|
On April 20 2013 15:48 MasterOfPuppets wrote:Show nested quote +On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ.
He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit.
First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another.
Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance.
I honestly don't see how my claims "are stupid", he replied to a post which was referencing win rates and he too was talking about win rates.
If you guys are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible.
But then to go as far as to say that my posts are "stupid" when you yourself seem to fail to understand the context of the discussion is actually laughable, truly laughable.
|
On April 20 2013 17:08 SlixSC wrote:Show nested quote +On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible.
I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB.
http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far.
And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"?
|
On April 20 2013 16:26 figq wrote:
Don't you dare question the mighty relevance of getting 3rd in WCS Denmark. The man is a legend.
While I do understand the comical aspect of this.
Bunny's points mainly stem from him smacking around the Danish scene and qualifying for WCS EU playing very strong players.
Our list rewards players who go on hot streaks with points. That is the basic premise of the system.
|
On April 20 2013 17:10 Grovbolle wrote:Show nested quote +On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"?
Zotac Finals are professional players only.
Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly
I think that's pro enough 
|
LucifroN certainly deserves his ranking.
|
On April 20 2013 17:26 SlixSC wrote:Show nested quote +On April 20 2013 17:10 Grovbolle wrote:On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"? Zotac Finals are professional players only. Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly I think that's pro enough
Our biggest "non-pro" is the DSCL qualifiers I think. But all players there have sub 1000 points.
Last month mainly consist of GSL, GSTL, Pro-League, WCS qualifiers, Acer Teamstory Cup, Gigabyte cups, Yegalisk Master something, Various Ro8/4's from G4SC2 cups etc.
|
On April 20 2013 17:26 SlixSC wrote:Show nested quote +On April 20 2013 17:10 Grovbolle wrote:On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"? Zotac Finals are professional players only.Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly I think that's pro enough
Should we disconsider DreamHack then, since they tend to have a few of amateur/semi-pro players in the early group stages?
Do you see how arbitrary this is? What if there are are "amateur" players who aren't on a proteam and don't practice 6 or 8 hrs a day but can still manage to beat a professional in some small daily/weekly cup? Should that not count? And then there are countless of players who *are* on teams, not big teams granted, but pro teams nonetheless, but they don't win tournaments. Should their matches not count either?
And then there's the fallacy of "I'm not familiar with this player therefore he's not a pro or simply terrible"... even when they've actually won stuff. ._.
|
On April 20 2013 17:39 MasterOfPuppets wrote:Show nested quote +On April 20 2013 17:26 SlixSC wrote:On April 20 2013 17:10 Grovbolle wrote:On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"? Zotac Finals are professional players only.Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly I think that's pro enough Should we disconsider DreamHack then, since they tend to have a few of amateur/semi-pro players in the early group stages? Do you see how arbitrary this is? What if there are are "amateur" players who aren't on a proteam and don't practice 6 or 8 hrs a day but can still manage to beat a professional in some small daily/weekly cup? Should that not count? And then there are countless of players who *are* on teams, not big teams granted, but pro teams nonetheless, but they don't win tournaments. Should their matches not count either? And then there's the fallacy of "I'm not familiar with this player therefore he's not a pro or simply terrible"... even when they've actually won stuff. ._.
Sure it's arbitrary. But this will always be the case. Ignoring diamond league players in your balance statistics isn't any less arbitrary than ignoring players you don't consider pro. You have to start at some point and if the early stages of dreamhack tend to have a significant amount of non-pro players then I think you would be justified in ignoring the first group stage for example.
On April 20 2013 17:31 Grovbolle wrote:Show nested quote +On April 20 2013 17:26 SlixSC wrote:On April 20 2013 17:10 Grovbolle wrote:On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"? Zotac Finals are professional players only. Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly I think that's pro enough Our biggest "non-pro" is the DSCL qualifiers I think. But all players there have sub 1000 points. Last month mainly consist of GSL, GSTL, Pro-League, WCS qualifiers, Acer Teamstory Cup, Gigabyte cups, Yegalisk Master something, Various Ro8/4's from G4SC2 cups etc.
Which is fair enough. I still think you should ignore showmatches though, for reasons I already mentioned.
