EPTOn August 07 2012 14:28 goswser wrote:
HAHAHA Mana is higher than nestea.
HAHAHA Mana is higher than nestea.
Yeah, just had a quick look at them and some of these seem way off.
Introducing the SC2 World Ranking - Page 10
| Forum Index > SC2 General |
|
Firesilver
United Kingdom1190 Posts
Yeah, just had a quick look at them and some of these seem way off. | ||
|
ragz_gt
9172 Posts
Maybe for SC2, we can do something like winrate/elo/earning | ||
|
GohgamX
Canada1096 Posts
| ||
|
Markwerf
Netherlands3728 Posts
On August 08 2012 09:03 Cascade wrote: Nice to see this progress.  I would still prefer to keep 100% from a tournament until that tournament is played again, but this is also fine. Some comments on this vs ELO: 1) ELO tries to estimate the skill of players right now. This measures recent achievements. These two are quite different, and you cannot really say that one is better than the other, as they measure different things. Taeja is a good example. He tears up skilled people right and left in smaller tournaments, but has not really placed very high in many big individual tournaments. Thus ELO ranks him high, this ranking not as much. On the other hand, a player like MC really knows how to put himself together and win (or place high in) big events, but can drop games to lower ranked players in less important occasions. Thus MC gets a very high score for recent achievements, but maybe not as high in ELO. 2) Korea vs foreigners. The two communities are a bit separated, so it is hard to compare between the two. ELO solves this by not comparing at all. Thus two separate rankings. This point ranking tries to compare them, but runs into the problem of balancing the points of GSL vs other tournaments, and the problem that some players participate in more tournaments than others. Compare to tennis where all top players play in all big tournaments. None of the two solutions are very impressive. 3) The two systems have different flaws, so it is useful to look at both. ELO is very good because it has little subjective bias. A win is a win, no matter the context. Some bias in which games you include, but probably not a huge deal. This point system, as has been pointed out a lot of times, is very subjective in terms of how the points are distributed. The OP has done a good job of trying to find a distribution that most people can accept, but even with the full support of the community, you cannot get away from the fact that the points are distributed in a subjective manner. The "a win is a win" in ELO is also one of it's main problems. beating MC in the GSL finals is worth as much as beating him in the round of 8 in a weekly tournament. Many would argue that winning in the GSL finals is a much larger indicator of skill than beating them in a weekly tournament. The argument would be that people prepare for the GSL, and will always bring his very best game, while in a weekly, people may not play at their very best in every game, due to various circumstances. This is better taken into account in a point system, where big tournaments count more. So putting them together, ELO can be seen more as a "current raw skill" in some sense, while the point system is more "recent achievements". Also note that ELO tries to measure skill right now, while the point system is counting over the last 6 months, and will lag behind a bit. So a new player coming out of nowhere will jump up in ELO very fast, but will not reach his peak in the point system until after about half a year after his first win. Thus the point system measures consistency a bit as well. Best picture can be seen by looking at both. High ELO but low point ranking means either a rising top player, or a skilled player that is not very clutch in big tournaments. Low ELO but high point ranking is either a declining top player, or a good "tournament player". you really have no clue what you're talking about.. both measure the same thing, current skill of the player.. making an ELO or similar system (glicko etc.) for world ranking would work fairly well now since there are enough matches between foreigners and koreans. The match is a match thing is simply not true either, ELO can easily be configured to have different K-values for different types of matches/tournaments... Overall ELO is just FAR superior to the method given in this thread and combining results is not better.. it's just worse than using ELO alone. TLPD is just not the only way to make an ELO system, it could be vastly improved.. it is in the right direction though | ||
