|
On March 28 2012 00:05 Big J wrote:Show nested quote +On March 27 2012 23:48 Ghanburighan wrote:On March 27 2012 23:37 Big J wrote:On March 27 2012 23:27 Ghanburighan wrote:On March 27 2012 23:23 Mehukannu wrote:On March 27 2012 23:13 Big J wrote:On March 27 2012 22:44 FirstGear wrote:On March 27 2012 22:12 Ghanburighan wrote:
1) Lower sc2 ranks numbers for terran in several leagues.
Counter: No change in overall, so there just seem to have been less terrans for a long time. Actually when the game was released Terran had the most players significantly and Zerg was woefully underrepresented. As I remember T had the most players for the first 3 seasons. The argument back then was that most people were familiar with terran due to the campaign and decided to start with them. Only around platinum (and somewhat gold and diamond) Terrans are decreasing faster than than their GM/Master/Silver/(Bronze, though I think we shouldn't put too much thought into this) colleagues or than Protoss generally. From this data alone, the question should somewhat be: where do all the zergs come from. Zergs could have come from either people switching races or/and more terran and protoss players would left the game which then would show as a rise in zerg players. It is most likely a combination of both of those that caused it to happen, unless there is something else that could be the cause of it. Well, there is a theory available but I don't know how to test it. Considering that beginners play T, many people that make it higher might switch from T to some other race. This would mean that if 10 T's get promoted, after the switch, only let's say 8 would remain in the higher league. This covers this part of the data. The problem is, why are they switching to Z instead of P? So there should be more to it than just this. In fact, the key here seems to be figuring out the reason why P numbers are stable. nope, they aren't stable. apart from this "platinum-crack" where Terrans seem to decrease rather hard, Protoss players decrease somewhat in the same speed as Terran players. They just always had the most players, so they are still the most played race at a lot of levels, but less played than ever before. Well, that's not entirely true. In Diamond, Protoss had one larger (1.2%) decrease between patches 1.2 and 1.3 (KA removal and Super-fungal), but besides that stayed stable for many patches (<0.5% change). The overall 2.5% change is way less than the 5% change from T. This might suggest a correlation between nerfs and race-switching. correct me if I'm wrong, but those are the numbers I got from 1.3.0 --> 1.4.2 for Terran and Protoss (globally) ProtossMaster: 32,72 --> 33,52 = +0,8 Diamond: 33,36 --> 31,92 = -1,44 Platinum: 34,05 --> 32,06 = -1,99 Gold: 35,16 --> 32,60 = -2,46 Silver: 35,62 --> 32,89 = -2,73 TerranMaster: 30,86 --> 29,60 = -1,26 Diamond: 28,10 --> 26,44 = -1,66 Platinum: 29,24 --> 25,45 = -3,79 Gold: 31,18 --> 27,22 = -3,96 Silver: 33,59 --> 30,96 = -2,63 so for me this looks somewhat like I said; Silver, Diamond are same decrease; Gold, Platinum the Terran decrease is harder; Master (and GM), Protoss is slightly increasing, Terran decreasing. Still Protoss is decreasing as well apart from the highest level.
Where have all the protoss gone? ㅜ.ㅜ
|
On March 28 2012 02:09 Ghanburighan wrote:Show nested quote +On March 28 2012 01:48 Plansix wrote:On March 28 2012 00:38 Ghanburighan wrote:On March 28 2012 00:05 Big J wrote:On March 27 2012 23:48 Ghanburighan wrote:On March 27 2012 23:37 Big J wrote:On March 27 2012 23:27 Ghanburighan wrote:On March 27 2012 23:23 Mehukannu wrote:On March 27 2012 23:13 Big J wrote:On March 27 2012 22:44 FirstGear wrote:
[quote]
Actually when the game was released Terran had the most players significantly and Zerg was woefully underrepresented. As I remember T had the most players for the first 3 seasons. The argument back then was that most people were familiar with terran due to the campaign and decided to start with them. Only around platinum (and somewhat gold and diamond) Terrans are decreasing faster than than their GM/Master/Silver/(Bronze, though I think we shouldn't put too much thought into this) colleagues or than Protoss generally. From this data alone, the question should somewhat be: where do all the zergs come from. Zergs could have come from either people switching races or/and more terran and protoss players would left the game which then would show as a rise in zerg players. It is most likely a combination of both of those that caused it to happen, unless there is something else that could be the cause of it. Well, there is a theory available but I don't know how to test it. Considering that beginners play T, many people that make it higher might switch from T to some other race. This would mean that if 10 T's get promoted, after the switch, only let's say 8 would remain in the higher league. This covers this part of the data. The problem is, why are they switching to Z instead of P? So there should be more to it than just this. In fact, the key here seems to be figuring out the reason why P numbers are stable. nope, they aren't stable. apart from this "platinum-crack" where Terrans seem to decrease rather hard, Protoss