Forged Alliance Forever Forged Alliance Forever Forums 2014-03-15T11:22:20+02:00 /feed.php?f=2&t=6249 2014-03-15T11:22:20+02:00 2014-03-15T11:22:20+02:00 /viewtopic.php?t=6249&p=68974#p68974 <![CDATA[Re: 1v1 Ladder Ranking]]> Statistics: Posted by Pierto — 15 Mar 2014, 11:22


]]>
2014-03-15T10:22:30+02:00 2014-03-15T10:22:30+02:00 /viewtopic.php?t=6249&p=68971#p68971 <![CDATA[Re: 1v1 Ladder Ranking]]>
"It doesn't matter WHY I'm losing, just that I'm losing."
I don't think this is a valid statement.

"Since everyone loses for BS reasons, this cancels out in the end".
You don't actually know this. Neither you nor I have the data to defend such a proposition. You should have the results of sufficiently enough games, of sufficiently enough players and be able to prove that bs losses didn't affect the true rating in the end. Neither you nor I can claim that have such data, so that "this cancels out in the end" is at least unknown.

"Your ranking is approximate. If it says I'm rank 1000, it knows I'm actually 950-1050 or so and just hasn't pegged me in just yet. However, if I've only played 2 games, my 1000 rating is actually 700-1300 and the best it can say is I'm 1000 ~for now. But if I have a million games, and it says I have 1000 ranking, it actually means 999-1001. Now I've been pegged and it's fairly sure I'm 1000 now."
Yes it is approximate, and tries to use an ever greater statistical sample to produce meaningful results.It also uses the "games played" concept, but for different reasons. Since the quality of the statistical sample can't be controlled some results turn out to be biased in the end. So, Yes I am praising the rating system and pointing some of its weaknesses; no contradiction here.

"I really wasn't trying to be mean the first time -cause a lot of people fall for that trap. But you responded arrogantly while riddled with spelling errs.".
Allow me to be as arrogant as I wish, because my "aggressiveness" didn't come from my ego only(ego should exist, it's not always bad). I usually display it to poke people too sure of their intellectual baggage. Being very sure of the validity of your knowledge is always a bad thing and handling it like this can lead to real deaths. Regarding my "spelling errs", well use your superior pattern recognition abilities to get the true meaning of my writings . Anyway, english has such a simple spelling, I haven't ever really understood, why you people think it's such a big deal.

"Er, right? That was the plan of the first argument? I honestly can't tell cause you starting spouting off some big words about AI not having the ability to be human -or something. Blah blah, supcom is hard blah blah, defining who is good is hard blah blah cause any random action can lead to a win/loss and the game can't say 'that move right there won the game, he needs to rank up' blah blah.
Something like that maybe? Again, you didn't really form a thought there. So I'm guessing.

Point is, none of that made any sense and what did has no relevance."
Err, no it was not. Sorry if I didn't express myself adequately. But I do think I formed a very concrete thought in the first argument and a very feeble one in the second. Shortly, yes supcom is more complex than chess and this is no bla bla.When you search for new strategies or tactics you actually search for new patterns.The more you know of them, the more efficient you are in game.
Even the low level functions, such as handling of controls improve over time or with more games. It's common sense that practising (more games played) make you better because they make you better at applying learnt patterns.

And one last thing related to the validity of the statistical sample.People don't play ladder that often, and though the matchmaking system tries its best to match you with fitting opponents, this will not frequently be the case because not much people play the game. ( this doesn't mean the gpg travesty was better; you were consistently matched with excellent players, because only they were consistently playing the game).And when they do, they usually engage in 4v4 or 6v6 fights that it takes them more time to "balance" than to play them.

Statistics: Posted by prodromos — 15 Mar 2014, 10:22


]]>
2014-03-15T02:46:36+02:00 2014-03-15T02:46:36+02:00 /viewtopic.php?t=6249&p=68965#p68965 <![CDATA[Re: 1v1 Ladder Ranking]]>
prodromos wrote:
errorblankfield wrote:^This this this OMG this.

No offense prodromos, but you fell for the mental rating trap. A 700 player with 5 games just as likely slightly better or slightly worst than a 700 player with 5000000E10000000 games.

SomeoneAUS says it better than I, but as you play more, your rating is more absolute. Once you've reached a lot of games, by definition you are a prime example of said rank. So our resident 700 guy with millions of games is shown time and time again unable to best a 701 opponent.


1. SupCom is much more than 2d chess and the reason why ai on supcom still sucks. We can talk all day about this, but shortly experience, ie the assimilation of greater number of pattern behaviours and the application of them, as understood by the modern science of ai, is what parts human intelligence from machine intelligence( for now). Yes, an experienced player may lack the neuromotor responses that contribute to the speed of execution of commands(for an array of different causes evey time), but neuromotor responses is the low level of life and no real match for superior intelligence. Insisting on the opposite view is like saying that a lion is smarter than you because it can defeat you consistenly on 1v1 body fights.

