|
On February 23 2013 00:13 Jealous wrote:
I don't think Birdie said he wants to use ICCup Elo for DR rankings... And I don't see why that would be logical, given that DR is a self-contained pool of players like progamer-license holding players were in KeSPA and therefore in TL's Elo calculator.
Maybe I'm misunderstanding, but he did say he wants to "include every single iCCup game" with this new ranking, which to me implies hes going to be including Iccup stats.
|
We just need some people to retire from D Ranks like me so we can populate C ranks and actually get something going there 
|
On February 23 2013 01:30 Artanis[Xp] wrote:We just need some people to retire from D Ranks like me so we can populate C ranks and actually get something going there
i would like to...just my opponents dont let me win :/
|
On February 23 2013 01:30 Artanis[Xp] wrote:We just need some people to retire from D Ranks like me so we can populate C ranks and actually get something going there
Or you could realize that anything C above is mostly dead and you might as well just go play Fish or something. 
|
On February 23 2013 01:30 Artanis[Xp] wrote:We just need some people to retire from D Ranks like me so we can populate C ranks and actually get something going there
That would require some people in D ranks to be capable of winning games at C- level, of which there is little evidence so far (otherwise the moaners would have dug it up by now).
|
On February 23 2013 01:39 Zergneedsfood wrote:Show nested quote +On February 23 2013 01:30 Artanis[Xp] wrote:We just need some people to retire from D Ranks like me so we can populate C ranks and actually get something going there Or you could realize that anything C above is mostly dead and you might as well just go play Fish or something.
Pfft, I'm sure there'd be enough players to fill a league if people just signed up
On February 23 2013 02:29 Jealous wrote:Show nested quote +On February 23 2013 01:30 Artanis[Xp] wrote:We just need some people to retire from D Ranks like me so we can populate C ranks and actually get something going there That would require some people in D ranks to be capable of winning games at C- level, of which there is little evidence so far (otherwise the moaners would have dug it up by now).
I think there's a few that could hold their own in C- in here. I'm not going to point fingers though, it's up to everyone to make their own decision.
|
On February 22 2013 23:39 Sentenal wrote:Show nested quote +On February 22 2013 23:27 t0ssboy wrote:On February 22 2013 23:25 Sentenal wrote:On February 22 2013 18:18 Birdie wrote:And on the topic of stats/iCCup rankings... Birdie, update your ELO rankings... :D It's not a matter of updating it; to include every single iCCup game I have to recode an entire website spidering system, parse the data into a database, process the data using the Glicko2 ranking system, then format the data. The last two steps are the simple bits, the first two are the bits I haven't done before and am having a bit of a learning curve as I learn to code it. Plus procrastination and being busy with this that and the other. Not to mention whoever coded the iCCup website needs rocks thrown at them. Repeatedly.EDIT I take that back, it's not so bad. I found a nice library that lets me parse the data easily. The only problem I have now is that each page takes a few seconds to download, probably because the iccup server is in russia and I'm in New Zealand. So doing 200000 or however many games will take several days, if not weeks. That's not really feasible, so I'm going to see if I can get the iccup games database off them... would make things a lot easier. Woah woah woah, the fabled New Birdie ELO for D-Rank Leagues is going to use Iccup stats now? What the fuck? Why? You are essentially wanting to include practice games into an ELO ranking....? Why don't you like it,i thinks it's a good idea to get to know the player in general rather than only the games he played in D Leagues.Besides this way we can see how much you practice your 9/9's Practice games don't matter in an ELO system, to think otherwise is to be completely delusional. Practice games are just that, practice. I'll give you a great example why practice games don't matter.. Lets take the best player in these leagues, Prophecy_. On Birdie's first ELO ranking, he was ranked #3. Then he went 6-1 in DRTL3. Then went 10-1 in DRATL. And is now set to play in the finals of DRIT5 after having to go through a Death Bracket. Hes clearly one of the top players in the league. Go look at his ladder stats and tell me they represent his skill. Unless you think a 0-3 D- player is the best player in the league.