players decrease somewhat in the same speed as Terran players. They just always had the most players, so they are still the most played race at a lot of levels, but less played than ever before. Well, that's not entirely true. In Diamond, Protoss had one larger (1.2%) decrease between patches 1.2 and 1.3 (KA removal and Super-fungal), but besides that stayed stable for many patches (<0.5% change). The overall 2.5% change is way less than the 5% change from T. This might suggest a correlation between nerfs and race-switching. correct me if I'm wrong, but those are the numbers I got from 1.3.0 --> 1.4.2 for Terran and Protoss (globally) ProtossMaster: 32,72 --> 33,52 = +0,8 Diamond: 33,36 --> 31,92 = -1,44 Platinum: 34,05 --> 32,06 = -1,99 Gold: 35,16 --> 32,60 = -2,46 Silver: 35,62 --> 32,89 = -2,73 TerranMaster: 30,86 --> 29,60 = -1,26 Diamond: 28,10 --> 26,44 = -1,66 Platinum: 29,24 --> 25,45 = -3,79 Gold: 31,18 --> 27,22 = -3,96 Silver: 33,59 --> 30,96 = -2,63 so for me this looks somewhat like I said; Silver, Diamond are same decrease; Gold, Platinum the Terran decrease is harder; Master (and GM), Protoss is slightly increasing, Terran decreasing. Still Protoss is decreasing as well apart from the highest level. Yeah, there are differences in the global statistics. I guess it's a EU quirk that Protoss does not dip as much before 1.2-1.3 as it does in the rest of the world. Globally, from 1.0.3 to 1.4.2 P dips 3.7 in diamond and Terran 4.3 percent. But an interesting finding, there are also significantly less Random players in higher leagues. They seem to have fallen into Gold and Silver, though, as unlike P and T, their numbers increase in G and S between 1 and 1.4.2. Zerg got 10-13 percent increases at the expense of Random!, Protoss and Terran decreases in Gold to Diamond. With the highest increase being in Platinum. There, terrans seems to be the most displaced with 6.5% change.As for Master, protoss and zerg gain a little (0.3 and 2.8, respectively), terran and random lose (2 and 1.1, respectively). I stand behind the idea that this is the point where a casual terran player(aka, he who does not seek builds on TL) either quits or seeks out a more efficent way to play. When we get into the lower leagues, we find more and more of thesse folks who do not know build orders, what MLG is or how many SCV is the right amount to build. Or that going mass maruader is not a good idea. All three races have this point, after all, there is only so much someone can learn on their own. The higher leagues are a different story. I would expect there to be fewer randoms. But the increase and decreases are well within any standard margin of error. Most of them may have little to do with the game itself. People switch races for any number of reasons or drop out of masters because they just don't have time to keep up with the meta game. I personally play less this year than last year. You need to explain what you mean by margin of error. There is no measurement error, as far as I can see. And there is virtually no fluctuation, only a steady trend. How would you calculate a margin of error?
A margin of error for most standard pools is 3%-7%, which means that anything that within those amounts is considered to be "equal" under the goals of the poll. It means within any sample set there are reasons people do thing that are not accounted for in the poll.
In this case, I was saying that the amount of change is so small that is well below any margin of error used in a standard poll. Because the amount is so small, sometimes just around 1000 people world wide, it would be nearly impossible to find a trend beyond the fact that the amounts are going down. Trying to link it to a patch, gameplay, the release of a new game or even the fact that SC2 is 2 years old at this point would be nearly impossible.
Unless you had the phone numbers of those 1000 people. Then we would need a whole new thread.
|
On March 27 2012 21:21 karpo wrote:Show nested quote +On March 27 2012 21:03 Kakaru2 wrote:
The biggest flaw with the above theory is that you fail consistently to explain why:
1. only late game TvP is questioned?
2. GM guys (merz, beasty) shared the same feeling for late game TvP. Yet you always conveniently forget about this and try to spin it that is only a gold-platinum Terran thing.
Until you incorporate these 2 points into your reasoning we have every right to call this "theory" biased. And you consistently get protoss and zerg pros complaining about their matchups being unbalanced in either early, mid, or lategame. Terrans complain about lategame TvZ, zergs claim to still have trouble against T/P in the earlier stages and protoss complain about motherships being a requirement in lategame PvZ else they "straight out lose". Pros are knowledgeable about the matchups but i think they too have a hard time being objective in their analysis.
Hmm, what exactly is your argument? That all players, no matter the race or pro/amateur status are simply whining? And for that reason we shoouldn't listen to them?
I don't have a problem with your argument, my problem is then with Blizzard for not listening to you earlier. If Blizzard didn't do the last two patches then this thread wouldn't exist.
