ladder's a joke

[zz-]

please make it so that when i play perfect i win

[Clapon]

bro your perfect<<<<<<<<<<<<<<majority of all good players on alty's cruise control.

get over it you not that good brah soz.

[zz-]

whatever

i'd recommend 5v5 so that an individual can have an impact on the game rather than being a slave to the rest of his teammates sucking

i have 13 losses and 12 of them i have most kills and most goals

[Clapon]

lol please just stop now before i die of laughter

[SkyGirl]

lol my sides are hurting!!!! Please stop please!!!!

[Herodadotus]

Originally Posted by zz-

please make it so that when i play perfect i win

Ladder is about teamwork. Just because you had most kills on your team, doesn't mean your team was better. I've been on teams that should've won easily, but lost due to poor tactics, a mistake, or dumb luck. Gotta play with the hand dealt to you, I think.

[zz-]

my point is really that there are no players with more than 50 games played that have even a 60% winning percentage. i realize that ladder balances teams according to rating and makes games progressively harder as you move up in the ranks but the point of ladder is to measure individual accomplishment.

i almost feel that making teams completely random and allowing the better players to be on the better team as often as they're on the worse team would be a more accurate measure of how good a player is --- working for a very strong winning percentage, rather than scrapping and scrapping and losing game after game climbing uphill against your own team and trying to eek out a 55% wins in progressively more unfair matches.

[eth]

You make a ****ty topic and expect people to take you seriously. wat

It's weird you say the system doesn't work, because I only see good players at the top. Instead of insisting everything around you is doing it wrong, maybe it's something else..

Ladders aren't meant for people to have 100 % winning percentages. If you want, I can set your percentage to 100 %, but then you'll have to never play again.

[zz-]

alright please set 12 of my losses to wins, i did deserve one loss.

i do agree that the system works fairly well, but it's subject to such a wild amount of variance with the current setting. I mean, right now the #1 player had a 12game losing streak at one point. I realize the ladder is all about winning long-term, but a system that names a player as better than everyone else when he is only 20 games or so removed from losing 8 games in a row just speaks to how streaky and ridiculous it is. Don't tell me none of you have never been downright frustrated with the lack of control you have over your ladder ranking over the course of a given week.

There should be some sort of decrease in the penalty your rating takes when you lose and finish the game with most goals and/or most kills. I understand that stats are not completely representative of a players contribution to his team but at the moment the ladder fails to differentiate between a player that goes 38-35-20 and scores 3 goals and a player that accomplishes nothing.

[tyr]

Originally Posted by zz-

i'd recommend 5v5 so that an individual can have an impact on the game rather than being a slave to the rest of his teammates sucking

I suggest you play the real mode, which is played 5v5.
It may suit your exceptional skillz better.

[mlopes]

I think 1v1 ladder would be just fit for you.

[Evan20000]

This thread gave me my morning lulz.

Also, having seen the OP play, I can confirm that he indeed does suck. (If you're ingame name is ZZ anyway. There was a kid using that name)

[silent skies]

Originally Posted by mlopes

I think 1v1 ladder would be just fit for you.

Is duel ladder ever going to come out? I'm kind of looking forward to it tbh.

[tyr]

As far as I know no real work has been done and there are no plans to do so at the moment. That being said, eth included the duel category to the website right from the beginning and I'm pretty sure the code for the log parser thingy won't be hard to write. So if enough people express interest in it and if someone make one or two good servers available I see no reason not to create it.
You can start a petition or something if you really want.

[elxir]

Has nobody noticed that one bad player hurts teams sooooo badly, and we play 5v5 or 6v6, and lamster has repeatedly said the game is balanced around 7v7 or 8v8?

[tgleaf]

Originally Posted by elxir

Has nobody noticed that one bad player hurts teams sooooo badly, and we play 5v5 or 6v6, and lamster has repeatedly said the game is balanced around 7v7 or 8v8?

You wanna play 7v7 TBD and 8v8 BALL? Oooh, fun.

