| peekn United States. September 17 2011 13:40. Posts 1145 | Profile Blog # |
| Wow amazing post, thanks for the data really nice read. |
| |
|
| CustomKal Canada. September 17 2011 13:45. Posts 437 | Profile # |
I think one of the most impressive things about that to realize the the difference from masters and diamond at this point is not as much in their macro skill as it is in their decision making. Up to that point however the macro steadily improves and does so quite drastically between the leagues.
This almost reinforces the good macro will help you get better mentality for it will get you to those higher leagues, and it isn't until your macro is quite up their that decision making starts to make a difference (hence the difference between diamond and masters, decision making and reaction).
Grandmasters to masters difference is also quite interesting, but that may be just because that shows all of masters which waters down the top of masters who are probably on par with grandmasters. |
| |
|
| ChineseWife United States. September 17 2011 13:49. Posts 371 | Profile # |
| this seems to just tell you your score is higher if you have longer games, doesnt really tell you how good your macro is. I just did the SQ test for a longer game where i had sloppy macro and i got a higher score than most 10-12 minute games i played despite the ratio of unspent resources to average collection rate being lower |
| |
|
| Rhaegar99 Australia. September 17 2011 13:52. Posts 1000 | Profile Blog # |
| Probably one of the best data analysis article involving sc that Ive read. Good work! |
| |
|
| Kring Portugal. September 17 2011 13:57. Posts 67 | Profile # | |
| | Evolution complete |  |
|
|
| yawnoC United States. September 17 2011 13:58. Posts 2569 | Profile # |
| amazing post ... I have to reread this because I am sure there is something I missed o_O |
| | GG | |
|
|
| TheUnderking Canada. September 17 2011 14:09. Posts 158 | Profile # |
This confirms what I and my friends already knew, my macro is weak, but my micro and decision making is good. This is indeed an excellent analysis
I'm top of platinum, but my SQ is barely gold using my last 10 games |
| | THE PACT IS SEALED! |  |
|
|
| Wolf Korea (South). September 17 2011 14:12. Posts 3046 | Profile Blog # |
| This is really cool. I'd love to see even more data for this. |
| | What's essential is invisible to the eye. It's only with a scout that one sees rightly. | http://twitter.com/proxywolf | |
|
|
| xOff United States. September 17 2011 14:26. Posts 245 | Profile # |
What was really interesting to me are the rare cases of certain players in very low leagues that performed better than grandmasters players when it comes to unspent resources. Does that mean that they spent all their money because they simply mined wayyy less money? or maybe theres some monsters of macro that cannot micro for the life of them  |
| | Anything can be accomplished through sheer discipline. |
|
|
 Seeker  Korea (South). September 17 2011 14:29. Posts 17735 | Profile Blog # |
| This guy deserves a medal or a star or something.... this is beyond expectations awesome... |
| | FXO.NA SC2 Manager / FXO Translator |  |
|
|
| Crypt United States. September 17 2011 14:29. Posts 17 | Profile # |
Sorry if this has already been posted, but there is an aspect that significantly exaggerates this data.
That is: lower skilled players stay in games longer after they've already lost. This goes beyond things like command center floating and building hiding, which I would believe happens more as the skill level is lower. The player in the position of winning stop or slows creation of workers and spending, since trying hard isn't as necessary any longer. This situation is much more common in lower league play and is not a determinant of their win/loss performance and should not be viewed as a skill-based cause of their league placement. Idra is the perfect example of the opposite extreme. There is very little "dead time" near the end of his games. If he's lost, he leaves immediately without slowing down macro rather than dragging out the game and lowering his macro/time "score."
Result: The gaps between leagues in all presented graphs is exaggerated. |
|
|
| LetoAtreides82 United States. September 17 2011 14:34. Posts 1188 | Profile # |
On September 17 2011 03:58 GabrielB wrote:
Show nested quote +On September 17 2011 03:49 perestain wrote:
It would be nice if sc2gears was picking up on the SQ-quotient so everybody can conveniently track his development.
Overall the most beautiful thing about it is that you dont even need to parse a replay to get the most practical information. Ingenuity beats data overkill.