In my opinion it's all about justice, if Blizzard buffed two races in the last 2 patches and now the third is crying it's only fair that they buff them as well in the very next patch. And that is why I'm so curious when they post that 1.50 notes up.
|
On March 28 2012 02:20 Plansix wrote:Show nested quote +On March 28 2012 02:09 Ghanburighan wrote:On March 28 2012 01:48 Plansix wrote:On March 28 2012 00:38 Ghanburighan wrote:On March 28 2012 00:05 Big J wrote:On March 27 2012 23:48 Ghanburighan wrote:On March 27 2012 23:37 Big J wrote:On March 27 2012 23:27 Ghanburighan wrote:On March 27 2012 23:23 Mehukannu wrote:On March 27 2012 23:13 Big J wrote:
[quote]Only around platinum (and somewhat gold and diamond) Terrans are decreasing faster than than their GM/Master/Silver/(Bronze, though I think we shouldn't put too much thought into this) colleagues or than Protoss generally.
From this data alone, the question should somewhat be: where do all the zergs come from. Zergs could have come from either people switching races or/and more terran and protoss players would left the game which then would show as a rise in zerg players. It is most likely a combination of both of those that caused it to happen, unless there is something else that could be the cause of it. Well, there is a theory available but I don't know how to test it. Considering that beginners play T, many people that make it higher might switch from T to some other race. This would mean that if 10 T's get promoted, after the switch, only let's say 8 would remain in the higher league. This covers this part of the data. The problem is, why are they switching to Z instead of P? So there should be more to it than just this. In fact, the key here seems to be figuring out the reason why P numbers are stable. nope, they aren't stable. apart from this "platinum-crack" where Terrans seem to decrease rather hard, Protoss players decrease somewhat in the same speed as Terran players. They just always had the most players, so they are still the most played race at a lot of levels, but less played than ever before. Well, that's not entirely true. In Diamond, Protoss had one larger (1.2%) decrease between patches 1.2 and 1.3 (KA removal and Super-fungal), but besides that stayed stable for many patches (<0.5% change). The overall 2.5% change is way less than the 5% change from T. This might suggest a correlation between nerfs and race-switching. correct me if I'm wrong, but those are the numbers I got from 1.3.0 --> 1.4.2 for Terran and Protoss (globally) ProtossMaster: 32,72 --> 33,52 = +0,8 Diamond: 33,36 --> 31,92 = -1,44 Platinum: 34,05 --> 32,06 = -1,99 Gold: 35,16 --> 32,60 = -2,46 Silver: 35,62 --> 32,89 = -2,73 TerranMaster: 30,86 --> 29,60 = -1,26 Diamond: 28,10 --> 26,44 = -1,66 Platinum: 29,24 --> 25,45 = -3,79 Gold: 31,18 --> 27,22 = -3,96 Silver: 33,59 --> 30,96 = -2,63 so for me this looks somewhat like I said; Silver, Diamond are same decrease; Gold, Platinum the Terran decrease is harder; Master (and GM), Protoss is slightly increasing, Terran decreasing. Still Protoss is decreasing as well apart from the highest level. Yeah, there are differences in the global statistics. I guess it's a EU quirk that Protoss does not dip as much before 1.2-1.3 as it does in the rest of the world. Globally, from 1.0.3 to 1.4.2 P dips 3.7 in diamond and Terran 4.3 percent. But an interesting finding, there are also significantly less Random players in higher leagues. They seem to have fallen into Gold and Silver, though, as unlike P and T, their numbers increase in G and S between 1 and 1.4.2. Zerg got 10-13 percent increases at the expense of Random!, Protoss and Terran decreases in Gold to Diamond. With the highest increase being in Platinum. There, terrans seems to be the most displaced with 6.5% change.As for Master, protoss and zerg gain a little (0.3 and 2.8, respectively), terran and random lose (2 and 1.1, respectively). I stand behind the idea that this is the point where a casual terran player(aka, he who does not seek builds on TL) either quits or seeks out a more efficent way to play. When we get into the lower leagues, we find more and more of thesse folks who do not know build orders, what MLG is or how many SCV is the right amount to build. Or that going mass maruader is not a good idea. All three races have this point, after all, there is only so much someone can learn on their own. The higher leagues are a different story. I would expect there to be fewer randoms. But the increase and decreases are well within any standard margin of error. Most of them may have little to do with the game itself. People switch races for any number of reasons or drop out of masters because they just don't have time to keep up with the meta game. I personally play less this year than last year. You need to explain what you mean by margin of error. There is no measurement error, as far as I can see. And there is virtually no fluctuation, only a steady trend. How would you calculate a margin of error? A margin of error for most standard pools is 3%-7%, which means that anything that within those amounts is considered to be "equal" under the goals of the poll. It means within any sample set there are reasons people do thing that are not accounted for in the poll. In this case, I was saying that the amount of change is so small that is well below any margin of error used in a standard poll. Because the amount is so small, sometimes just around 1000 people world wide, it would be nearly impossible to find a trend beyond the fact that the amounts are going down. Trying to link it to a patch, gameplay, the release of a new game or even the fact that SC2 is 2 years old at this point would be nearly impossible. Unless you had the phone numbers of those 1000 people. Then we would need a whole new thread.
