wanderrful

Points: 5,659, Level: 51

Activity: 0%

Yes! That is precisely why the upper bound on allowed forfeits must be n/2. I already explained that a few posts up to shadowpuppet.

http://commforums.com/showpost.php?p=51811&postcount=21

I already posted about this too!

http://commforums.com/showpost.php?p=51778&postcount=19

I think you're slipping man. Have you had your coffee?

To elaborate more on my answer to that question, though... it is a matter of probability — and that was not my favorite subject in mathematics, unfortunately. I'm not saying that it was my worst... I'm just saying that I'm not as motivated to do that stuff.

Yes! The success of the league is a direct result of the activity of its participants. I see no problem with this :x

I agree!

I agree! Do a Season of a Ladder to see which teams are where on the rankings and then use that information to divide people more fairly into leagues. Oh wait, we already have those. They're called Pre-Season games. Wait... I might be lying there — are preseason games predetermined or are they just a free for all? I'll never kno

Hawkeye

Points: 5,054, Level: 48

Activity: 50.0%

preseason games are FFL that don't count against anyone..

wanderrful

Points: 5,659, Level: 51

Activity: 0%

guess what guys? get ready to want to bear my children...


...because i just solved the problem of fairly deciding which cut to make! turns out its not a probability question:

click here because the forum re-sized the smart picture




oh yeah baby

wanderrful

Points: 5,659, Level: 51

Activity: 0%

i totally feel like john hodgman during his "your welcome" skits on the daily show right now

edit - i should get a fuckin internet Ph.D for that shit

wanderrful

Points: 5,659, Level: 51

Activity: 0%

im such a fucking baller


edit -- so yeah. with this, you don't need a rating algorithm to match teams up based on their stats!



double edit -- btw sorry if it seemed like i was being rude a few posts up hawkeye. i was just flowing there and i just wrote what came out naturally without thinking too much about it. WOO!

triple edit -- holy shit am i going to have to try to get a patent on this idea? i've never heard of this kind of thing for a tournament before. i hope its original!

wanderrful

Points: 5,659, Level: 51

Activity: 0%

fyi guys i was writing out explicitly the logic for helping shdwpuppet to understand how to program this and i ended up finding a small error in the whole "figuring out the fairest way to decide on the first cut" part.

i will rethink and let you guys know

wanderrful

Points: 5,659, Level: 51

Activity: 0%

so yeah, here's the most fair and objective idea that i've come up with so far to replace the "let the forfeit-winning team choose one of their own matches to cut" thing:

Create a list before the regular matches begin containing each of the relevant teams. This list will be cycled through every time a forfeit match occurs and the team that is currently selected will be the one to choose one of their own future matches to remove. After the choice is made, the next team on the list will get their turn whenever the next forfeit happens.

Intuitively, if a team is disqualified and replaced by the BYE team, they will also be removed from the list.





The remaining question is: how do you decide on the order of teams in such a list?

My suggestion: base the order on the pre-season games. For each team that did not participate, place them randomly after the teams that did. The best records are put at the top of the list and it goes in descending order.

If two teams have the same record at the end of the pre-season, randomly choose who will go first, second, third, etc. amongst the tied teams until there are no more left.






PS: i've finished writing out the logic behind the whole removing of matches thing to help you write the code for it. I'll make another long picture deal for you so its easier on the eyes.

wanderrful

Points: 5,659, Level: 51

Activity: 0%

Oh yeah, baby YEAH!

Here you go shdwpuppet and probably Hawkeye because I'm sure he'd be very interested in seeing how I did with trying to translate this mathematics into programmer talk.

click here because the forum resized the smart picture



God, I'm so handsome...

defiance

Points: 2,864, Level: 34

Activity: 8.3%

you need a life nerd

wanderrful

Points: 5,659, Level: 51

Activity: 0%

says the competitive TF2 player



anyway i was taking a nap earlier going through the logic and i realized that the pseudcode-ish thing at the end of the last picture is missing some conditions. its cool but its not complete. i will try to think more about it


PS: why the hell did they take away my super moderator abilities - i want to ban this asshat.

wanderrful

Points: 5,659, Level: 51

Activity: 0%

god i'm handsome

defiance

Points: 2,864, Level: 34

Activity: 8.3%

Prolly cuz you post saying how everyone is an ignorant asshole, even though you literally define the term.












but how mad?

wanderrful

Points: 5,659, Level: 51

Activity: 0%

you're a fucking troll. you have no room to speak. with me, my arrogance is actually called for because i'm smarter than you. now get the fuck out from under my bed and go make something of yourself.

the problem with you trolls is that nobody is allowed to be happy, feel positive, or even find solace about themselves when you're around. there is no gray area for you on these things.

everyone who is happy or feels an ounce of self-satisfaction is viewed as not only comparable to but indistinguishable with ric flair in your eyes:

MjrNuT

Points: 1,634, Level: 24

Activity: 0%

wanderrful,

I've been following your approach and it looks great. I like numbers too, but also detest probability.