CosmicSpiral
United States15275 Posts
On August 09 2012 03:42 Markwerf wrote: you really have no clue what you're talking about.. both measure the same thing, current skill of the player.. making an ELO or similar system (glicko etc.) for world ranking would work fairly well now since there are enough matches between foreigners and koreans. The match is a match thing is simply not true either, ELO can easily be configured to have different K-values for different types of matches/tournaments... Overall ELO is just FAR superior to the method given in this thread and combining results is not better.. it's just worse than using ELO alone. TLPD is just not the only way to make an ELO system, it could be vastly improved.. it is in the right direction though Glicko is a good system. | ||
|
SpaceYeti
United States723 Posts
Mainly, weightings. The weightings between events is something that no one person can determine without many others disagreeing. For between tournament weights to be credible, these would have to be determined in the very least by a committee of trusted authorities who can properly estimate just how more difficult certain tournaments are over others, quantitatively. Simply weighting events from different leagues however is now good enough, because the same event in a series from the same league is likely harder at some times than at others (think early seasons of GSL or MLG to today). Therefore, every event would have to be weighted INDIVIDUALLY for the list to remain credible. To further complicate this same issue, the difficulty of a particular event is in large part determined by who else is competing in that same event. Therefore, the best way to determine the proper weights of finishing in different tournaments is going to have to rely on some metric of the skill of the players registered for that event. This is problematic since quantifying player skill is essentially the end goal of this exercise to begin with! Lastly, weights for finishing in different places within a given tournament have to be given a similar treatment to the two points above. First place in the GSL Finals isn't universally difficult between seasons, and the difference between 1st and 2nd, or 2nd and 3rd is not universally equal either. What's more, you could argue that within a given season, the weighting of getting 1st place may change depending on who is left in the tournament at any given point in time. If by some "fluke" of chance most of your best players get eliminated early, the suddenly getting 1st place isn't as difficult an achievement. I think these issues, each in their own way, would need to be addressed for a system of this nature to be a reliable indicator of true ranking and skill. At the end of the day, the truth of the matter is we all HATE the BCS system because it is convoluted and not obvious to the lay person watching games. The sort of measures that a system like this would require to be a reliable and valid system is plagued by that same fault. :\ | ||
|
Metalteeth
United States115 Posts
Inside the spoiler tag, you will see data on various upcoming tournaments, and ways I'm trying to weight them. + Show Spoiler + Ok, first one I'm taking a look at is TSL4. 32 players. Using the current up to date data. These are the player points of each player in the tournament: 5474.06 3670.64 3371.29 3230.43 3182.12 2909.78 2888.05 1502.80 1441.61 1378.23 1155.48 1081.93 873.42 842.69 818.58 738.44 702.41 626.61 581.28 573.48 572.08 426.83 424.55 345.60 334.56 319.40 305.17 247.32 243.04 133.55 113.42 78.82 Giving a mean of 1268.36, and a median of 720.42. Now, the first thing I thought of was just adding together the mean and median. This ensures that a single high ranked player won't overpower a tournament. The problem though is that this way means that a 4 person invite with the top 4 players in the world for a small prize pool would be worth more than a GSL or MLG could possibly be. So there has to be some sort of equalizing factor. So then I thought....prize money. Take TSL4. 