Well, the sample size is very large, tens of thousands of people per race per league. This means that any margin of error is near 1-2%, right? I don't see where you get 3-7%. Even with three, Terrans going down 5.3 percentage points from patch 1.2 - now is still significant as that's an over 20% change. So, I repeat, how do you calculate this margin of error?
|
On March 28 2012 02:30 Ghanburighan wrote:Show nested quote +On March 28 2012 02:20 Plansix wrote:On March 28 2012 02:09 Ghanburighan wrote:On March 28 2012 01:48 Plansix wrote:On March 28 2012 00:38 Ghanburighan wrote:On March 28 2012 00:05 Big J wrote:On March 27 2012 23:48 Ghanburighan wrote:On March 27 2012 23:37 Big J wrote:On March 27 2012 23:27 Ghanburighan wrote:On March 27 2012 23:23 Mehukannu wrote:
[quote]
Zergs could have come from either people switching races or/and more terran and protoss players would left the game which then would show as a rise in zerg players. It is most likely a combination of both of those that caused it to happen, unless there is something else that could be the cause of it. Well, there is a theory available but I don't know how to test it. Considering that beginners play T, many people that make it higher might switch from T to some other race. This would mean that if 10 T's get promoted, after the switch, only let's say 8 would remain in the higher league. This covers this part of the data. The problem is, why are they switching to Z instead of P? So there should be more to it than just this. In fact, the key here seems to be figuring out the reason why P numbers are stable. nope, they aren't stable. apart from this "platinum-crack" where Terrans seem to decrease rather hard, Protoss players decrease somewhat in the same speed as Terran players. They just always had the most players, so they are still the most played race at a lot of levels, but less played than ever before. Well, that's not entirely true. In Diamond, Protoss had one larger (1.2%) decrease between patches 1.2 and 1.3 (KA removal and Super-fungal), but besides that stayed stable for many patches (<0.5% change). The overall 2.5% change is way less than the 5% change from T. This might suggest a correlation between nerfs and race-switching. correct me if I'm wrong, but those are the numbers I got from 1.3.0 --> 1.4.2 for Terran and Protoss (globally) ProtossMaster: 32,72 --> 33,52 = +0,8 Diamond: 33,36 --> 31,92 = -1,44 Platinum: 34,05 --> 32,06 = -1,99 Gold: 35,16 --> 32,60 = -2,46 Silver: 35,62 --> 32,89 = -2,73 TerranMaster: 30,86 --> 29,60 = -1,26 Diamond: 28,10 --> 26,44 = -1,66 Platinum: 29,24 --> 25,45 = -3,79 Gold: 31,18 --> 27,22 = -3,96 Silver: 33,59 --> 30,96 = -2,63 so for me this looks somewhat like I said; Silver, Diamond are same decrease; Gold, Platinum the Terran decrease is harder; Master (and GM), Protoss is slightly increasing, Terran decreasing. Still Protoss is decreasing as well apart from the highest level. Yeah, there are differences in the global statistics. I guess it's a EU quirk that Protoss does not dip as much before 1.2-1.3 as it does in the rest of the world. Globally, from 1.0.3 to 1.4.2 P dips 3.7 in diamond and Terran 4.3 percent. But an interesting finding, there are also significantly less Random players in higher leagues. They seem to have fallen into Gold and Silver, though, as unlike P and T, their numbers increase in G and S between 1 and 1.4.2. Zerg got 10-13 percent increases at the expense of Random!, Protoss and Terran decreases in Gold to Diamond. With the highest increase being in Platinum. There, terrans seems to be the most displaced with 6.5% change.As for Master, protoss and zerg gain a little (0.3 and 2.8, respectively), terran and random lose (2 and 1.1, respectively). I stand behind the idea that this is the point where a casual terran player(aka, he who does not seek builds on TL) either quits or seeks out a more efficent way to play. When we get into the lower leagues, we find more and more of thesse folks who do not know build orders, what MLG is or how many SCV is the right amount to build. Or that going mass maruader is not a good idea. All three races have this point, after all, there is only so much someone can learn on their own. The higher leagues are a different story. I would expect there to be fewer randoms. But the increase and decreases are well within any standard margin of error. Most of them may have little to do with the game itself. People switch races for any number of reasons or drop out of masters because they just don't have time to keep up with the meta game. I personally play less this year than last year. You need to explain what you mean by margin of error. There is no measurement error, as far as I can see. And there is virtually no fluctuation, only a steady trend. How would you calculate a margin of error? A margin of error for most standard pools is 3%-7%, which means that anything that within those amounts is considered to be "equal" under the goals of the poll. It means within any sample set there are reasons people do thing that are not accounted for in the poll. In this case, I was saying that the amount of change is so small that is well below any margin of error used in a standard poll. Because the amount is so small, sometimes just around 1000 people world wide, it would be nearly impossible to find a trend beyond the fact that the amounts are going down. Trying to link it to a patch, gameplay, the release of a new game or even the fact that SC2 is 2 years old at this point would be nearly impossible. Unless you had the phone numbers of those 1000 people. Then we would need a whole new thread. Well, the sample size is very large, tens of thousands of people per race per league. This means that any margin of error is near 1-2%, right? I don't see where you get 3-7%. Even with three, Terrans going down 5.3 percentage points from patch 1.2 - now is still significant as that's an over 20% change. So, I repeat, how do you calculate this margin of error?