2. Even if point 1. were bullshit, there are still reasons that can compromise the useful system Zep had the kidness to apply for our own pleasure.A player may lose a match because he has things to do and the match goes on for long, but the server doesn't know that. A player may lose a match because someone's at the door or the phone rings, but the server doesn't know that. A player may lose a match, because he's pissed off about how the match progresses but the server doesn't understand that. A player may lose a match because his connection is lost but the server doesn't know that. A player may lose a match because he tries a novel strategy(maybe a risky one), but the server assumes the player always follows the safe route.
and on and on....
Shortly, though the logic behind the current rating system is sound, it assumes that the pool of a player's results is valid, even if there's no fucking way to verify this, not only for the current system but for every rating system.Thus, the "games played" and so indirectly the experience is a better measure for intelligence and thus for true skill. On a different level even the reflexes improve with more games albeit at different pace for every one of us.
Rushing to bash me in self content because you know the right meaning of a statistical definition is not gonna work.I am profoundly "cerebral", my curse and my salvation. Chances are that I see more aspects of a certain subject before you even
fathom their existence.
So I maintain that " games played" should be displayed together with the rating, if you want to better assess your opponent ,pre game that is. Now, there are some " quit"ters that defy even the games played concept, but hey, it's human intelligence it hates being confined by definitions.


Oh, fun.

Let's see let's see...

1. I'm glad the first part of #2 says 'even if all of my first point is wrong' cause it is.
If a lion beat me repeatedly in a 1v1 body-fight, then I would rank him higher than me -mentally. As a human, I should dominate this fight -I have an actual brain and all... But you meant like 'spur of the moment fight' so I can't you know, bring a gun.
But even thing, the 1v1 fight does give him a better ranking than I, in combat ability. Assuming you realize that such a metric would only matter for combat (no intellect), so you where trying to imply that I was also supplanting the interpretation of ranking?

Er, right? That was the plan of the first argument? I honestly can't tell cause you starting spouting off some big words about AI not having the ability to be human -or something. Blah blah, supcom is hard blah blah, defining who is good is hard blah blah cause any random action can lead to a win/loss and the game can't say 'that move right there won the game, he needs to rank up' blah blah.
Something like that maybe? Again, you didn't really form a thought there. So I'm guessing.

Point is, none of that made any sense and what did has no relevance. Our ranking is a measure of how well one plays supcom. It doesn't matter how good/bad the metric is at measuring this, but the fact remains it's some reflection of skill playing supcom. And if said metric is more confident at saying [person] is rated [rating] than I am of [other person] being rated at [same rating] then we have to take it's word for it. Now we can say the metric (ranking) is terrible, but if it's telling you that it's pretty sure I'm a 700 and only 50% sure it's 700 we have to think about what that actually means. If we like the metric, that means we know my (700) rating better. If we don't like the metric, we know my rating is really off.

Long short, the first argument attacked the validity of the metric NOT if Mr. 700 and 100000 games is better than Mrs. 700 and .2 games.

2. You can lose a game for a million reason, everyone is more or less equally affected by this. Heck, I'm lying, we aren't equally affected by this. My dog makes me lose more matches compared to those that don't have a distracting dog around. But that literally should be reflected in my rating anyway. It doesn't matter WHY I'm losing, just that I'm losing.

Since everyone loses for BS reasons, this cancels out in the end. If on average, people have ten BS losses, and you have 11, your ranking will show that you are relatively down in rank an amount equivalent to 1 BS loss. (Wow, that's meta... sorry if it's hard to follow, hard to explain in a logical way seeing as it's common sense we are trying to nail in.)

But again, this is attacking the metric. You then went on to praise the metric.

So how about actually looking at how the metric works?

Your ranking is approximate. If it says I'm rank 1000, it knows I'm actually 950-1050 or so and just hasn't pegged me in just yet. However, if I've only played 2 games, my 1000 rating is actually 700-1300 and the best it can say is I'm 1000 ~for now. But if I have a million games, and it says I have 1000 ranking, it actually means 999-1001. Now I've been pegged and it's fairly sure I'm 1000 now.

As mentioned, this is why I say a rank 1000 player with 5 games is just as good as a 1000 with 1000000 games. Why? Cause I know it's really a 940-1060 player vs a 1000 player. Which is more or less a 50% chance of winning -or the goal ratio ranking is suppose to create.