All 3 of his losses are disconnects btw
|
On February 23 2013 04:31 arb wrote:Show nested quote +On February 22 2013 23:39 Sentenal wrote:On February 22 2013 23:27 t0ssboy wrote:On February 22 2013 23:25 Sentenal wrote:On February 22 2013 18:18 Birdie wrote:And on the topic of stats/iCCup rankings... Birdie, update your ELO rankings... :D It's not a matter of updating it; to include every single iCCup game I have to recode an entire website spidering system, parse the data into a database, process the data using the Glicko2 ranking system, then format the data. The last two steps are the simple bits, the first two are the bits I haven't done before and am having a bit of a learning curve as I learn to code it. Plus procrastination and being busy with this that and the other. Not to mention whoever coded the iCCup website needs rocks thrown at them. Repeatedly.EDIT I take that back, it's not so bad. I found a nice library that lets me parse the data easily. The only problem I have now is that each page takes a few seconds to download, probably because the iccup server is in russia and I'm in New Zealand. So doing 200000 or however many games will take several days, if not weeks. That's not really feasible, so I'm going to see if I can get the iccup games database off them... would make things a lot easier. Woah woah woah, the fabled New Birdie ELO for D-Rank Leagues is going to use Iccup stats now? What the fuck? Why? You are essentially wanting to include practice games into an ELO ranking....? Why don't you like it,i thinks it's a good idea to get to know the player in general rather than only the games he played in D Leagues.Besides this way we can see how much you practice your 9/9's Practice games don't matter in an ELO system, to think otherwise is to be completely delusional. Practice games are just that, practice. I'll give you a great example why practice games don't matter.. Lets take the best player in these leagues, Prophecy_. On Birdie's first ELO ranking, he was ranked #3. Then he went 6-1 in DRTL3. Then went 10-1 in DRATL. And is now set to play in the finals of DRIT5 after having to go through a Death Bracket. Hes clearly one of the top players in the league. Go look at his ladder stats and tell me they represent his skill. Unless you think a 0-3 D- player is the best player in the league. All 3 of his losses are disconnects btw
Thats because his Starcraft crashes every time a game ends, forcing him to have to relog. Hes one of the most manner players in this league, so I dunno if you are trying to say hes a discer or something lol.
|
On February 23 2013 04:36 Sentenal wrote:Show nested quote +On February 23 2013 04:31 arb wrote:On February 22 2013 23:39 Sentenal wrote:On February 22 2013 23:27 t0ssboy wrote:On February 22 2013 23:25 Sentenal wrote:On February 22 2013 18:18 Birdie wrote:And on the topic of stats/iCCup rankings... Birdie, update your ELO rankings... :D It's not a matter of updating it; to include every single iCCup game I have to recode an entire website spidering system, parse the data into a database, process the data using the Glicko2 ranking system, then format the data. The last two steps are the simple bits, the first two are the bits I haven't done before and am having a bit of a learning curve as I learn to code it. Plus procrastination and being busy with this that and the other. Not to mention whoever coded the iCCup website needs rocks thrown at them. Repeatedly.EDIT I take that back, it's not so bad. I found a nice library that lets me parse the data easily. The only problem I have now is that each page takes a few seconds to download, probably because the iccup server is in russia and I'm in New Zealand. So doing 200000 or however many games will take several days, if not weeks. That's not really feasible, so I'm going to see if I can get the iccup games database off them... would make things a lot easier. Woah woah woah, the fabled New Birdie ELO for D-Rank Leagues is going to use Iccup stats now? What the fuck? Why? You are essentially wanting to include practice games into an ELO ranking....? Why don't you like it,i thinks it's a good idea to get to know the player in general rather than only the games he played in D Leagues.Besides this way we can see how much you practice your 9/9's Practice games don't matter in an ELO system, to think otherwise is to be completely delusional. Practice games are just that, practice. I'll give you a great example why practice games don't matter.. Lets take the best player in these leagues, Prophecy_. On Birdie's first ELO ranking, he was ranked #3. Then he went 6-1 in DRTL3. Then went 10-1 in DRATL. And is now set to play in the finals of DRIT5 after having to go through a Death Bracket. Hes clearly one of the top players in the league. Go look at his ladder stats and tell me they represent his skill. Unless you think a 0-3 D- player is the best player in the league. All 3 of his losses are disconnects btw Thats because his Starcraft crashes every time a game ends, forcing him to have to relog. Hes one of the most manner players in this league, so I dunno if you are trying to say hes a discer or something lol.