As some people have said, it seems hard to automate this from the replay file (the replay file doesn't seem to save that information). However, sc2ranks probably can get this done pretty easily, as it gathers its info from the match history.
Yes I think SC2Ranks would be able to do this. |
| |
|
| LetoAtreides82 United States. September 17 2011 14:38. Posts 1188 | Profile # |
On September 17 2011 03:42 xai_death wrote:Updated my tool at http://beijers.eu/share/sq.html so you can calculate average etc. It's almost like the Excel someone posted, but Excel takes such a long time to start  .
Cool thanks |
| |
|
| fuzzy_panda New Zealand. September 17 2011 14:46. Posts 1148 | Profile Blog # |
| wow that sure puts my 500th post to shame. epic post man some real nice analysis. |
|
|
| GMonster September 17 2011 14:49. Posts 679 | Profile Blog # |
Im ranked 198 in Grandmaster... so i can't answer any of your questions  |
| | GrandMaster Terran NA Server / Mod @ justin.tv/incontrol |
|
|
| docvoc United States. September 17 2011 14:49. Posts 3679 | Profile Blog # |
| I understand that this looks very comprehensive, but where are the error bars? surely this work isn't infallible also i love that this has been done, but wouldn't it be much more useful to have done say, 300 people? 100 is not enough to get rid of systematic errors and such. Also what we see here is the top 100 best bronze players and so on correct? I understand that you used scatter plots and then inserted a trendline, but wouldn't it have been beneficial to take a myriad of level players, mid level to higher instead of just the 100 best? I would like to say that i'm very happy to see this caliber of math on a forum like this :D |
| | User was warned for too many mimes. |  |
|
|
| HostileEgo United States. September 17 2011 14:51. Posts 6 | Profile # |
| Neat, pretty accurate for me. My SQ is 71 and I'm Diamond. |
|
|
| Azzur Australia. September 17 2011 14:52. Posts 5852 | Profile Blog # |
On September 17 2011 13:49 ChineseWife wrote:
this seems to just tell you your score is higher if you have longer games, doesnt really tell you how good your macro is. I just did the SQ test for a longer game where i had sloppy macro and i got a higher score than most 10-12 minute games i played despite the ratio of unspent resources to average collection rate being lower
In a longer game, there is a greater chance of being sloppy since it's not difficult to play like a pro in the early stages. |
| |
|
| Azzur Australia. September 17 2011 14:53. Posts 5852 | Profile Blog # |
On September 17 2011 14:29 Crypt wrote:
Sorry if this has already been posted, but there is an aspect that significantly exaggerates this data.
That is: lower skilled players stay in games longer after they've already lost. This goes beyond things like command center floating and building hiding, which I would believe happens more as the skill level is lower. The player in the position of winning stop or slows creation of workers and spending, since trying hard isn't as necessary any longer. This situation is much more common in lower league play and is not a determinant of their win/loss performance and should not be viewed as a skill-based cause of their league placement. Idra is the perfect example of the opposite extreme. There is very little "dead time" near the end of his games. If he's lost, he leaves immediately without slowing down macro rather than dragging out the game and lowering his macro/time "score."
Result: The gaps between leagues in all presented graphs is exaggerated.
Nope, I don't buy this. I believe there are definite gaps between leagues. The empirical data supports what many higher level players know intuitively. |
| |
|
| MarKeD Afghanistan. September 17 2011 14:56. Posts 153 | Profile # |
On September 17 2011 14:49 docvoc wrote:
I understand that this looks very comprehensive, but where are the error bars? surely this work isn't infallible also i love that this has been done, but wouldn't it be much more useful to have done say, 300 people? 100 is not enough to get rid of systematic errors and such. Also what we see here is the top 100 best bronze players and so on correct? I understand that you used scatter plots and then inserted a trendline, but wouldn't it have been beneficial to take a myriad of level players, mid level to higher instead of just the 100 best? I would like to say that i'm very happy to see this caliber of math on a forum like this :D
Ready more closely, only the grandmasters are the top 100, the other leagues are a randomly selected division. |
|
|
| Prev 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 Next |
|
|
|
Store