Work on your basics before you try to act all high and mighty. (Though I suspect that former wasn't an act.) I really wasn't trying to be mean the first time -cause a lot of people fall for that trap. But you responded arrogantly while riddled with spelling errs.
Do people not yet realize this doesn't work well with me? I'm quite stubborn when it comes to facts. Everyone is entitled to opinions, but if you tell me the sky is red I'll come at your throat if your only defense is 'cause I think so'.

The sky is blue. Ranking works kinda like the way I explained -but more complexity.

Statistics: Posted by errorblankfield — 15 Mar 2014, 02:46


]]>
2014-03-15T01:32:25+02:00 2014-03-15T01:32:25+02:00 /viewtopic.php?t=6249&p=68961#p68961 <![CDATA[Re: 1v1 Ladder Ranking]]>
errorblankfield wrote:
^This this this OMG this.

No offense prodromos, but you fell for the mental rating trap. A 700 player with 5 games just as likely slightly better or slightly worst than a 700 player with 5000000E10000000 games.

SomeoneAUS says it better than I, but as you play more, your rating is more absolute. Once you've reached a lot of games, by definition you are a prime example of said rank. So our resident 700 guy with millions of games is shown time and time again unable to best a 701 opponent.


No offense taken. But I am afraid you fell victim of the statistics theoretical meaning trap. A player with "1000000" games as you say , if he/she doesn't quit every one of them(like some very smart "mates") has huge experience even if he is a slow learner. Yes, he is absolutely incosistent, but if you happen to face them on their good moments, you will whine all over the forums about the weakness of the rating system or of the bad habit of smurfing.
Don' t get me wrong, I am obsessed with numbers and have great respect for statistics, but I feel you miss 2 important points:
1. SupCom is much more than 2d chess and the reason why ai on supcom still sucks. We can talk all day about this, but shortly experience, ie the assimilation of greater number of pattern behaviours and the application of them, as understood by the modern science of ai, is what parts human intelligence from machine intelligence( for now). Yes, an experienced player may lack the neuromotor responses that contribute to the speed of execution of commands(for an array of different causes evey time), but neuromotor responses is the low level of life and no real match for superior intelligence. Insisting on the opposite view is like saying that a lion is smarter than you because it can defeat you consistenly on 1v1 body fights.

2. Even if point 1. were bullshit, there are still reasons that can compromise the useful system Zep had the kidness to apply for our own pleasure.A player may lose a match because he has things to do and the match goes on for long, but the server doesn't know that. A player may lose a match because someone's at the door or the phone rings, but the server doesn't know that. A player may lose a match, because he's pissed off about how the match progresses but the server doesn't understand that. A player may lose a match because his connection is lost but the server doesn't know that. A player may lose a match because he tries a novel strategy(maybe a risky one), but the server assumes the player always follows the safe route.
and on and on....
Shortly, though the logic behind the current rating system is sound, it assumes that the pool of a player's results is valid, even if there's no fucking way to verify this, not only for the current system but for every rating system.Thus, the "games played" and so indirectly the experience is a better measure for intelligence and thus for true skill. On a different level even the reflexes improve with more games albeit at different pace for every one of us.
Rushing to bash me in self content because you know the right meaning of a statistical definition is not gonna work.I am profoundly "cerebral", my curse and my salvation. Chances are that I see more aspects of a certain subject before you even
fathom their existence.
So I maintain that " games played" should be displayed together with the rating, if you want to better assess your opponent ,pre game that is. Now, there are some " quit"ters that defy even the games played concept, but hey, it's human intelligence it hates being confined by definitions.

Statistics: Posted by prodromos — 15 Mar 2014, 01:32


]]>
2014-03-14T06:30:59+02:00 2014-03-14T06:30:59+02:00 /viewtopic.php?t=6249&p=68914#p68914 <![CDATA[Re: 1v1 Ladder Ranking]]>
No offense prodromos, but you fell for the mental rating trap. A 700 player with 5 games just as likely slightly better or slightly worst than a 700 player with 5000000E10000000 games.

SomeoneAUS says it better than I, but as you play more, your rating is more absolute. Once you've reached a lot of games, by definition you are a prime example of said rank. So our resident 700 guy with millions of games is shown time and time again unable to best a 701 opponent.

Statistics: Posted by errorblankfield — 14 Mar 2014, 06:30


]]>
2014-03-14T04:07:46+02:00 2014-03-14T04:07:46+02:00 /viewtopic.php?t=6249&p=68911#p68911 <![CDATA[Re: 1v1 Ladder Ranking]]>
If you played 2000 games and are rated 700 then you are rated 700 + /- say 1% more games means that the certainty that you are worth a 700 rank is higher... it doesn't automatically make you better.

I would be more worried about playing someone ranked 1000 with 50 games than someone ranked 800 with 1000 games... that 800 rank player is a pretty solid 800 that 1000 managed to get to 1000 in just 50 games... and will probably end up much higher once they have more experience...

