Nullpo Tetris League: SEASON TWO HAS BEGUN

[Rosti_LFC]

I screwed up. I kept writing 'played' when I meant 'won', I think. You answered one question, though; now the other: at what point should points from losses overtake points from wins? Say, if player A plays twice as many games as player B but wins the same amount, should they be ranked over player B? 2.5 times as many? 1.5?

If player A has played more games than B and they have the same wins, then B should be ranked above A usually.

I'm not even going to bother thinking about how many times they should be ranked over each other - it's a meaningless question. In a league, by the end of the season, everyone should have played everyone else, so debating where people should be if they've played more/less matches is pointless.

Also, if only a minority of the total matches have been played , then "wins" are a also pointless statistic. You could have two players who are the same level and placed roughly in the middle of the division. If one of them plays most of the top players in the division, and the other plays most of the bottom players, then clearly one is likely to have far more wins than the other one (and more points), and it'll only be when they've both played most of their matches that things really start to show through.

And you can't define a point where points from losses should overtake points from wins - it's not that simple (and even if it was, it's still just a constant that I'm changing mostly arbitrarily). One loss is not necessarily equal to another loss, in the same way that one win is not equal to another win. It depends on the skill if the person you're playing against, and of the points margin you manage to score against them, whether you win or lose. I'd be happier with a 15-10 defeat against Blink than a 15-13 victory against someone like me. You'll get more points from the 15-13 win against me, but that's not the beginning and end of it, because what's really important is what everyone else manages to score against the same player, and that's something that will be reflected regardless of how I tweak the scoring system.

[myndzi]

I'm not debating what should be - I'm asking what you want. You seemed unhappy with the arbitrariness of our current solution, so I am trying to get you to define the parameters, in words, of how you want it to work. If you don't want to, then just say so.

As for not every game being equal, you can hardly hold that against me, since I was approaching things the way you are already doing them. You *already* have a win/loss balance, you just haven't worked out what it is.

For example, if you play 8 games and someone else plays 7, and 6 of those games are equivalent in the points they give you, then if you lost both the remaining games by 5 points and the other person wins their game by 5 too, your two games are equivalent to his one. This equation is not linear, I think, but it is still present regardless of what you say about not caring how losses stack up against wins.

Another for-instance: loss points give some incentive, but there is no incentive to play, say, the person you are likely to place under. No matter the score, you will be increasing their potential lead over you if they win.

There are other options, but you don't seem to want to bother deciding how your scoring system should behave, so I have no way to know which ideas are relevant. Here's two anyway.

Alternate losing game incentive: final score is win points * percentage of games played. This dissociates the incentive from playing any particular player, and also makes moot in some ways the question of wins vs games played.

Points taken from better players are worth more: this one is simple really. Modify loss-points by the number of wins the opponent ends the season with. To make the reward more closely approximate the player's skill, you could modify it by the sum of their win margins instead. This method needs careful examination for balance, but it wouldn't be difficult.

[Rosti_LFC]

Even if I am unhappy with the current scoring system being arbitrary, how exactly is switching to another arbitrary system going to solve that (especially if it's hideously complicated)? And I'm not unhappy with it. On the condition that all games get played, I'm perfectly happy with it.

I wanted an incentive for people to play as many games as possible, and with the exception of 15-0 defeats, I have that. I wanted the scores to matter, to value a 15-13 scoreline below a 15-3 scoreline for the winner, and value it above for the loser and I have that.

[myndzi]

Neither of those suggestions are "horribly complicated" - in fact, they are very easy to implement in a spreadsheet, which was the idea. Neither are they arbitrary - they are based in a mathematical representation of a particular desired behavior.

I did not suggest switching to either of those as "the scoring system"; I suggested defining the behavior of your ideal system so that it would be possible to approach the how of implementing it. I gave a couple examples of ways that the current system could be changed to better meet certain hypothetical requirements.

My very first post was only made because it seemed from your posts in this thread that you weren't quite content with the scoring system. If that is not the case, then there's no reason for me to attempt to contribute, eh? All you needed to say was "I like it the way it is."

[Rosti_LFC]

Neither of those suggestions are "horribly complicated" - in fact, they are very easy to implement in a spreadsheet, which was the idea.

They might be easy to shove into a spreadsheet, but they both detatch the player from being able to know how for sure how the outcome of the game will affect their rank without waiting for me to update or running a parallel spreadsheet. I could code some fancy recursive retrospective ELO-based points system if I really wanted to, but I doubt it'd make much difference to the outcome of the league, and it'd make players completely unable to follow their own scores from match to match.

Maybe I should have clarified that I was content with the way things were having run a few experimental runs and tried out various other ways of working the points system, but your initial post came across to me as more patronising  (like I didn't understand the finer mechanics and implications of the points system I was using) than suggestively helpful.
If I misunderstood, then hey, sh** happens. I apologise for the needless arguing.

[coolmaninsano]

Rating system from a Scrabble site I played on:

Code Select Expand


r1 + k * [w - 1/1+10^((r2-r1)/400)

k = 20+(|d|/10)

d = Score difference*

r1 is your rating, r2 is the opponent's

Let w be 1 if you win and 0 if you lose.

*since it's a Scrabble formula, it's designed for score differences from 1 to 200. For the score differences only from 2 to 15, maybe have |d|/5 or 3.

[myndzi]

...your initial post came across to me as more patronising  (like I didn't understand the finer mechanics and implications of the points system I was using) than suggestively helpful.

Ah, you mean the "magic of math" one? Yeah, I was going more for whimsical, but I gbuess I can see how it could be misinterpreted. Sorry for making a mess in your thread needlessly

P.S. - understanding the results of your match down to the point is overrated  I'm actually serious here - i'd trade knowing easily for a system where, say, wins vs blink got me more than wins vs some noob in a heartbeat... but then again i'd probably put in the effort to understand the details anyway.

[Shizi]

shizi - 41.91

[Rosti_LFC]

i'd trade knowing easily for a system where, say, wins vs blink got me more than wins vs some noob in a heartbeat...

Except in a league where everyone plays the same matches against the same people, there's no reason to value winning against one person against winning against another.

[!--quoteo--][div class=\\\'quotetop\\\']QUOTE[/div][div class=\\\'quotemain\\\'][!--quotec--]hes like 30.03 on bb and 29.92 on lj[/quote]
Then why the hell is Paradox only quoting his TF score when he's clearly got faster times? Especially given he's one of the people who is against using TF, yet such a slow time will place him in the TF division.

[BentoBoxer]

BentoBoxer Sprint: 30.26

[Paradox]

lj? and the Blockbox scores don't even work

I told you already my best time is 28.407 on Keyblox- no deep drop but you wouldn't accept it so what are you complaining for.

[coolmaninsano]

lj?

LJ = LockJaw

Also, this topic should be pinned if it's the official league thread.

[chopin]

This post has been edited by Rosti_LFC: An hour ago

How did you do that

[coolmaninsano]

This post has been edited by Rosti_LFC: An hour ago

How did you do that

Moderator fail xD.

[crzy242]

lj? and the Blockbox scores don't even work

I told you already my best time is 28.407 on Keyblox- no deep drop but you wouldn't accept it so what are you complaining for.

your score is on the records page.
why would he accept keyblox when you cant use a 40 key layout on nullpo/tf?