1st place gets $15,000. So a possibly way is to average the prize pool (in US dollars) with the mean+median of a tournament, then divide that average by a number, lets try 5. For TSL:, that means that winning TSL4 would net a player 3397.75 points, with dwindling results further down (which would be dependent on the structure of the tournament). Ok, this seems decent, let's try another upcoming tournament. IEM has a tour stop soon, with a 1st place prize of $6,500. Many would consider this IEM stop a lower player pool, so with the same formula, what does winning IEM Cologne give? 1728.31. Ok, this seems like a decently weighted result. Try this on another group, the next MLG. Because Open play is a strange beast (and we don't know who will be there fully), I will only use those from pool play. Results: 5878.43 for winning MLS Summer. Wow, that is quite a lot. A GSL would be a lot more though, thanks to the larger prize pool for winning a GSL. What do you guys think about this? I'm thinking the equation needs work, but it's a decent start. To recap: Possible way to weight the WINNER of a future tournament: (Mean of players' ratings+Median of players' ratings+Prize pool in US Dollars)/5. Alright, trying another one. Let's imagine that GSL S4 was about to start, with the same players as GSL S3, along with the same prize pool. Winner of that tournament would get: 9581.27. Ok, that is a massive point hall. If that sticks, then we will get massive point inflation as tournaments run. I want to balance it such that points won't substantially increase with time to a stupid amount. A solution to this? Divide everything in half. Instead of dividing by 5, try 10. That makes TSL4 1698.88, IEM Cologne 864.155, MLG Summer 2939.22, and GSL 4790.64. Those are better numbers, much better. Hmm. Thinking about how this would work. A tournament with a large prize pool relative to the player quality (see some of the WCS Nationals without as much quality players, like New Zealand or Colombia) would be weighted too high, whereas a tournament with a good pool but little prize would underrate the tournament. However, this does provide a decent way of measuring the importance of the tournament. Plus, with the largest tournament (GSL) not super large, there shouldn't be point inflation, barring a massive influx of money into SC2. And if that happens, the formula can always be adjusted. | ||
|
Markwerf
Netherlands3728 Posts
On August 09 2012 12:20 Metalteeth wrote: So it seems that a lot of people think that the tournament weightings are subjective. Which is true. But people want a method of ranking the tournaments, not by a (seemingly) random weighting, but by measuring how strong the players there are. So I'm trying something for FUTURE tournaments. I will not be changing how past tournaments are weighted (part of the problem being how do you weigh players from before a tournament? TLPD is already updated, and if I start back rating tournaments, eventually I get no weight to determine the first tournament), only testing out rating future tournaments. I will definitely take ideas, but right now I'm testing out various ways of combining the mean and median of a player pool to determine a "weight" for the tournament. Inside the spoiler tag, you will see data on various upcoming tournaments, and ways I'm trying to weight them. + Show Spoiler + Ok, first one I'm taking a look at is TSL4. 32 players. Using the current up to date data. These are the player points of each player in the tournament: 5474.06 3670.64 3371.29 3230.43 3182.12 2909.78 2888.05 1502.80 1441.61 1378.23 1155.48 1081.93 873.42 842.69 818.58 738.44 702.41 626.61 581.28 573.48 572.08 426.83 424.55 345.60 334.56 319.40 305.17 247.32 243.04 133.55 113.42 78.82 Giving a mean of 1268.36, and a median of 720.42. Now, the first thing I thought of was just adding together the mean and median. This ensures that a single high ranked player won't overpower a tournament. The problem though is that this way means that a 4 person invite with the top 4 players in the world for a small prize pool would be worth more than a GSL or MLG could possibly be. So there has to be some sort of equalizing factor. So then I thought....prize money. Take TSL4. 1st place gets $15,000. So a possibly way is to average the prize pool (in US dollars) with the mean+median of a tournament, then divide that average by a number, lets try 5. For TSL:, that means that winning TSL4 would net a player 3397.75 points, with dwindling results further down (which would be dependent on the structure of the tournament). Ok, this seems decent, let's try another upcoming tournament. IEM has a tour stop soon, with a 1st place prize of $6,500. Many would consider this IEM stop a lower player pool, so with the same formula, what does winning IEM Cologne give? 1728.31. Ok, this seems like a decently weighted result. Try this on another group, the next MLG. Because Open play is a strange beast (and we don't know who will be there fully), I will only use those from pool play. Results: 