Might be worth some weight for the Visual Learners to use your system on a mock season. Or take existing season completions (eg., CEVO S-5) and run through it to see how it rolls out. Would be great testing yes? TWL would even be another great example to mock up?

I'll help if you'd like. ;) At least you're putting up. :D

__________________
MjrNuT
Flames and Ash Gaming

wanderrful

Points: 5,659, Level: 51

Activity: 0%

Excellent idea!

Does TWL keep a good enough record of match results for me to reference while I make a picture of how it plays out?

Or how do you think it should be done

MjrNuT

Points: 1,634, Level: 24

Activity: 0%

In looking at TWL, it seems since playoffs started all the regular season stats are not available. That was my cursory look and I didn't bother to login.

CEVO S5 appears to still be in tact though..at least until S6 starts...??

btw, I can't see the image in #38 at that well. Your hyposthesis here is to improve the pairing, especially in the midst of FFs, correct?

So my suggestion was to put some fiction or nonfiction data through it to test it. Does that make sense to prove it? How else would you put stock into to your effort here? ;)

__________________
MjrNuT
Flames and Ash Gaming

improperdancing

Points: 380, Level: 7

Activity: 21.2%

Forfeits aren't independent to TWL, though. We had like four FFWs in ESEA and one or two in CEVO-A last season.

I think the main way to prevent teams from constant forfeits is for the admins of all the leagues to pay closer attention and make sure that teams that forfeit several matches in a row don't keep getting scheduled for matches. This is one of the big problems with ESEA, where they keep scheduling a team even if they have two people on their roster, and it basically just steals money from all the other teams that paid to play, not to get six people in a server at match time and type ".reportffw"

CEVO does the best job of this I think. ESEA ignores it completely. TWL is somewhere in the middle.

Hawkeye

Points: 5,054, Level: 48

Activity: 50.0%

Absolutely agree here, it just seems very prevalent in TWL over the past couple of seasons.

Yup, if you forfeit twice during a season, you are gone, If scheduling or something was simply an issue, see what can be worked out with the admins or whatever.

The way TWL slits the teams into skill divisions, it actually makes the problem seem a lot bigger, since you have significantly smaller pools, having 2 teams nasty, while CEVO has one large bucket basically that handles things.

wanderrful

Points: 5,659, Level: 51

Activity: 0%

lol i think that you don't understand what a proof is becuase i actually did that in the first 3 picture posts.


i think that what you meant to say was that i should demonstrate it, or that i should prove that it works in a single case, rather than in general (which is what i did).


but yes -- do you think i should just make another picture post with how it plays out or make an HTML page or something or what?

i think i'll go for the picture post


edit - this is a LOT of fucking effort. is this -really- necessary? it works the exact same way as a round robin, but when a forfeit happens you just remove select matches so that everybody gets credit for the same number of matches

shdw.puppet

Points: 3,599, Level: 39

Activity: 66.7%

to be adoptable, either a league would need to bite the bullet or someone would need to run through a whole league with the history of a TWL division (I chose because it is smallest at only 8) to show that it is fair etc.

It would also need to be done with a huge number of teams (aka ESEA/CEVO) to show that it can work in a limited number of weeks with such a large number of teams. Round robin isnt used for a reason bro :D

MjrNuT

Points: 1,634, Level: 24

Activity: 0%

That is exactly what I was saying b/t the lines Mr. Wanderrful, since it seemed others were not "catching" on to your proof judging by the lack of responses to it. Hence my suggestion....






Also what I was inferring. How else does a system become adopted without mock testing? You can be sure that the Swiss pairing was done extensively. Most importantly having its assumptions predetermined as well as insitu.


Carry on then

__________________
MjrNuT
Flames and Ash Gaming

wanderrful

Points: 5,659, Level: 51

Activity: 0%

umm I don't think I understand what you guys are talking about when you say that I should "run this through a twl division as an example to be sure that its fair". I fucking proved it.

Besides, how am I to know which match THPS would have chosen to eliminate from their future to-do list? Its not possible to do using twl's history (which isn't even available from what i can tell anyway)

shdw.puppet

Points: 3,599, Level: 39

Activity: 66.7%

there is proof and then there is a good testing routine. I am not doubting the solution, I am saying that it needs to be run through a season before any league will pick it up.

And match histories are available on the team pages.

Thefriend

Points: 7, Level: 1

Activity: 9.1%

I played TWL back in my DOD:S glory days. Didn't know things fell apart while I was away.

improperdancing

Points: 380, Level: 7

Activity: 21.2%

TWL's scheduling would probably be fine if they only allowed a week to play matches (which they should). Allowing two weeks just fucks everything up when it comes to Swiss Pairing, and that's why the scheduling was such a disaster this season.

Look at UGC. Not the best league, but they use pretty much the same scheduling system and it works just fine because teams only have one week to play their matches. We won every week last season and pretty much always played another team in the top five.

Then again, there's no accounting for competence I suppose. In TWL our team played the top-ranked team one week and the bottom-ranked team the next in a division with fifteen or sixteen teams. How that happens I have no clue.