[Esoteric]

This post is an explanation about Altitude and the ladder as well as a response to zz-, personally.


The ladder doesn't take kills/assists/bomb hits/everything else to disincentivize people from being even more obsessed with personal glory over the success of the team. Let your better-suited teammate take the bomb. Don't bother wasting shots shooting people who are already stalled and headed towards the wall. Hang back with a shield and reflect the bomb that was headed towards your base, rather than trying to get some extra kills with it. Play another plane, which you're objectively worse at, to help out the team. Act as a meatshield for the bomb-runner. Steal the neutral bomb out the enemy's clutches and then drop it. Leave defenders badly wounded instead of killing them outright in ball, so that a charge in 10 seconds will rip through them. Pass the ball, even when you won't get an assist.

These are just some of the many situations where there would be an incentive to play badly if the game cared about individual ranking. People would be playing for personal glory rather than for the team. With our system, playing correctly in all of those situations, even when they require you to be selfless, is rewarded. Give your TEAM good stats rather than trying to take them yourself.


As for win rates, with our ladder system EVERYONE'S win rate will, after thousands of games, end up in the 49-51% range. It is one of the most useless stats that we display. If you're obsessed with getting a high win rate, ladder is not the place for you. There are many newbie servers that afford you the opportunity to win far more than you lose.


In regards to variance, yes. There's lots of it. Consider people to be within +/- 200 points of their "true" rating once they've settled. This means that people might have a 16-game streak as the go from the top of that range to the bottom. This is a problem where there's no good solutions, only tradeoffs.
However, even in the best of circumstances we could never make it so that people didn't streak 10 games occasionally. Losing at least 10 games in a row with 800 played (assuming they are roughly balanced.) has a probability of 1-((1-(2^10))/(2^10))^800 = 0.542341368 They will have either a win or a lose streak with a 0.790708627 probability, over 79% of the time. This is the cost of doing business when you're trying to make games have equal chances of either team winning.


And finally, of course playing perfectly doesn't mean you always win. That wont be true in anything where you're on a team. If you are looking for this style of play investigate duels. Though you'll be able to complain about powerups and spawn locations even there. You may want to look into Chess or Go if games without chance interest you. Most of us enjoy a little chaos from time-to-time.

[nesnl]

Originally Posted by silent skies

Is duel ladder ever going to come out? I'm kind of looking forward to it tbh.

Dont think it's every going to happen because there is no match making system. People would just become rating snobs and refuse to play anyone lower than them, etc.

[hurripilot]

I'd take on any comers :P

[elxir]

Originally Posted by tgleaf

You wanna play 7v7 TBD and 8v8 BALL? Oooh, fun.

i'm just sayin, there's a reason 1 player ****s you so badly

[silent skies]

Originally Posted by hurripilot

I'd take on any comers :P

Ditto.

10char

[Boko]

Originally Posted by silent skies

Ditto.

10char

On topic: you can also play ladder for the good games and not for the rating?

[beefheart]

Wow esoteric nice explanation man!

I do have a question though:

How is the individual rating composed and is this fair?

I can see from observation of the ladder list that 2 factors matter: 1) win percentage and 2) number of games played.

I understand the importance of the second factor, because the more games you played the more reliable is the winning ratio.

But, the weight that the "number of games played" contributes to the total rating seems constant and this would be unfair imo, because then you can be number 1 in ladder with only a slight winning ratio, but a lot more games played then the rest.

What i would like to propose (if the previous part is close to the truth ) is that factor 2 (number of games played) contributes to the total rating with diminishing marginal weights. So that means that the more games you play the less "number of games played" contributes to the total ratio each game.

[tgleaf]

Originally Posted by beefheart

But, the weight that the "number of games played" contributes to the total rating seems constant and this would be unfair imo, because then you can be number 1 in ladder with only a slight winning ratio, but a lot more games played then the rest.