5878.43 for winning MLS Summer. Wow, that is quite a lot. A GSL would be a lot more though, thanks to the larger prize pool for winning a GSL. What do you guys think about this? I'm thinking the equation needs work, but it's a decent start. To recap: Possible way to weight the WINNER of a future tournament: (Mean of players' ratings+Median of players' ratings+Prize pool in US Dollars)/5. Alright, trying another one. Let's imagine that GSL S4 was about to start, with the same players as GSL S3, along with the same prize pool. Winner of that tournament would get: 9581.27. Ok, that is a massive point hall. If that sticks, then we will get massive point inflation as tournaments run. I want to balance it such that points won't substantially increase with time to a stupid amount. A solution to this? Divide everything in half. Instead of dividing by 5, try 10. That makes TSL4 1698.88, IEM Cologne 864.155, MLG Summer 2939.22, and GSL 4790.64. Those are better numbers, much better. Hmm. Thinking about how this would work. A tournament with a large prize pool relative to the player quality (see some of the WCS Nationals without as much quality players, like New Zealand or Colombia) would be weighted too high, whereas a tournament with a good pool but little prize would underrate the tournament. However, this does provide a decent way of measuring the importance of the tournament. Plus, with the largest tournament (GSL) not super large, there shouldn't be point inflation, barring a massive influx of money into SC2. And if that happens, the formula can always be adjusted. ugh you're acting so stupid. You're basically trying to invent ELO again... You want tournaments that are automatically ranked by the quality of players that are attending?? Ie you want a player that beats difficult opponents on average to gain more points than beating weaker opponents? That is what ELO does... difference being that with ELO you can also lose points, which just happens to solve the problem of certain players (MC?) attending way more tournaments than others... Stop trying to reinvent the wheel, points systems suck for any sport which doesn't have rigourous rules like tennis.. Even in tennis ELO like systems are used at lower level.. only the ATP pro's are ranked by the point system.. | ||
|
Trasko
Sweden983 Posts
| ||
|
Penke
Sweden346 Posts
| ||
|
CosmicSpiral
United States15275 Posts
On August 09 2012 12:20 Metalteeth wrote: So it seems that a lot of people think that the tournament weightings are subjective. Which is true. But people want a method of ranking the tournaments, not by a (seemingly) random weighting, but by measuring how strong the players there are. So I'm trying something for FUTURE tournaments. I will not be changing how past tournaments are weighted (part of the problem being how do you weigh players from before a tournament? TLPD is already updated, and if I start back rating tournaments, eventually I get no weight to determine the first tournament), only testing out rating future tournaments. I will definitely take ideas, but right now I'm testing out various ways of combining the mean and median of a player pool to determine a "weight" for the tournament. Inside the spoiler tag, you will see data on various upcoming tournaments, and ways I'm trying to weight them. + Show Spoiler + Ok, first one I'm taking a look at is TSL4. 32 players. Using the current up to date data. These are the player points of each player in the tournament: 5474.06 3670.64 3371.29 3230.43 3182.12 2909.78 2888.05 1502.80 1441.61 1378.23 1155.48 1081.93 873.42 842.69 818.58 738.44 702.41 626.61 581.28 573.48 572.08 426.83 424.55 345.60 334.56 319.40 305.17 247.32 243.04 133.55 113.42 78.82 Giving a mean of 1268.36, and a median of 720.42. Now, the first thing I thought of was just adding together the mean and median. This ensures that a single high ranked player won't overpower a tournament. The problem though is that this way means that a 4 person invite with the top 4 players in the world for a small prize pool would be worth more than a GSL or MLG could possibly be. So there has to be some sort of equalizing factor. So then I thought....prize money. Take TSL4. 