no i'm saying it looks like he lost 3 games because his internet crashed
tbh
|
Every single D Ranks tournament and league, there's an argument about which players are allowed in based on skill and current ladder ranking and so on. This is futile because of the woeful inaccuracy of the iCCup ladder ranking system. If I reformat the iCCup ladder 1v1 games (not every iCCup game, just ladder 1v1s on official iCCup maps) then every single player will be ranked according to skill. Then we can decide what points equal D ranks (on the new ranking I'm doing, it might be, say, 0-1500 points = E to D+ ranks, something like that) and then there can be no arguing about players being too good for the league.
The only inaccuracy of the system is when there are players who don't play ladder games, e.g. sGs.shoyo (not that he plays in D ranks events).
There's nothing wrong with having a separate D Rank Event ELO ranking but I don't see the use, as it only compares the players in the small D Ranks Event playerbase, when the arguments are caused by comparisons to overall iCCup player skill.
|
Rank: D/D+
ICCUP account: awerti[BSG]
Race: protoss
|
On February 23 2013 05:04 Birdie wrote:
Every single D Ranks tournament and league, there's an argument about which players are allowed in based on skill and current ladder ranking and so on. This is futile because of the woeful inaccuracy of the iCCup ladder ranking system. If I reformat the iCCup ladder 1v1 games (not every iCCup game, just ladder 1v1s on official iCCup maps) then every single player will be ranked according to skill. Then we can decide what points equal D ranks (on the new ranking I'm doing, it might be, say, 0-1500 points = E to D+ ranks, something like that) and then there can be no arguing about players being too good for the league.
The only inaccuracy of the system is when there are players who don't play ladder games, e.g. sGs.shoyo (not that he plays in D ranks events).
There's nothing wrong with having a separate D Rank Event ELO ranking but I don't see the use, as it only compares the players in the small D Ranks Event playerbase, when the arguments are caused by comparisons to overall iCCup player skill.
I see. But even doing that, there alot of holes/problems. What about smurfs (as in, are you going to track down everyone's alternate accounts)? What about people who /clearstats? What about people who don't ladder (Prophecy is a good example of this, since he does like 2-3 ladder games per season, tops)? And even then, it won't remove people arguing about who is too good or not to be in the league, due to a number of reasons (people not understanding the system, people not agreeing with the system, etc etc).
I was thinking it was going to be just a D Rank Event ELO thing, since people were mentioning it when talk of that All-Star League thing came up. But it looks like it wasn't your intention to do that, which is fine. I think what you are intending to do would require alot of work to solve all possible holes/inaccuracies that might come up.
|
What about smurfs (as in, are you going to track down everyone's alternate accounts)? What about people who /clearstats? What about people who don't ladder (Prophecy is a good example of this, since he does like 2-3 ladder games per season, tops)?
Players would have to also say what their max rank is on their other accounts, same as it is currently. It's hard to get rid of smurfing no matter what system is used, except for live LANs xD People who don't ladder I already mentioned as an inaccuracy, but they're already a problem anyway. /clearstats same sort of thing, nothing I can do about it. But for people who DO ladder, they'll be ranked much more accurately.
|
this is a problem that happens in any event with a max skill cap. there's not much you can do about it except have people play for fun instead of prize, and hope that they don't have fun by stomping people badder than they are.
|
Canada8036 Posts
On February 23 2013 05:04 Birdie wrote:
Every single D Ranks tournament and league, there's an argument about which players are allowed in based on skill and current ladder ranking and so on. This is futile because of the woeful inaccuracy of the iCCup ladder ranking system. If I reformat the iCCup ladder 1v1 games (not every iCCup game, just ladder 1v1s on official iCCup maps) then every single player will be ranked according to skill. Then we can decide what points equal D ranks (on the new ranking I'm doing, it might be, say, 0-1500 points = E to D+ ranks, something like that) and then there can be no arguing about players being too good for the league.
The only inaccuracy of the system is when there are players who don't play ladder games, e.g. sGs.shoyo (not that he plays in D ranks events).
