EPTDo you macro like a pro? - Page 63
| Forum Index > SC2 General |
|
Cyro
United Kingdom20345 Posts
| ||
|
Numiro
3 Posts
http://ggtracker.com/players/690687#?page=1 is my WoL account, inactive untill today. http://ggtracker.com/players/952593 is my HotS beta account. | ||
|
dsjoerg
United States384 Posts
On March 02 2013 20:15 Numiro wrote: How accurate is the GGtracker? We've looked at the Spending Skill data several different ways and it is accurate. There are many skills to the game, Spending is only one of them. So there are players who are much better in their Spending than in the other parts of the game, you might be one of those. | ||
|
Numiro
3 Posts
On March 05 2013 10:34 dsjoerg wrote: We've looked at the Spending Skill data several different ways and it is accurate. There are many skills to the game, Spending is only one of them. So there are players who are much better in their Spending than in the other parts of the game, you might be one of those. Okey thank you for the reply, I always felt like I was spending stuff to late (like seeing 400 minerals, sending probe when it hits 450 and expansion goes down at like 800 minerals and so on) If you're responsible for ggtracker in any way, shape or form I'd like to thank you alot, you have no clue how useful that has been for me who've just started playing this game! | ||
|
MattMannion
United States25 Posts
| ||
|
Deleted User 245622
184 Posts
On March 06 2013 02:34 MattMannion wrote: for sq take your total collected resources and divide it by the game length then use your average unspent to get your sq im sorry but thats not what sq sq is: SQ= 35*(0.00137*I-LN(U))+240 | ||
|
dsjoerg
United States384 Posts
On March 06 2013 02:34 MattMannion wrote: for sq take your total collected resources and divide it by the game length then use your average unspent to get your sq How do you get your total collected resources? | ||
|
Race is Terran
United States382 Posts
edit: heres one that does | ||
|
Faster69
69 Posts
| ||
|
nucLeaRTV
Romania822 Posts
| ||
|
Sufinsil
United States760 Posts
| ||
|
Jezebeth
United States23 Posts
| ||
|
DinosaurPoop
687 Posts
| ||
|
HaloLegend98
United States54 Posts
I.e. downloading a data set to a program, say excel, the going through it? | ||
|
Coldblackice
United States49 Posts
On February 16 2015 13:13 HaloLegend98 wrote: Is there anyway that this analysis could be updated with more recent info from HOTS? I.e. downloading a data set to a program, say excel, the going through it? I'd be interested in this, as well! | ||
|
dchaudh
58 Posts
I wanted to get folks thoughts on an "accounting adjustment" to the unspent resources metric that's used to calculate SQ. As has been mentioned, overqueuing can skew SQ. However, we can adjust for this skew by adding back the resource cost of overqueued units. Let's take a simple example at a specific moment during the game: The concept: Let's say that at a particular moment in the game, I have 300 unspent minerals and my production facilities are one command center and one reactored barracks. Let's assume that the CC has 1 SCV in production and 2 more SCVs in queue; let's also assume that the reactored rax has 2 marines in production and 4 more in queue (i.e., two production cycles in queue, in addition to the pair of marines that's currently under production). At the moment, the calculation of SQ take into account the 300 unspent minerals (along with the unspent minerals at every other point during the game, ofc). My proposal is that a more accurate value for unspent minerals (at this point in time) would be as follows: 300 unspent minerals (as-reported figure) + 50 mineral add-back for the second SCV in the CC queue + 100 mineral add-back for the third and fourth marines in the rax queue -------------------------------------------------------------- 450 unspent minerals (pro-forma figure) In this example, I implicitly apply a rule that it's okay to have one production cycle in queue at any given point (in addition to the current production cycle). In performing the calculation above, we're simply "pretending" that the SCV and the two marines were not overqueued to begin with. This rule is of course up for debate (e.g., it's debatable whether any queuing ought to be allowed - even of one production cycle). Conceptually, we can apply this pro-forma adjustment (to use an accounting term) to unspent resources at every time-unit for which data is reported in replays. Using adjusted unspent resources (rather than as-reported unspent resources) would yield an SQ which is a better indication of the quality of macro. Feasibility: At the moment, I'm not too familiar with the format of replay data. From using SCELight, it seems that the unspent resources at any given point is known. However, I'm not sure whether it's possible to determine the number of units in queue at each production structure at any given time. (Note: even if it's the case that we're not explicitly provided the queue per structure at any point in time, we may be able to deduce it based on other provided data - I'm just not sure atm.) My questions: 1) For those of you who are familiar with replay data formats: is it possible to infer the queue per structure at any given moment in a replay? 2) (Really only worth discussing if the answer to #1 is yes) What is the appropriate queue threshold to use when making the adjustment to unspent resources (e.g., no queuing allowed at all vs. 1 production cycle allowed in queue vs. 2 production cycles allowed in queue..., etc.) | ||
|
Cascade
Australia5405 Posts
 | ||
| ||