Statistics: Posted by SomeoneAUS — 14 Mar 2014, 04:07


]]>
2014-03-14T01:01:35+02:00 2014-03-14T01:01:35+02:00 /viewtopic.php?t=6249&p=68903#p68903 <![CDATA[Re: 1v1 Ladder Ranking]]> Statistics: Posted by prodromos — 14 Mar 2014, 01:01


]]>
2014-03-13T19:58:23+02:00 2014-03-13T19:58:23+02:00 /viewtopic.php?t=6249&p=68884#p68884 <![CDATA[Re: 1v1 Ladder Ranking]]> Statistics: Posted by Gorton — 13 Mar 2014, 19:58


]]>
2014-03-13T19:49:26+02:00 2014-03-13T19:49:26+02:00 /viewtopic.php?t=6249&p=68883#p68883 <![CDATA[Re: 1v1 Ladder Ranking]]>
Just saw another thread (now locked) where ZePilot says ratings will be removed in the next update.

What does this mean specifically? When we join a game lobby, there won't be a number indicating a player's possible skill? Does it mean that the ladder will operate differently?

I sure don't understand why ratings are such a problem and such an area of contention. I guess I don't see the scale and damage that some are causing by either smurfing or trying to lower their rating. Maybe I'm assuming a higher level of maturity than our player base is exhibiting as a whole. I'd love to get some background on this problem.

Was this decision reached by Zep only, or is there some group that is involved with decisions like these? I ask because it seems like Zep is getting angry (my inference, definitely not stating this as a fact). I'd hate to see Zep get burned out leave and/or just shut this whole thing down. Is that even possible? Having to rely on steam for matchmaking would definitely suck.

Statistics: Posted by Flipper — 13 Mar 2014, 19:49


]]>
2014-03-13T16:16:09+02:00 2014-03-13T16:16:09+02:00 /viewtopic.php?t=6249&p=68852#p68852 <![CDATA[Re: 1v1 Ladder Ranking]]> Statistics: Posted by Gorton — 13 Mar 2014, 16:16


]]>
2014-03-13T16:08:45+02:00 2014-03-13T16:08:45+02:00 /viewtopic.php?t=6249&p=68850#p68850 <![CDATA[Re: 1v1 Ladder Ranking]]> Statistics: Posted by Flipper — 13 Mar 2014, 16:08


]]>
2014-03-10T14:00:17+02:00 2014-03-10T14:00:17+02:00 /viewtopic.php?t=6249&p=68475#p68475 <![CDATA[Re: 1v1 Ladder Ranking]]> Statistics: Posted by Vee — 10 Mar 2014, 14:00


]]>
2014-03-10T12:16:43+02:00 2014-03-10T12:16:43+02:00 /viewtopic.php?t=6249&p=68473#p68473 <![CDATA[Re: 1v1 Ladder Ranking]]>
E8400-CV wrote:
Sometimes people have a very good day and play seemingly above their rating
Sometimes people are a bit too intoxicated and play below their rating
And then some people smurf their way to a certain rating... like IKatherine, so they can gang up on lower rated players.


Sometimes?

Statistics: Posted by Cuddles — 10 Mar 2014, 12:16


]]>
2014-03-09T15:59:56+02:00 2014-03-09T15:59:56+02:00 /viewtopic.php?t=6249&p=68379#p68379 <![CDATA[Re: 1v1 Ladder Ranking]]> Ladder ranking/ladder position is your position in a ladder division (Snoop, Striker, etc.). That position is determined as described in above mentioned link.
1v1 rating is your Trueskill rating based on matchmaker 1v1 games ("ladder games"). This is the rating that you see in-game when you play an automatch game. 1v1 rating is updated after every game.
There is a list of all players ordered by rating. Your "rank" is your position in the ordered list.
Global rating is the Trueskill rating based on all "ranked" matches, which are all custom games (1v1, 4v4, etc.) without restrictions and mods, + as far as I know all matchmaker games ("ladder games"). Global rating is rounded and updated weekly.
Global "ranking" is the position in the ordered global rating list, before rounding.

I think that the matchmaker matches (combines people to play eachother) based on 1v1 ("ladder") rating, not on ladder division.

Statistics: Posted by ax0lotl — 09 Mar 2014, 15:59


]]>
2014-03-08T05:13:43+02:00 2014-03-08T05:13:43+02:00 /viewtopic.php?t=6249&p=68278#p68278 <![CDATA[Re: 1v1 Ladder Ranking]]>
Also the ladder wiki page is out of date. The new ladder works like this http://www.faforever.com/?p=1262

Statistics: Posted by Deering — 08 Mar 2014, 05:13


]]>