The link to the wiki I posted gives a pretty good run down of how they are calculate. I do not work in statistics and I am not even sure we could calculate the margin of error for this data set. We would need to know the sample size, the small relative to the population and random data I don't think we can get. There is a reason people are paid a lot of money to take accurate polls. Most margins of error are taken from polls of 1000 people and a good error rate is around 3%-7%, and those amounts are standard in most political and marketing statistics. I have never heard of a margin of error below 3%.
|
I believe his null hypothesis is that race changes are random but there is an equal chance for each race to switch to another. At least that's the only null hypothesis that makes sense.
Plansix is just overlooking a small detail:
This would only work if the sc2ranks numbers are a small subset of a much much much larger bnet population. I'm not entirely sure on that point, but I believe that sc2ranks makes database dumps and therefore catches everyone. Thus there is no margin of error, as we have not statistic. The numbers account for 100% of the population. We do not conclude from a small dataset to an infinite population and thus there is no margin of error calculation needed.
Simply speaking, if I want to divine the number of white balls in a bowl with white, red and black balls in it, I do not need to do statistics, if I can draw each and every ball and simply count.
I have the feeling that 'It's within the margin of error' is his way of discrediting the data and all conclusions drawn from it.
|
On March 11 2012 11:24 Faust852 wrote:
TvP is too hard for a lot of Terran, so they stop playing because they don't want to lose and get demoted.
This literally the reason why I switched to P, I was stuck at High masters for almost 3 months and I was just done with the silly matchup because im clearly no pro. Now I play P and am r2 in my league. T is way to hard right now for under skilled players.
|
Since we are mainly talking about foreign Terrans, I don't think that looking at global race distributions is all that helpful. We all know Korean Terrans are doing fine. Aggregating with Korean stats masks the trend occuring in Europe and the Americas. Let's look at Europe and America from 1.2 to 1.4.2 then
America
Terran
Masters: 30.2 --> 28.8 = -1.4
Diamond: 28.0 --> 24.9 = -3.1
Platinum: 30.2 --> 24.3 = -5.9
Gold: 30.9 --> 26.8 = -4.1
Silver: 31.6 --> 30.2 = -1.4
Europe
Terran
Masters: 29.8 --> 28.3 = -1.5
Diamond: 27.7 --> 25.8 = -1.9
Platinum: 30.4 --> 24.9 = -5.5
Gold: 32.3 --> 26.7 = -5.6
Silver: 34.0 --> 30.9 = -3.1
So we see here almost a 6% drop in both servers for Terrans at the platinum level between 1.2 and 1.4.2. This is huge amount of players switching. In America and Europe, they are the most underrepresented race in all leagues except silver and bronze. I would argue that a sub-25% race representation in leagues is a problem, much like how Zerg was in the early days.
Back in 1.2, Zerg populations were struggling in all leagues, but a combination of buffs, learning the race, and metagame shifts have catapulted Zerg populations in 1.4.2 to be most represented in the American and Europe servers, Platinum and up to and including Masters.
I'm not suggesting any imbalance at the highest levels of play. Watching MKP play Terran is mind boggling. Who needs tanks against zerg when you have bio + godly micro + close position spawning on Metalopolis. But it makes sense that a small population of the Starcraft community would stick with a race that they have the most success with. Others will be propelled to other races by nerfs and buffs, regardless of whether it actually affects how they play the game. Still others will switch races due to perceptions of race difficulty, real or imagined. Finally, some will just quit altogether out of frustration.
The point is, Terrans in the Platinum to Diamond are struggling and due to any of the factors listed above, have decided to switch races. This becomes a problem for the health of the game when the XvT matchup becomes less and less frequent. 1/3 of the variety of the game is taken out at certain levels of play with such large race population disparities.