What i would like to propose (if the previous part is close to the truth ) is that factor 2 (number of games played) contributes to the total rating with diminishing marginal weights. So that means that the more games you play the less "number of games played" contributes to the total ratio each game.

I've been thinking about this a lot, too. If your win percentage is 56%, for example, and you get 25 points per win (just as an example; I know that's not accurate), then for every x games, you would increase 56%*25*x = 14x points. If you keep playing and keep your win percentage constant, your rating keeps improving by that much, so that a player with an additional 10 games would be 140 points above someone else with the same win percentage.

Hence this situation:

[drunkguava]

but like eso said, win % isnt the point of ladder. as an individual, the point is to get wins. as your ranking goes up, it gets harder to earn wins, so your % goes down. Ideally, if your rank was perfect and there was no variance you would win 50% of your games once you reached your rank, and your score would not change.


overall, the point of ladder is to create balanced games. Winning ~50% of your games shows that the system is working.

[nobodyhome]

Originally Posted by tgleaf

I've been thinking about this a lot, too. If your win percentage is 56%, for example, and you get 25 points per win (just as an example; I know that's not accurate), then for every x games, you would increase 56%*25*x = 14x points. If you keep playing and keep your win percentage constant, your rating keeps improving by that much, so that a player with an additional 10 games would be 140 points above someone else with the same win percentage.

Hence this situation:

You're missing the balancing factor in this thought experiment. The thing is, if your rating is 2400 or something else really high, and your win percentage is still 56%, that means you're a RIDICULOUSLY STRONG player, because every time you play you get paired up with the lowest of the noobs and yet you still manage to win more than 50% of the time. If you manage to keep your win rate above 50%, then as you get higher in higher in rating, you get paired up with crappier and crappier players, eventually making it so hard to win anything above a 50% rate (at that point, your rating can be said to be accurate).

Yes, you can point to the fact that both #1 players on the ladder right now are the ones with the greatest amount of games as evidence to your theory, but I can just as easily say that they're the best because they've had the most practice.

[nesnl]

I think it's time for a ladder reset and a new ladder rating system as this old one seems to not work very well.

[tgleaf]

Thanks, nobo.

[Evan20000]

How would a new rating system work, Maimer?

[Boko]

Where did nobo get this rating system anyway? Starting at 1500, I'd guess WoW Arena?

[nesnl]

Originally Posted by Evan20000

How would a new rating system work, Maimer?

I would like to see a rating system that gives points out based on some system that gives you more or less points based on your rating as compared to the average rating. As it stands right now you can be rated 2400 and as long as the two teams have average ratings (even if the average is 1000), then you get 25 points. I would rather see a system that gives less points to people who are rated higher than the average and more to the people who are rated less. This would also encourage people to play in games with people around their similar rating in order to actually increase their rating.

[nesnl]

Originally Posted by Boko

Where did nobo get this rating system anyway? Starting at 1500, I'd guess WoW Arena?

Starting at 1500 is a pretty standard thing. I think it was first used by competitive chess when the Elo system was used. So WoW Arena was by no means the first computer game to use that system. It makes sense to use 1500 only because it's a standard that everyone seems to recognize. I know that if I play a game tomorrow and it has a rating system that most likely 1500 means "average."

[Sarah Palin]

@leaf

Every game is +/- 25 now that we have autobalance. At best when the server can't perfectly balance it's +/- 24 or 26.

It would be impossible for a player to maintain a 56% win percentage because every time he wins he is assigned 25 more points. He will be assigned worse and worse teammates, until he starts to lose more often. When his win percentage dips below 50% he will start to gain points on average, and so on.

You can see that ladder works because all the players with the MOST games (SSD, Pein, Apa) have almost exactly 50% wins.

Players who have lots of games and win percentages above 50% are actually overvalued on the ladder. The more they play, their win% will fall to 50% and their rank will decrease.

However you may have noticed that everyone in the top 100, even with lots of games, has a greater than 50% win record.