There's nothing wrong with having a separate D Rank Event ELO ranking but I don't see the use, as it only compares the players in the small D Ranks Event playerbase, when the arguments are caused by comparisons to overall iCCup player skill.
As far as I know, game lists aren't saved between seasons, so you're going to be missing a ton of information. The accuracy of your numbers will also be kinda questionable due to /clearstats fucking everything up. It's equivalent to removing information from the system, which isn't exactly ideal for ELO rankings.
Edit: I should clarify, if this ranking is going to be used to determine player legitimacy in the league, you reaaaally want your numbers to be accurate.
|
On February 23 2013 06:25 Spazer wrote:Show nested quote +On February 23 2013 05:04 Birdie wrote:
Every single D Ranks tournament and league, there's an argument about which players are allowed in based on skill and current ladder ranking and so on. This is futile because of the woeful inaccuracy of the iCCup ladder ranking system. If I reformat the iCCup ladder 1v1 games (not every iCCup game, just ladder 1v1s on official iCCup maps) then every single player will be ranked according to skill. Then we can decide what points equal D ranks (on the new ranking I'm doing, it might be, say, 0-1500 points = E to D+ ranks, something like that) and then there can be no arguing about players being too good for the league.
The only inaccuracy of the system is when there are players who don't play ladder games, e.g. sGs.shoyo (not that he plays in D ranks events).
There's nothing wrong with having a separate D Rank Event ELO ranking but I don't see the use, as it only compares the players in the small D Ranks Event playerbase, when the arguments are caused by comparisons to overall iCCup player skill. As far as I know, game lists aren't saved between seasons, so you're going to be missing a ton of information. The accuracy of your numbers will also be kinda questionable due to /clearstats fucking everything up. It's equivalent to removing information from the system, which isn't exactly ideal for ELO rankings.
My ranking will store info between seasons though so it's fine. Although I'm thinking I'll just reset every iccup season anyway, so I don't have to worry about ranking deterioration. Clearstats won't actually affect my system when I think about it; clearstats only clears iCCup ranking, not my ranking, so unless someone clearstats right before I grab the iCCup info, it won't matter.
|
It seems that the issue is that a lot of people won't play enough games to get a decent ranking (in terms of accuracy), or will avoid playing games specifically for that reason... I think if people really care about quality control to the level that they seem to (going off of the posts in all DR threads about rank and all that jazz) there could be a smaller, more dedicated league where each team has say 4-5 players tops, and each player is required to play at least 15 ladder games in a season before signing up. We'd have maybe 4 teams, but at least then people couldn't complain, and it would be the most dedicated players so there should be few W/O, etc.
|
I suggest requiring players to designate their account and play a minimum number of games (with a minimum game length). There would be a quota per week or so. This would weed out inactives and make it harder to smurf, since it would be boring to play 30 or so long games and intentionally lose. If someone has a complaint, you can then review accounts and decide from there. For legit players who want to be in the league, this shouldn't be a big problem since they should be practicing on their account anyway.
|
On February 23 2013 09:10 Savant wrote:
I suggest requiring players to designate their account and play a minimum number of games (with a minimum game length). There would be a quota per week or so. This would weed out inactives and make it harder to smurf, since it would be boring to play 30 or so long games and intentionally lose. If someone has a complaint, you can then review accounts and decide from there. For legit players who want to be in the league, this shouldn't be a big problem since they should be practicing on their account anyway.
sounds like a pain in the ass and would drive away most people who only have time to play a few games a week
|
On February 23 2013 09:22 rauk wrote:Show nested quote +On February 23 2013 09:10 Savant wrote:
I suggest requiring players to designate their account and play a minimum number of games (with a minimum game length). There would be a quota per week or so. This would weed out inactives and make it harder to smurf, since it would be boring to play 30 or so long games and intentionally lose. If someone has a complaint, you can then review accounts and decide from there. For legit players who want to be in the league, this shouldn't be a big problem since they should be practicing on their account anyway. sounds like a pain in the ass and would drive away most people who only have time to play a few games a week
A weekly quota could be removed, and instead make this ICCup season the test season for the next DR event. That way they won't be mandated to play anything during the DR season and only need 30 games throughout this whole ICCup season to qualify. Like I said, it would weed people out but then at least no one can complain.
|
|
|
|
|
|
EPT
