|
Always being a Zerg enthusiast, I strongly believe that the zergling is a formidable starting unit. Being a melee unit, it has to travel to its enemy in order to inflict damage. This does not create a problem when facing the Protoss base unit, the zealot, but poses an interesting dilemma when facing the Terran marine. Luckily, there are several upgrades that can be researched that help the zergling overcome the travel distance and inflict rapid damage on the enemy marine.
But how much do these upgrades help the zergling? What sort of odds should Zerg players expect when assaulting the Terran menace? I present to you the
The Zergling/Marine Efficiency Experiment By The_Overmind_77
Objective: To measure the efficiency of the zerg base unit, the zergling, versus the terran marine as the zergling takes on various upgrades.
Procedure:
1. Determine the Zergling/Marine Ratio
First, it is necessary to decide on the number of zerglings to which we should combat the same number of marines. I first considered having same mineral value ratios (two zerglings for each marine, a fifty mineral value for each race) but then came the complication.
Under these conditions, the zerglings most often completely overwhelmed the marines with zero losses. When finding the efficiency in this condition (efficiency being number of marines killed per number of zerglings killed represented as a percent), the ratio yields a number over zero, which gives an efficiency of undefined parameters. When this ratio is received, it indicates that we need to increase the number of marines until at least one zergling is killed.
I therefore decided to use a 1:1 ratio. In combat there would be one marine present for each zergling. After a few preliminary tests this ratio was determined to be effective.
2. Determine Parameters
After finding the ratio to use, I looked to what one would find in a common Starcraft match to determine on what numbers of zerglings and marines to test. Since a person man only select twelve units at once in Starcraft, I decided that I would test up to twelve marines and twelve zerglings. Finding battles above this number in-game are rare and the presence of other units (hydralisks, lurkers, medics) would warp the results.
Since zerglings are hatched in twos, the numbers to be tested are 2-12 marines and zerglings.
In considering what upgrades to test, I will start with a test group of no upgrades and advancing up to crackling status (which means Metabolic Boost + Adrenal Glands). I will also add a special group for my own personal interest, the Burrow group, testing the effectiveness of burrow in comparison to the other groups.
Therefore we have four test groups: No upgrades, Metabolic Boost, Metabolic Boost + Adrenal Glands, and Burrow. Adrenal Glands by itself was not tested because zerglings with this upgrade but without Metabolic Boost are rarely (if at all) seen in-game.
3. Create Map
A custom UMS setting map was created to test this procedure. It would be run by triggers to prevent any human interference. Being run by triggers, however, does create several limitations which should be noted. See Things to Consider at the bottom of this post.
For groups 1-3, triggers were run to create the set number of marines and zerglings at a set distance from each other. The zerglings were given their specified upgrades. The zerglings were then set to patrol to the marine location. The number of surviving zerglings and marines were counted after the conflict was completed and the number placed in excel. All scenarios were run under fastest speed. Six trials were run and the efficiency was averaged.
For the burrowing group, I set up the zerglings to ambush the marines. They would be burrowed and then spring up from under the marines. If you create zerglings already burrowed with the triggers, however, the zerglings are all clustered on one single spot. Since this is rarely the case in actual games, I first created the zerglings unburrowed with a trigger. I manually selected and burrowed them, causing them to be in the same formation as when initially created by the trigger but not on top of each other when burrowed. Then, I set a switch to spawn the marines on top of them after I had completed this. The zerglings where then ordered, via trigger, to patrol to the location where they were created, causing them to spring up and attack the marines. The survivors would then be counted and documented in excel. Six trials were run and the efficiency was averaged.
Results:
Below is a chart of the average number of survivors per group per number of marines and zerglings. Following is a chart showing the change of efficiency. Click to enlarge!
Note: The results for Burrowing with 2 marines and zerglings was not shown on the efficiency chart because it yielded the ratio of a number over zero, which is inaccurate results. This shows that the Zerg has very high (yet difficult to measure) efficiency percent for this test run.
Analysis (Interesting things to note!):
1. As a general trend, the efficiency of zerglings decreases as the number of combatants increases. This may be common sense seeing as the zerglings can only engage the marines on the outside of the cluster, but is still interesting to note.
2. Burrowing ambushes yields the highest efficiency increase from the baseline of unupgraded zerglings. At some instances, it yields over a 350% increase. It allows the zerglings to engage the marines without traveling a large distance. I never thought that burrowing was that advantageous! Of course, the limitations are that an ambush must be utilized (which isn't always the easiest thing to do).