1st place gets $15,000. So a possibly way is to average the prize pool (in US dollars) with the mean+median of a tournament, then divide that average by a number, lets try 5. For TSL:, that means that winning TSL4 would net a player 3397.75 points, with dwindling results further down (which would be dependent on the structure of the tournament). Ok, this seems decent, let's try another upcoming tournament. IEM has a tour stop soon, with a 1st place prize of $6,500. Many would consider this IEM stop a lower player pool, so with the same formula, what does winning IEM Cologne give? 1728.31. Ok, this seems like a decently weighted result. Try this on another group, the next MLG. Because Open play is a strange beast (and we don't know who will be there fully), I will only use those from pool play. Results: 5878.43 for winning MLS Summer. Wow, that is quite a lot. A GSL would be a lot more though, thanks to the larger prize pool for winning a GSL. What do you guys think about this? I'm thinking the equation needs work, but it's a decent start. To recap: Possible way to weight the WINNER of a future tournament: (Mean of players' ratings+Median of players' ratings+Prize pool in US Dollars)/5. Alright, trying another one. Let's imagine that GSL S4 was about to start, with the same players as GSL S3, along with the same prize pool. Winner of that tournament would get: 9581.27. Ok, that is a massive point hall. If that sticks, then we will get massive point inflation as tournaments run. I want to balance it such that points won't substantially increase with time to a stupid amount. A solution to this? Divide everything in half. Instead of dividing by 5, try 10. That makes TSL4 1698.88, IEM Cologne 864.155, MLG Summer 2939.22, and GSL 4790.64. Those are better numbers, much better. Hmm. Thinking about how this would work. A tournament with a large prize pool relative to the player quality (see some of the WCS Nationals without as much quality players, like New Zealand or Colombia) would be weighted too high, whereas a tournament with a good pool but little prize would underrate the tournament. However, this does provide a decent way of measuring the importance of the tournament. Plus, with the largest tournament (GSL) not super large, there shouldn't be point inflation, barring a massive influx of money into SC2. And if that happens, the formula can always be adjusted. I understand you are trying to find some way in which the prestige of a tournament enhances/lessens the performance of a player within it. However, the prize pool of a tournament is independent of a player's performance within said tournament. I don't think you will find many people who believe IEM Cologne is half as prestigious as TSL4. Being live is important for a tournament's prestige. | ||
|
Metalteeth
United States115 Posts
Now, how the players ranked lower than first will work: Second will get 50% of the points (201.42). In a single elimination, semifinals would get half of that, but with a 3rd/4th placement because of a double elimination, 3rd will get 66.6% of 2nd, 4th will get 33.3% (134.28 and 67.14). 5th/6th will get 25% of 2nd (50.36), 7th/8th 50% of that (25.18), 9-12 hafl that (12.59), and 13-16 half that (6.3). WCS Oceania will depend on who qualifies. | ||
|
Metalteeth
United States115 Posts
By winning WCS Australia, PiG moves up to 124th on the rankings. mOOnGLaDe is very close, at 121st. 3rd place Mafia is down in 204th. With WCS Oceania next up, there is a slightly bigger points prize for winning. Compared to Australia's 402.83, Oceania is worth 648.14 points! Total points breakdown is in the spoiler: + Show Spoiler + Again, 1st place prize is not known, but estimated at 40% of the total prize pool of $15,000. 1st: 648.14 2nd: 324.07 3rd: 216.05 4th: 108.02 5/6: 81.02 7/8: 40.51 As you can see, the winner of WCS Oceania WILL break into the top 100 (current gateholder of the top 100 is Sound with 566.95). JazBas, the highest ranked played coming in and the New Zealand champion, will make the top 50 (50th exactly) should he win. mOOnGLaDe winning Oceania would have him flirting with the top 50, in fact he would slot in between Monster and Bomber for 53rd. Australia champion PiG would become 55th with a win, right behind Keen. The player with the biggest to gain is tgun. His performance of 5th/6th at Australia was his first entry in the rankings, and he is currently the lowest ranked player at Oceania, down at 381st. A win would move him all the way up to 73rd, ahead of Creator. Rankings of players at WCS Oceania: 105: JazBas 121: mOOnGLaDe 124: PiG 162: KnighT 204: MaFia 303: Light 340: YoonJY 381: tgun | ||
|
Darkhorse
United States23455 Posts
On August 08 2012 06:38 Metalteeth wrote: The points degrade over time. So a 1200 point second gets smaller over time. Oh okay that makes sense TY! | ||
|
Metalteeth
United States115 Posts