So in response to "Where did all the Terrans go [in the mid-high levels]?" In the Americas and Europe, they switched or quit. This is not good for the game and willl probably not be addressed until HoTS. In the meantime, as a newly promoted Platinum Terran, I'm getting my butt kicked a little bit. It is a bit disheartening, but I'm still learning a lot and make too many mistakes to name. I can improve everywhere. However, real or imagined, my perception of the skill discrepency at this level is affecting my play. Can't wait for the tanky unit, the battle-hellion.
|
On March 28 2012 02:53 Thrombozyt wrote:
I believe his null hypothesis is that race changes are random but there is an equal chance for each race to switch to another. At least that's the only null hypothesis that makes sense.
Plansix is just overlooking a small detail:
This would only work if the sc2ranks numbers are a small subset of a much much much larger bnet population. I'm not entirely sure on that point, but I believe that sc2ranks makes database dumps and therefore catches everyone. Thus there is no margin of error, as we have not statistic. The numbers account for 100% of the population. We do not conclude from a small dataset to an infinite population and thus there is no margin of error calculation needed.
Simply speaking, if I want to divine the number of white balls in a bowl with white, red and black balls in it, I do not need to do statistics, if I can draw each and every ball and simply count.
I have the feeling that 'It's within the margin of error' is his way of discrediting the data and all conclusions drawn from it.
Dicrediting is a strong word. The amount of change in the highest leages is so small, it could be almost anything for any race. There are changes, but saying it is for x or y reason has serious pitfalls.
|
So it seems Protoss is the way to go if you want to be Grandmaster. Though the statistics indicate that the few Terran GMs have a way better winrate.
I'm looking at the statistics for the last few days since I can't reach GM with Terran. I feel that I'm too bad for the advanced mechanics and will never improve to GM level. Maybe if I quit my life and everything but this isn't an option. So currently I'm in a phase where I want to play Protoss/Zerg for a few weeks to decide my further approach to SC...
|
On March 28 2012 03:02 slane04 wrote:
Since we are mainly talking about foreign Terrans, I don't think looking that at global race distributions is all that helpful. We all know Korean Terrans are doing fine. Aggregating with Korean stats masks the trend occuring in Europe and the Americas. Let's look at Europe and America from 1.2 to 1.4.2 then
America
Terran
Masters: 30.2 --> 28.8 = -1.4
Diamond: 28.0 --> 24.9 = -3.1
Platinum: 30.2 --> 24.3 = -5.9
Gold: 30.9 --> 26.8 = -4.1
Silver: 31.6 --> 30.2 = -1.4
Europe
Terran
Masters: 29.8 --> 28.3 = -1.5
Diamond: 27.7 --> 25.8 = -1.9
Platinum: 30.4 --> 24.9 = -5.5
Gold: 32.3 --> 26.7 = -5.6
Silver: 34.0 --> 30.9 = -3.1
So we see here almost a 6% drop in both servers for Terrans at the platinum level between 1.2 and 1.4.2. This is huge amount of players switching. In America and Europe, they are the most underrepresented race in all leagues except silver and bronze. I would argue that a sub-25% race representation in leagues is a problem, much like how Zerg was in the early days.
Back in 1.2, Zerg populations were struggling in all leagues, but a combination of buffs, learning the race, and metagame shifts have catapulted Zerg populations in 1.4.2 to be most represented in the American and Europe servers, Platinum and up to and including Masters.
I'm not suggesting any imbalance at the highest levels of play. Watching MKP play Terran is mind boggling. Who needs tanks against zerg when you have bio + godly micro + close position spawning on Metalopolis. But is makes sense that a small population of the Starcraft community would stick with a race that they have the most success with. Others will be propelled to other races by nerfs and buffs, regardless of whether it actually affects how they play the game. Still others will switch races due to perceptions of race difficulty, real or imagined. Finally, some will just quit altogether out of frustration.
The point is, Terrans in the Platinum to Diamond are struggling and due to any of the factors listed above, have decided to switch races. This becomes a problem for the health of the game when the XvT matchup becomes less and less frequent. 1/3 of the variety of the game is taken out at certain levels of play with such large race population disparities.
So in response to "Where did all the Terrans go [in the mid-high levels]?" In the Americas and Europe, they switched or quit. This is not good for the game and willl probably not be addressed until HoTS. In the meantime, as a newly promoted Platinum Terran, I'm getting my butt kicked a little bit. It is a bit disheartening, but I'm still learning a lot and make too many mistakes to name. I can improve everywhere. However, real or imagined, my perception of the skill discrepency at this level is affecting my play. Can't wait for the tanky unit, the battle-hellion.
Another thing to note in this analysis. I don't know if anyone bothered to do the math, but it seems people take this 3% reduction lightly. When you drop from 28% to 25% of total players, you actually lost 11% from your previous players. When you drop from 30% to 24% you lost 20% from your previous players.