|
On April 20 2013 17:50 SlixSC wrote:Show nested quote +On April 20 2013 17:39 MasterOfPuppets wrote:On April 20 2013 17:26 SlixSC wrote:On April 20 2013 17:10 Grovbolle wrote:On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"? Zotac Finals are professional players only.Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly I think that's pro enough Should we disconsider DreamHack then, since they tend to have a few of amateur/semi-pro players in the early group stages? Do you see how arbitrary this is? What if there are are "amateur" players who aren't on a proteam and don't practice 6 or 8 hrs a day but can still manage to beat a professional in some small daily/weekly cup? Should that not count? And then there are countless of players who *are* on teams, not big teams granted, but pro teams nonetheless, but they don't win tournaments. Should their matches not count either? And then there's the fallacy of "I'm not familiar with this player therefore he's not a pro or simply terrible"... even when they've actually won stuff. ._. Sure it's arbitrary. But this will always be the case. Ignoring diamond league players in your balance statistics isn't any less arbitrary than ignoring players you don't consider pro. You have to start at some point and if the early stages of dreamhack tend to have a significant amount of non-pro players then I think you would be justified in ignoring the first group stage for example. Show nested quote +On April 20 2013 17:31 Grovbolle wrote:On April 20 2013 17:26 SlixSC wrote:On April 20 2013 17:10 Grovbolle wrote:On April 20 2013 17:08 SlixSC wrote:On April 20 2013 15:48 MasterOfPuppets wrote:On April 20 2013 11:16 SlixSC wrote:You are including show matches and a significant amount of amateur games. Including games that have nothing to do with pro play is kind of defeating the whole purpose of a balance graph. Why not take it to the extreme and also include bronze matches and practise matches? You are including amateur matches and showmatches after all so it doesn't seem that far off. Wat. This isn't a balance graph. It's a system that attempts to predict the outcome of a match with some degree of accuracy, based upon a player rating, which is calculated from their match history. If you seriously think showmatches between pro players are not remotely representative of their skill level at that given point in time, then... I suggest you stop posting in this thread, you're doing nothing but making a fool out of yourself with stupid claim upon stupid claim none of which would happen if you had even read the information provided in the OP or in the site's FAQ. He was talking about win rates. So yes, he thinks including amateur matches and showmatches has statistical relevance when talking about win rates. Which is bullshit. First of all, show matches are completely different to competitive matches, players will more likely try to play longer games (it's a show match) which will inevitably skew the outcome of the game one way or another. Amateur matches are amateur matches, they tell you exactly bupkis about the current state of balance. I honestly don't see how my claims "are stupid", you can't make an argument and as soon as someone calls you out for it go "that's not what he/I actually meant, read the FAQ". If you are using language incorrectly or ambigously you can't just assume that people will interpret your posts in the most charitable way possible. I do agree. However showmatches with money on the line do hold relevance to most people. If you want "by event" balance stats (although they have very few matches) we do offer those on every event in our DB. http://aligulac.com/results/events/12652-WCS-2013-Season-1-Korea-Code-S/ This is Code S matches and balance so far. And the guy also includes stuff like Zotac Final as well, is that considered "Pro enough"? Zotac Finals are professional players only. Hyun, beastyqt, Arthur, Nerchio, revival, tarra, happy and bly I think that's pro enough Our biggest "non-pro" is the DSCL qualifiers I think. But all players there have sub 1000 points. Last month mainly consist of GSL, GSTL, Pro-League, WCS qualifiers, Acer Teamstory Cup, Gigabyte cups, Yegalisk Master something, Various Ro8/4's from G4SC2 cups etc. Which is fair enough. I still think you should ignore showmatches though, for reasons I already mentioned.
I disagree with your assesment of showmatches which are sponsored, thus having money on the line are, since playing for money is the bread and butter of most proplayers.
|
Our policy is to include rounds where we can be reasonably sure of the identities of most of the players involved, and to include players who we can be reasonably sure will not keep playing in a bubble (that is, they will either keep interacting with the greater pool, or they will play only a few matches and then vanish in obscurity). There's a fair amount of leeway still, of course.
For example, for the WCS EU qualifiers, the Ro64 tended to have maybe 70-80% well known players, so that's where we started adding.
A quick look indicates that showmatches make up 0.44% of the matches in the database (250). That seems like an unlikely source of error IMO.
|
On April 20 2013 07:10 Grovbolle wrote:Show nested quote +On April 20 2013 06:21 gondolin wrote:
From what I remember the system use something similar to true skill, so it is not "flawed", just it will take some time to converge, especially with the arrival of a lot of new players (the kespa player). What would make it converge a lot sooner is to use something like true skill through time, to incorporate futre results into the present ranking. So guessing the skill in the future? Obviously we can always use forward filthering backwards smoothing once games are played and make the older rankings better, but that's not really doing us any good with the newest ranking.
Well yes You could filter backwards to make the older rankings better and use these older rankings to improve the current rankings.
Imagine that two new players "Life" and "Flash" appears out of nowhere. Their first match is a BO9 between them where Flash win 5-4. The system can determine they are pretty close in skill, but since they are new players their rating does not change much. Then a week latter, Life goes on to win the GSL without dropping a single map, while Flash does not play at all. Then in the current system Life will have a lot of points but Flash rating will stay the same. So the problem is that the system determined that Life is really good *after* his BO9 with Flash. If they had played the BO9 after the GSL, the system would have determined that Flash also is very good.
But what you can do is now that you know that Life is very good, since he was probably also very good last week is to update the current rating of Flash. You can look at True Skill through time or Whole history rating to see how to incorporate past results retroactively into present ratings in a converging manner.
One argument against this system is that it would change past rating, but what you can do is only change these past ratings internally to have better current ratings. (One drawback of this system would still be that the rating of a player will change even if he does not play any match, but that's exactly the point!)
|
I have a few suggestions such as new balance stats, a RacevRace win% graph that only counts games between the top X% of pros, say 33%, and another that disconsiders games between players with too much rating discrepancy if the player with more rating wins.
|
As someone that watches a lot of SC2 the graph showing the ebb and flow of race balance since 2010 looks very accurate. Will be very interesting to see what the next plot looks like. Whether Z and P have started to work out Terran or whether Terran makes even more use of its new units
|
|
|
|
|
|
EPT