Other than dodging games to preserve an unnaturally high ranking (coughMaimeronballladdercough), there is one other effect which keeps people's win%s from converging on 50%, and that's the constant influx of new players with 1400 ratings. A new player with 1400 points either has points he doesn't deserve (in which case above-average players will steal them) or he has fewer points than he deserves (he will take the remainder from below-average players).

The "inflationary" effect of new players each with 1400 newly minted ladder points constantly being inserted into the system means the best and worst players ratings cannot remain stable, but they will diverge: the top of the ladder will keep climbing and the bottom will keep sinking.

The players who have played the longest and against the most players on the ladder, benefit the most from this inflation, which means their ratings are slightly "pumped up" compared to those with fewer games.

This is not intentional but an unavoidable effect of letting new people join ladder.

[Sarah Palin]

BTW this is also why you should vote for Maimer for next Laddor Prezident

[tyr]

Maimer is already Laddor Prezident. People seem to forget that ..

[beefheart]

Originally Posted by nobodyhome

You're missing the balancing factor in this thought experiment. The thing is, if your rating is 2400 or something else really high, and your win percentage is still 56%, that means you're a RIDICULOUSLY STRONG player, because every time you play you get paired up with the lowest of the noobs and yet you still manage to win more than 50% of the time. If you manage to keep your win rate above 50%, then as you get higher in higher in rating, you get paired up with crappier and crappier players, eventually making it so hard to win anything above a 50% rate (at that point, your rating can be said to be accurate).

Yes, you can point to the fact that both #1 players on the ladder right now are the ones with the greatest amount of games as evidence to your theory, but I can just as easily say that they're the best because they've had the most practice.

OK, I like the rating system very much, dont get me wrong and also the balanced games and the incentive to play as a teamplayer that stem from this system. But still the ranking of players is a bit unfair. Look at ball ladder where void is #1 and dart #8. You can't say that void is better than dart because void played almost 4 times as many games. So the comparing of players with different number of games played is unfair. What i'm asking for then is maybe a correction for this (if possible). Maybe in the form of 2 lists, one like the one existant right now and the other one to compare player skills and to see which player is the best.

[zz-]

^ this is a problem with the current rating system. At the moment your ranking is really really close to simply being (W-L)x25 + 1500. So moving up is just a matter of playing a ton of games and capitalizing on the fiat points (lol) and/or running above the expected 50% win rate. Or simply shooting high from being rated at 1500 and winning a lot of games early because your rating is inaccurate (for example, pinkshift). I understand that if you're not good enough to maintain 50% wins with the more difficult team setups you receive you will drop -- this is the good characteristic of the current system and how the system aims to get good players at the top. its just that the other methods of rising up are equally significant.

all that said, my only suggestions for making it better is to decrease the game size, find a way to lessen the penalty for losers that play well, or have shorter ladder "seasons" where games played will be much closer between all the players. if the ladder standings are decently accurate after a couple hundred games and variance is as intense as you say, wouldnt most of the shifting around now be rooted in luck and choosing easy games to play in?

edited because im posting drunk againz, i dont really even want a reply to this post. it got muddier as i added more to it.

[zz-]

So i was looking at the top players ratings compared to their (W-L)*25. it seems there are some players who received significant ratings boosts from playing in a lot of games with large edges (im guessing before the balance was put in place.)

is there any plan to resetting scores now that all the games are autobalanced? seems pretty bad to have unbalanced games mixed with balanced games in the same rating.

[Sarah Palin]

I also looked at a similar stat for top 50 players:

(Score-1400)/(Wins-Losses)=?

This stat should be very close to 25 for any player but in fact it has a wide range, not sure why.

[Evan20000]

Originally Posted by Sarah Palin

I also looked at a similar stat for top 50 players:

(Score-1400)/(Wins-Losses)=?

This stat should be very close to 25 for any player but in fact it has a wide range, not sure why.

Because in the early days of the ladder, you could stack teams as much as you wanted and point drop would deviate from 25. It still does, check some of the games you've played. There are a few 24s and 23s.