|
Sc2Gears
Skillcraft.ca: A SC2 Expertise Study
Do you macro like a pro?
The above posts have one thing in common: they have inspired me to think about SC2 performance metrics I've never considered before! I would like to share my ideas with you:
1) Average unspent gas and minerals over time: Minimizing both unspent minerals and gas will ultimately minimize total average unspent resources. One of my major weaknesses is that I tend to be really good at spending one resource but have an excess of the other that I am unable to spend of due to the lack of the other resource. This metric would allow anyone with this issue to see which resource was contributing to the unspent more and how long it was doing so for.
2) Percentage of units hotkeyed divided by total number of units produced: I micro better when I remember to hotkey my units. This metric would allow players to track how many of units were hotkeyed during a game.
3) Average idle production facility time per production facility divided by game length: The less time your production facilities are idle, the more units they are producing, which means you are spending more and ultimately have a larger army. This metric would measure both 1) the optimality of the amount of production structures you have and 2) how well you are macroing from them. I realize that shorter games will skew this statistic since no production facilities are built at the start of the game, so it would be necessary to divide this metric into groups of games with varying lengths.
These next two metrics are very simple, self explanatory and more helpful for beginners but still useful in my mind:
4) Percentage of game time supply blocked
5) Average energy of CC, Queen or Nexus
I would love to hear if you guys have any more metrics to add, why you like or dislike my ideas and if you think there are any feasible ways to track these metrics.
Thanks for reading! 
|
I think you could create a ton of different of metrics, most of them don't become popular because they're hard to measure, like 2-5
|
Hopefully more people come up with statistically sound metrics so that we can better measure and identify factors that lead to better players, thus helping improvement process.
|
I feel strongly as if there's a build order tester/generator (or sc2gears itself) that measures number 3. Whatever it is, I haven't used it before... but I'm pretty sure one exists.
Also, Percentage of time with active larva injects on hatcheries would be a much more reliable metric than queen energy, as one or two creep spreading queens, or extra queens to defend air would skew the metric, where %time with active injects on hatcheries wouldn't.
I'd also like a metric for amount of time hatcheries have 3+ larva. I do wonder how many larva I miss out on because I don't spend the ones I've got fast enough.
|
a player who makes only lings and drones all game off one base would excel with flying colors over all of your metrics.
I think the best (but difficult) metric to measure would be action completion time. There was a good thread on it not too long ago: http://www.teamliquid.net/forum/viewmessage.php?topic_id=250059
I think map awareness would be something to add to your list though, or amount of time scrolling around the map versus using hotkeys.
|
Thanks for your feedback guys.
I think you could create a ton of different of metrics, most of them don't become popular because they're hard to measure, like 2-5
I agree that 2 and 3 would be hard to measure but I dont think 4 and 5 would be.
Also, Percentage of time with active larva injects on hatcheries would be a much more reliable metric than queen energy, as one or two creep spreading queens, or extra queens to defend air would skew the metric, where %time with active injects on hatcheries wouldn't.
I'd also like a metric for amount of time hatcheries have 3+ larva. I do wonder how many larva I miss out on because I don't spend the ones I've got fast enough.
Couldnt agree with you more man, although a Queen that is using energy to spread creep is good as well! :p
a player who makes only lings and drones all game off one base would excel with flying colors over all of your metrics.
Not if they have 1 extractor going, forget to make overlords and inject larva, :p I will def check out the ACT metric, sounds interesting!
|
I think that the existing queen injection metrics are quite interesting and deserve further study. For those that are unfamiliar, if you play zerg sc2gears can determine the average gap between between injections and the percentage of time that a hatchery has that spell cast on it. There is also a graphic that depicts your injection status over time.
What is interesting is to look at what level of injection gap is necessary as a minimal level of competency. Contrary to what is sometimes suggested by casters, pros don't have absolutely perfect injection timings, so if your injection timings are not absolutely perfect, but within the range shown by someone like Nestea, then maybe working on injection timing is not the main priority in terms of training at the moment. I feel that research into norms for this statistic seems to be lacking at the moment.
|
#2 seems pretty useless. Hotkeying more or less is a playstyle difference, not directly responsible for wins.
But, just like baseball, starcraft 2 seems very suitable for a deep numeric analysis.
How about something like this: amount of enemy supply killed per unit of your supply per minute?
|
The question is: how deep can you reach into the starcraft engine to get these numbers?
Another one would be: value of enemy units/structures killed/destroyed per 100 minerals spent per minute?
|
All of these are fairly useful, but there are legitimate circumstances to be on the "bad" side of them. Saving minerals and gas can be good - stocking up for a big warp-in, an expansion, etc. Not all units need to be hotkeyed; clever hotkey set-ups are really well-thought-out, not when everything is hotkeyed, but when the important things are hotkeyed (even more than once, in some cases), and the important things aren't eating up hotkeys. Idle production facilities are also not inherently bad; the average 3-gate expand includes a lot of idle time on warpgates, not out of lack of skill, but because the units aren't needed yet. The 3 gates are just insurance in case the opponent gets aggressive. And there are legitimate reasons to stock energy on CC's (scans, especially in TvT), queens (mostly for transfuse), and nexi (save chrono for an important upgrade, or to chrono all warpgates in case of a big attack).
Extractor tricks aside, supply block is pretty universally bad, though
|
The "Do you macro like pro?" link is broken 
|
|
|
|
|
|
EPT