About Plansix point. Margin of error is only applied when you take a small sample from a complete dataset. sc2ranks, as far as I know, is the whole dataset, so applying margin of error doesn't make sense.
BigJ mentioned that the average points is almost the same for every race and that is an indication that every race is equally active in the ladder. That's not necessarily true. The bonus pool plays a big role in this situation. You can have 200 points having played 1000 games and having played 30 games. We would only know that for sure if sc2ranks had "average number of games per race" data.
|
On March 28 2012 03:13 VoO wrote:
So it seems Protoss is the way to go if you want to be Grandmaster. Though the statistics indicate that the few Terran GMs have a way better winrate.
I'm looking at the statistics for the last few days since I can't reach GM with Terran. I feel that I'm too bad for the advanced mechanics and will never improve to GM level. Maybe if I quit my life and everything but this isn't an option. So currently I'm in a phase where I want to play Protoss/Zerg for a few weeks to decide my further approach to SC...
don't switch.
Gm means nothing. Being high master is enough.
|
purely anecdotal, but since MKP won I have seen a ridiculous surge in terrans. 50%+ of my games are against terrans.
Oh, and they're all doing the same broken, inefficient cheesy builds that marine king does. It's so hard to scout these guys, their buildings make no sense @_@
|
On March 28 2012 03:24 darkscream wrote:
purely anecdotal, but since MKP won I have seen a ridiculous surge in terrans. 50%+ of my games are against terrans.
Oh, and they're all doing the same broken, inefficient cheesy builds that marine king does. It's so hard to scout these guys, their buildings make no sense @_@
The ladder is incredibly trendy, whenever someone wins a tournament, the next day the ladder is full wannabe's.
|
Another thing to note in this analysis. I don't know if anyone bothered to do the math, but it seems people take this 3% reduction lightly. When you drop from 28% to 25% of total players, you actually lost 11% from your previous players. When you drop from 30% to 24% you lost 20% from your previous players.
Ya, that makes sense petro1987. Forgot about taking the entire league population into account. So for example
Terran NA Platinum
Platinum: 30.2 (29,203) --> 24.3 (9,527) = -5.9,
with total Terran population in parenthesis
and the total Platinum league population going from 96,730 -->39,130.
39,130 * 0.302 = 11817.3
39,130 * 0.243 = 9508.59 (yay numbers not matching)
Going from 11817.3 --> 9508.59 Terran players is a ~20.5% decrease. And it wasn't like Terran was over-represented at the Platinum level in 1.2, which makes this decrease all the more alarming and difficult to explain without invoking some degree of disparity in race difficulty at certain levels of play.
Good point to take account activity by race into account.
|
On March 28 2012 03:10 Plansix wrote:Show nested quote +On March 28 2012 02:53 Thrombozyt wrote:
I believe his null hypothesis is that race changes are random but there is an equal chance for each race to switch to another. At least that's the only null hypothesis that makes sense.
Plansix is just overlooking a small detail:
This would only work if the sc2ranks numbers are a small subset of a much much much larger bnet population. I'm not entirely sure on that point, but I believe that sc2ranks makes database dumps and therefore catches everyone. Thus there is no margin of error, as we have not statistic. The numbers account for 100% of the population. We do not conclude from a small dataset to an infinite population and thus there is no margin of error calculation needed.
Simply speaking, if I want to divine the number of white balls in a bowl with white, red and black balls in it, I do not need to do statistics, if I can draw each and every ball and simply count.
I have the feeling that 'It's within the margin of error' is his way of discrediting the data and all conclusions drawn from it. Dicrediting is a strong word. The amount of change in the highest leages is so small, it could be almost anything for any race. There are changes, but saying it is for x or y reason has serious pitfalls.
If discrediting is a strong word, which weak word would you like to see? Tainting?
While I agree, that attributing a reason to the changes is littered with pitfalls, you cannot argue with the fact that the Terran population is decreasing faster than any other race. It's consistent and it's significant. Your 'margin of error' approach doubted that there even was a decline - and it was wrong.
|
On March 28 2012 04:08 Thrombozyt wrote:Show nested quote +On March 28 2012 03:10 Plansix wrote:On March 28 2012 02:53 Thrombozyt wrote:
I believe his null hypothesis is that race changes are random but there is an equal chance for each race to switch to another. At least that's the only null hypothesis that makes sense.
Plansix is just overlooking a small detail:
This would only work if the sc2ranks numbers are a small subset of a much much much larger bnet population. I'm not entirely sure on that point, but I believe that sc2ranks makes database dumps and therefore catches everyone. Thus there is no margin of error, as we have not statistic. The numbers account for 100% of the population. We do not conclude from a small dataset to an infinite population and thus there is no margin of error calculation needed.
Simply speaking, if I want to divine the number of white balls in a bowl with white, red and black balls in it, I do not need to do statistics, if I can draw each and every ball and simply count.
