|
On September 17 2011 02:41 TheLight wrote:Show nested quote +On September 17 2011 02:34 Xaeldaren wrote: I'd love to calculate my SQ but I'm completely illiterate when it comes to Excel...I'm even using the 2000 version. Could anyone give me a hand and tell me how to go about it? Here's a Excel 2000 template: http://www.mediafire.com/?h1cd1f2kdj16p6fJust put in your numbers and it'll do the rest for you.
Thank you so much!
|
wow i have to give a big props for the hard work you put in running these regressions haha
I really thing making the chart with the sq of various top players would be interesting
|
This is an incredible post! You put in an amazing effort that yielded very interesting results. I look forward to seeing how these results are used to compare regional ladders in the future!
|
Omg... this is a nice analys. damn it, idont have excel. what can i use instead of Excel to calculate my SQ?
|
Wow, I'm absolutely amazed that you've actually found a somewhat useful way to track your macro progress. I may actually have to do this myself and start tracking my progress. The next thing we need to do is get someone to use the SC2gears plugin API to create a plugin for calculating your SQ.
|
On September 17 2011 02:13 MrBarryObama wrote: What surprised me the most was that protoss creates the least workers on average! I assumed it would be terran, since they have mules and protoss have chrono boost. Could it be that protoss expands the least, on average? Or that typically protoss loses the fewest probes?
I was wondering about this as well. My first thought was it must be counting Mules, but by the time the first mule is out, Protoss should have a worker lead unless they're chrono-ing tech/units.
The divergence appears to start at the 5-6 minute mark. It's possible for Terran to start the FE by then, but is it possible for the Terran to actually build from it to explain the difference? I know PvP is also the matchup with the latest expansion timing, so that could be dragging down Protoss as well?
Also, I wonder if this is being skewed by 4-gate, since you stop at 20 or so probes which is really early. Other than the 11/11, Terran openings/all-ins don't involve cutting workers do they (they super delay the expansion or pull workers to junk your economy)?
|
This is so goddamn awesome, thanks!
|
Is there a way to find out the data without having to watch every single replay until the end?
|
|
I just put the SQ formula in excel and its quite funny to compare something like this. Seems my SQ as Protoss is around 70 and I'm in silver, had a very small amount of data, only put in like 5 games or so.
And as Terran its between 50-60. Maybe I should have stayed Protoss ^^^
Anyway great job by OP. Really interesting stuff. Someone awesome should make a kickass spreadsheet!
Nvm, seems the awesome people are faster than me as well! GJ on spreadsheet!
|
I am going to go print these graphs out and hang them in my bedroom. BEAUTIFUL!!!
|
Just Calculated my own SQ, took my last 10 ladder games (both wins, and losses) and got an SQ or 67.8, not too shabby for a Platinum League player I suppose
|
Great post, but you did not account for if a player maxes out quickly and then wins without losing significant portions of it, which would cause average unspent resources to skyrocket, and you also did not account for workers lost - losing workers is a big reason why players in higher leagues would need to make more.
|
On September 17 2011 02:41 Impervious wrote:Wow. Awesome stuff! Show nested quote +On September 17 2011 01:52 Ihsahn wrote: Is there any way this index could be integrated into sc2gears? I think that could help a lot! Seriously, this is a good idea.
We should contact the wizzard of sc2gears!^^
|
I'm truly impressed by the level of detail put in to this post. It explains clearly how the macro difference in terms of descending leagues. Good job!
I do have a question though for the SQ(spending quotient) part. As a fellow academician doing a physics degree with experience in modelling. What are the constraints of your SQ formula. Also, could you explain further on the derivation of it? My guess is that the SQ formula is an empirical formula based on the existing data that you had in hand. If that is the case, my suggestion is to collect a different set of data and compare whether the distribution fits the formula or not.
Once again, thank you for this truly educational post!!
|
sickest post i've ever read here.
are you a wizard?
|
That was...an amazingly interesting read O-o. Truly eye-opening.
I applaud you, best sir.
|
You can get far more accurate measures using sc2gears and piles of replays. (mlg, the dozens of replay sites around, etc.)
The way data is calculated, the length of masters/ GM games and the time spent maxed on army values would skew the results lower than the actual, and comparative difference in usage of macro mechanics also would stratify the leagues far more. Additionally, 100 games per league is far too few to really encompass a proper range, the likelihood of having a predominant trend in games or matchups is far too likely. (Ex. 2/3 of the protoss replays could easily be PvP.)
The idea is good though, and props on putting in the effort! Though, I can't help but feel a lot of the time spent doing this could be reduces or automated to be more accurate.
|
This doesn't seem right, since I'm a gold player and after inputting 20 games my average comes in at 71, which I presume is pretty good?
|
Tried out that SQ-thing as a gold zerg over 10 games: SQ_avg = 57 SQ_max = 85 SQ_min = 37.
So if I'd take the SQ to it's edge one could say that at my best I macro like a GM and at my worst I macro like a bronzer! xD
|
|
|
|