A problem I've been having is WCS South Korea. How do I rate the KeSPA players? Just giving them 0 isn't fair, but giving them too high an amount doesn't work either. So what I am going to do is just not give points for the KeSPA players, act like they aren't there in determining the rankings. Another tournament already going on (that I'll do the points for now) is the ongoing OSL. Similar to WCS SK, the problem is what do I do for the KeSPA players, so I'll just take the points from the non-KeSPA players. Another problem is the unknown prize pool, so I have the mean and median of the players (3069.73 mean and 1585.20 median). Prize pool is not known right now, but if you take a previous OSL for reference ($35362.24), that would make the OSL worth 4001.69 points, just a little less than a GSL! Full points breakdown for TSL4 and WCS South Korea, among other events, can be found below in the spoiler. + Show Spoiler + TSL4 1st: 1698.88 2nd: 849.44 SF: 424.72 QF: 212.36 Ro16: 106.18 Ro32: 53.09 WCS South Korea 1st: 1434.79 2nd: 717.40 3rd: 478.27 4th: 239.13 5/6: 179.35 7/8: 89.68 9-12: 44.84 13-16: 22.42 17-24: 11.21 25-32: 5.60 FXOpen Invitiational #6 1st: 407.71 2nd: 203.86 SF: 101.93 QF: 50.96 Ro16: 25.48 Ro32: 12.74 GS 3rd: 6.37 GS 4th: 3.19 | ||
|
ReaperX
Hong Kong1758 Posts
| ||
|
Metalteeth
United States115 Posts
Let's take a look at how the Oceanianc championship changed those players' rankings! 1st: mOOnGLaDe 121 -> 52 2nd: Mafia 204 -> 111 3rd: tgun 381 -> 166 4th: Light 302 -> 202 5/6: PiG 124 -> 117 5/6: JazBas 105 -> 93 7/8: YoonYJ 340 -> 275 7/8: KnighT 162 -> 147 Another tournament has been calculated: GIGABYTE NVIDIA Invitational #3 1st: 232.21 2nd: 116.11 3rd: 77.41 4th: 38.70 QF: 29.03 | ||
|
Metalteeth
United States115 Posts
Also, with all players announced (assuming the two TBA in the each open bracket as just byes for the top koreans), here are the point spreads for IEM Cologne! :D IEM Cologne: 1st: 893.17 2nd: 446.59 3rd: 297.73 4th: 148.86 QF: 111.65 Ro12: 55.83 GS 4th: 27.92 GS 5th: 13.96 GS 6th: 6.98 Open Bracket 3rd: 3.49 Open Bracket 4th: 1.75 Open Bracket 5-6: 0.88 Open Bracket 7-8: 0.44 Yeah, the open bracket spots are REALLY not worth much, but considering how the tournament is structured, being in the open is the step below groups, so whatever. 893 points for winning is a decent amount, about half of winning TSL4. The main difference is the prize pool, but this makes sense, since this is just a tour stop for IEM. GL HF to all participants! As for the actual tournament: let's talk some stats! -The lowest ranked players is the French Terran AureS, way down in 649th place. Even getting last place would move him up to 640th. The lower end of the rankings are VERY close. Last place (Retro at 768) has 0 points, 100 places up at 669 (Mkengyn) is only at 4.91. -The highest ranked player is current #1, MC! MC dropped out of WCS South Korea to compete in this event, which originally may have meant giving up his #1 to close #2 DRG if he made a run, but with his surprising loss to EffOrt, MC will retain his top position regardless of what happens this weekend. - The competition for top player from Oceanian heats up, with both #117, the Australian champion PiG (who crashed out of WCS Oceanian), and Australia runner up and Oceanian winner #52 MOOnGLaDe. mOOnGLaDe is 576.24 points ahead of PiG. The only way PiG catches up to pass mOOnGLaDe for top Oceanian player is with a win and mOOnGLaDe getting anything worse than 2nd. A more likely way for PiG to move up is by preventing a last place finish in the groups, that would move PiG ahead of Oceanian #2 (in both rank and WCS finish) MaFia. -NaNiwa can break into the Top 10 (which is where many people think he should be, but only "good" results at GSL and not much in the foreigner tournaments is hurting his ranking) with a win here. With only a PvT and a PvP to get out of the Open Bracket pretty easily, but each group has at least 2 Zergs, so it will be tough for him to win. | ||
|
Cele
Germany4016 Posts
 | ||
|
Metalteeth
United States115 Posts
2011 Chinese Battle.Net Invitational Removed 2012 IEM Cologne Added Some Major changes: Nerchio +2: With his 2nd place, Nerchio goes from 12th to 10th, meaning there are now 2 foreigners in the top 10! VortiX +41: A run through to the Semis means that VortiX jumps from 142nd to 101st, he is now a mere .1 point from the top 100 (and he should get that soon due to other player's points dropping slowly). Bomber +3 sLivko +3 SuperNoVa +2 Kas +2 SortOf +5 ForGG +2 Minigun +11 PiG +3 | ||
| ||
