is missing for them.

Also Kels beats Darkrain/Kage every time they show up, I don't see how he's a bad player.

If you needed an example, you should have went with France being way too high, because we have a lot of tourneys processed.

 

[FoxLisk]

idk i thin unless u take some other factors into account the rankings r gonna b more inaccurate than they wud otherwise b

Please do some research into statistical ratings systems. If you still have concerns after you gain a little background on the issue, let us know.

 

[Zivilyn Bane]

I think everybody that thinks Kels is too high should MM him

 

[Bones0]

Not sure how anyone can complain about any issue with the ratings at all until a solid foundation of data is available. I've see a couple players with higher ratings than me that I've beaten head to head several times. I don't immediately assume this is because the algorithm is wrong. It seems a lot more likely that they just have more sets uploaded for them. If it's a year from now and we have people in the top 10 from farming points at locals, then something can be done about it. Until then, it all just points to a lack of data, nothing more.

 

[Kal]

It's also not the case that beating someone head-to-head, even consistently, guarantees you a higher Elo. People should keep in mind that Elo rankings are relative; if we have a small, closed off group of players who never compete outside of their local events, then it's entirely possible for one player in this group to have the same Elo as Armada, despite no one in the group being any good. It's necessary for players from different regions to interact so that "relative" rankings become "global" rankings.

 

[Bones0]

Well yeah, but it's not just a head to head issue. I also place much better and beat other players they would get destroyed by.

 

[Fortress | Sveet]

It's also not the case that beating someone head-to-head, even consistently, guarantees you a higher Elo. People should keep in mind that Elo rankings are relative; if we have a small, closed off group of players who never compete outside of their local events, then it's entirely possible for one player in this group to have the same Elo as Armada, despite no one in the group being any good. It's necessary for players from different regions to interact so that "relative" rankings become "global" rankings.

its funny you mention that, seeing as we've been tied in Elo for a while lol

 

[Kal]

Maybe I'm just... as good as you? >_>

<_<

Just kidding, of course. There's no way I'm as good as anyone.

Edit: I've got one point on you. I'm clearly better.

Also, FoxLisk, the search function appears to be a little buggy. Not sure if this has been pointed out before, but even if I type in someone's name with the exact matching characters, the website returns "Sorry, no matches found. Did you mean one of these players?"

 

[Grizzlpaw]

I will become... the greatest yoshi!!! mwuhaha! @Ambition. Love how natral he feels in the air

 

[Dekar289]

in the lifetime of competitive melee in aus, there's been 4 guest melee smashers: captain jack, ken, m2k and mango.
those 4 tournies aside our rankings gunna skyrocket yeyhz

 

[Nintendude]

There's still kind of a limit on how far you can get without facing out-of-region players. And so what if your top player is rated really highly? We can't just magically assume that he's way overrated until he actually plays vs. other top people, so the rating isn't invalid at all.

 

[FoxLisk]

Also, FoxLisk, the search function appears to be a little buggy. Not sure if this has been pointed out before, but even if I type in someone's name with the exact matching characters, the website returns "Sorry, no matches found. Did you mean one of these players?"

haha yeah the search function is not very well done. i'll get around to fixing it sometime soon.

in the lifetime of competitive melee in aus, there's been 4 guest melee smashers: captain jack, ken, m2k and mango.
those 4 tournies aside our rankings gunna skyrocket yeyhz

that's great and all but you guys have GOT to start using tio 1.2. all these 1.1.3 version files are giving me headaches

 

[OkamiBW]

One thing I'm wondering about. Might not affect too much stuff. But, basically, if you have a bunch of people who have lost a bunch. Who are like, say, rank 700. If someone who's say 1100 or whatever beats them, no big deal. But what if all of a sudden a bunch of new players who are bad come in and play at the tourney. The 1100 player beats them all. Does his rank increase a bunch since the new players haven't lost much yet?

 

[FoxLisk]

One thing I'm wondering about. Might not affect too much stuff. But, basically, if you have a bunch of people who have lost a bunch. Who are like, say, rank 700. If someone who's say 1100 or whatever beats them, no big deal. But what if all of a sudden a bunch of new players who are bad come in and play at the tourney. The 1100 player beats them all. Does his rank increase a bunch since the new players haven't lost much yet?

itll increase a decent amount, yeah. but a) not much, b) it'll even out, and c) that's fine; if we have a bigger pool of people, he's going to be better than a larger number of them, so it makes sense for his rating to rise up a little to indicate that.