I have the feeling that 'It's within the margin of error' is his way of discrediting the data and all conclusions drawn from it. Dicrediting is a strong word. The amount of change in the highest leages is so small, it could be almost anything for any race. There are changes, but saying it is for x or y reason has serious pitfalls. If discrediting is a strong word, which weak word would you like to see? Tainting? While I agree, that attributing a reason to the changes is littered with pitfalls, you cannot argue with the fact that the Terran population is decreasing faster than any other race. It's consistent and it's significant. Your 'margin of error' approach doubted that there even was a decline - and it was wrong.
The funny thing is that he actually tried to discredit the data with this "margin of error" thing. Then you came down and bursted his bubble. He had no option left but to take it back and tried to "smooth" things out.
You know someone is out of arguments in a discussion when they try to come up with things that don't even make sense to prove their claims/opinions. I guess the bottomline is: why is he trying so desperately to discredit it?
|
On March 28 2012 04:08 Thrombozyt wrote:Show nested quote +On March 28 2012 03:10 Plansix wrote:On March 28 2012 02:53 Thrombozyt wrote:
I believe his null hypothesis is that race changes are random but there is an equal chance for each race to switch to another. At least that's the only null hypothesis that makes sense.
Plansix is just overlooking a small detail:
This would only work if the sc2ranks numbers are a small subset of a much much much larger bnet population. I'm not entirely sure on that point, but I believe that sc2ranks makes database dumps and therefore catches everyone. Thus there is no margin of error, as we have not statistic. The numbers account for 100% of the population. We do not conclude from a small dataset to an infinite population and thus there is no margin of error calculation needed.
Simply speaking, if I want to divine the number of white balls in a bowl with white, red and black balls in it, I do not need to do statistics, if I can draw each and every ball and simply count.
I have the feeling that 'It's within the margin of error' is his way of discrediting the data and all conclusions drawn from it. Dicrediting is a strong word. The amount of change in the highest leages is so small, it could be almost anything for any race. There are changes, but saying it is for x or y reason has serious pitfalls. If discrediting is a strong word, which weak word would you like to see? Tainting? While I agree, that attributing a reason to the changes is littered with pitfalls, you cannot argue with the fact that the Terran population is decreasing faster than any other race. It's consistent and it's significant. Your 'margin of error' approach doubted that there even was a decline - and it was wrong.
I will agree that they are decreasing like every other race. I will even agree that they have had larger decreases. However, I will not agree that anyone has provided conclusive evidence that it is some sort of gameplay difficulty that is the main and overwhelming cause the decrease. Specificlly, I don’t think the highest leagues have the drop off people are describing, since masters has the lowest decrease out of all the leagues for both protoss and terran.
|
On March 28 2012 04:36 Plansix wrote:Show nested quote +On March 28 2012 04:08 Thrombozyt wrote:On March 28 2012 03:10 Plansix wrote:On March 28 2012 02:53 Thrombozyt wrote:
I believe his null hypothesis is that race changes are random but there is an equal chance for each race to switch to another. At least that's the only null hypothesis that makes sense.
Plansix is just overlooking a small detail:
This would only work if the sc2ranks numbers are a small subset of a much much much larger bnet population. I'm not entirely sure on that point, but I believe that sc2ranks makes database dumps and therefore catches everyone. Thus there is no margin of error, as we have not statistic. The numbers account for 100% of the population. We do not conclude from a small dataset to an infinite population and thus there is no margin of error calculation needed.
Simply speaking, if I want to divine the number of white balls in a bowl with white, red and black balls in it, I do not need to do statistics, if I can draw each and every ball and simply count.
I have the feeling that 'It's within the margin of error' is his way of discrediting the data and all conclusions drawn from it. Dicrediting is a strong word. The amount of change in the highest leages is so small, it could be almost anything for any race. There are changes, but saying it is for x or y reason has serious pitfalls. If discrediting is a strong word, which weak word would you like to see? Tainting? While I agree, that attributing a reason to the changes is littered with pitfalls, you cannot argue with the fact that the Terran population is decreasing faster than any other race. It's consistent and it's significant. Your 'margin of error' approach doubted that there even was a decline - and it was wrong. I will agree that they are decreasing like every other race. I will even agree that they have had larger decreases. However, I will not agree that anyone has provided conclusive evidence that it is some sort of gameplay difficulty that is the main and overwhelming cause the decrease. Specificlly, I don’t think the highest leagues have the drop off people are describing, since masters has the lowest decrease out of all the leagues for both protoss and terran.
@petro:
Let him be. It's not an easy distinction to make when statistics apply and what statistics.
@plansix
So then.. why do you think Terran numbers are decreasing much stronger than the number of the two other races?
|
maybe because terran had the most players at the beginning and people are bored with them?
not everyone is playing this game to become a pro
|
|
|
|
|
|
EPT
