1-4 & ranked 17th

retrospekcja_jeza defeated [REGL] majokun 1866538 12/28/2011 7:34:26 AM retrospekcja_jeza: 2072
Soyrice defeated [REGL] majokun 1871795 12/28/2011 7:17:36 AM Soyrice: 2113
[WM] Dazed & Insane defeated [REGL] majokun 1866542 12/27/2011 5:48:52 PM [WM] Dazed & Insane: 2131
[REGL] majokun defeated FBG-Chadstroma 1866532 12/27/2011 10:33:34 AM FBG-Chadstroma: 1769
[WM] Nord defeated [REGL] majokun 1866528 12/26/2011 8:11:10 AM [WM] Nord: 2144

There are several people that have scores way higher/lower than they should. Every win/loss should count for the same amount on the seasonal ladder.

|> *Every win/loss should count for the same amount on the seasonal ladder.*

If everyone played everyone, like in a round robin tournament, that would work and produce fair ratings. However, given that you only play a subset of the players, it's much more fair for wins against strong opponents to give you more rating points than wins against weak opponents.

And if you don't get to face stronger opponents you have no chance to make it to the top of the ladder. Also it puts the people who started first at a disadvantage since the players who started later, all faced people with much higher ratings.

I've seen several players that are 5-0+ and aren't even in the top 25. Not a fair scoring system at all.

Isn't it the same as the 1v1 ladder? Same issues apply except for the limited number of games. Each game should stabilize the rankings a bit more.

The big difference is that the game assignment is random, correct? Matches are not assigned based on rating but randomly among all participating in the ladder. I am interesting in seeing how things sort out with each successive game. I'd say, save the analysis until we have a few more games finished. However, someone who knows the math behind the rating system better than I might step up with an explanation.

Seems to me to be an issue of sample size.

Most players are only at 5 completed games which is meaningless statistically. Granted, the end total of 15-16 games is also rather meaningless but that is a known limitation of such a short season. Even then, it should provide far more information and more accurate rankings by the 15th game than it did for the 5th.

D&I, part of the issue is that everyone starts at the same rating, which will be inaccurate. And the relative ratings are calculated based on the ratings of your opponent. So initially everyone's win loss adjustments are very skewed by the inaccurate ratings. These only settle down when as a whole player's ratings adjust to closer to their 'true' value.

Me: 6 wins, 1 loss & rated 54th. *shrugs*
Kinda sucks. The current ladder system is suitable for a continous ladder, but I fails for a seasonal ladder with limited games. Players cannot get to the top solely by skill, but are largely dependent on pairing luck.
What we need is divisions, say 20 players per division, playing eachother in 19*19 games. New players would have to start in the lowest division, with top three moving up, lowest 3 moving down at the end of each season.
Or at the very least have some kind of playoffs between the top ten rated by Elo rating and those who won all games but didn't make it to the top ten because they were pitted against too weak players.

i thought it was interesting to play with less fog, had my fun, and have since quit the seasonal ladder. i made a double-elimination tournament of 16 and invited 50 guys i thought were the best at 1v1s. that is enough for me.

if you don't like the ladder, make a tournament or create your own league.

@Diabolicus, I really like the promotion relegation idea. Adds a lot of interest at both ends of the spectrum and might be able to handle a straight foward win/loss system.

What to do about ties though. No goal differential to settle a tie. If a tie breaker could be determined, it could work. Maybe a straight win/loss with a tie broken by the ELO rating?

Just randomly lost 12 points while several people ranked below me gained points, though they didn't win any new matches. The more I look into, the crappier the seasonal rating system gets.

Dazed, I would assume that their past opponents did well while yours faltered. Makes sense to me. Same thing happens in the normal ladder.

what about the gap in # of games??? I saw somebody with 19 games where I'm at 9 atm

I think we should wait until the seasons over and all of the games are played until we criticize the current system, because it could even out by then.

Not like I'm part of the ladder, but if the point of the seasonal ladder is to give a meaningful rating only after all games have been completed, it is indeed unfair to criticize it before the season has finished.

I believe the seasonal ladder should have the swiss system, developed exactly for events like these...

I believe the seasonal ladder should have different leagues, with a smaller amount in each.. mainly to offset this problem.. 130 players playing 16 games each is not enough *Imho* to give an accurate rating for noticeable amount of players..
in the 1v1 ladder you have time to get to your place and then play games at your place that will determine your scores.. sometimes it takes more games due to bad luck in game placement, but it will happen..
in the seasonal ladder, someone could literally get lucky and get 0-16 and have a pretty decent score..
I might have to field a test with the program you showed us previously to see how it might turn out..

Useless tidbit of information.. the likelihood of the current 1v1 ladder occuring is -3727.47...

I just did a lil' ghetto test with a fictitious league of 34 players... my 0-16 guy only ended up in 25th with a 1300 ELO..

so then I decided to input a new player in a replica of the seasonal ladder and give it losses against 8 of the top 9 players..
the result is that player is now ranked 57/130 with a 1538 rating and a 0% w/r...

does anyone know why the ratings are much higher on the seasonal ladder than they are on the 1v1 ladder? and why the ratings on the 2v2 ladder are much lower?