3. Metabolic Boost provides a larger increase in efficiency than Adrenal Glands.
4. Around 6 marines to 6 zerglings, the efficiency of groups 1-3 converge to around 40%. The efficiency flatlines for these three groups after this with only a slight variation in percentages.
5. The fourth group, the Burrowing group, did not receive such a significant decrease as the number of marines decreased. The efficiency for this group varied quite widely in a dampering simple harmonic oscillator motion. I have yet to explain this.
6. When in low numbers, zerglings almost always defeat marines in efficiency.
7. Remember: If you want to get actual mineral value efficiency, multiply these values by two. This is because in the test I use a 1:1 ratio, whereas the actual mineral value ratio is two zerglings to one marine. This means that when the efficiency flatlines fro groups 1-3, it is actually stabilizing around 100%, meaning that for every 50 minerals you spend in on zerglings you are destroying approximately 50 minerals worth of marines for the Terran player.
Quick Points to Note (If you are too lazy to read through the experiment):
- Zerglings are better against marines when in small scale battles (1-6 units per team)
-A burrowing ambush provides the largest benefit to the zergling efficiency
-Metabolic boost provides the zerglings with the largest increase of the non-ability upgrades
-A Zerg will almost always kill AT LEAST the same mineral value of enemy units when engaging zerglings with non-upgraded marines across the same elevation of terrain
Things to Consider
1. The tests were run under very controlled situations. The results will of course vary under different combat formations and will further vary when additional units or intensive micro is added.
2. These tests did not include the two marine upgrades (stimpack and range upgrade). Range would most definitely change the trials of group 1-3 (not by an incredible amount, I believe, as each marine would at most only gain an additional hit) but would not affect the burrowing trial (group 4) because the zerglings are already in range at the start of combat.
3. The burrowing trial is dependent on very particular formations. However, it is important to note that when ambushed, the marines might be moving in single-file and thus this might cause a higher jump in efficiency.
4. No medics or firebats (God help the zerglings) are introduced in these cases. Enter: Lurkers!
Extended Study: Burrowing Graph:
Micronesia got me irked when he told me when he started talking about the Simple Harmonic Motion of the burrowing graph. I had used it to describe the graph's initial appearance..and he said that he doubted it continued like that.
So I decided to extend the burrowing range to 24 units on each side. I repeated the testing procedure and graphed the data.
And what wacky results! It looks nothing like a dampened oscillator. The surviving zerglings varied GREATLY...but it seems that efficiency is increasing as the number of zerglings increases even though the number of marines increases as well. This is probably because the large number of zerglings cause the marines to move about when they unburrow, giving plenty of time for free hits by the zerglings. Interesting, eh?
Conclusion
Although many of you might not think that this data is worth anything, I found it very interesting to actually have hard number to crunch when viewing how zergling efficiency increases as they are upgraded in various situations.
Now why, do you ask, did I not investigate zealot combat? Well, zealot/zergling micro depends very much on formation and the surrounding of the zealots, and many of the upgrades (burrowing, metabolic boost) would not provide as much of a noticeable effect because the units are both melee. Travel time is not as much as an issue when considering just the combat (but can affect many things such as flanking, formation, etc). Obviously this would be a MONSTER to study and thus I avoided it.
Hope you enjoyed this small, nifty experiment and my results! FOR THE SWARM!
|
+ Reserved +
Oops my bad i posted below, at least i got first post. why did you post when i posted my TvZ Vessel theory _ _
Thread in overall, is amazing and insightful that people can use.
|
Very nice, but I think the main problem with early lings has always been more a matter of larvae efficiency than mineral efficiency. Also, I think T/P have more minerals in the beginning? All these factors warp the results and contribute to the evolution of today's strategy.
|
Well I didn't try to worry myself about when the lings were being produced or the money, just the hardcore combat and effects of the upgrades ><.
|
You have way too much time on your hands to analyze all this. Statistics are neat, but don't win battles. I think the important thing is your point that the more ranged units there are, the less effective melee units are agaisnt them is pretty accurate though among most RTSs.
On a side note one time in a game I played my zergling fought a marine and they both killed each other at the same time O.O
|
On May 11 2007 16:32 aokces wrote:
You have way too much time on your hands to analyze all this. Statistics are neat, but don't win battles. I think the important thing is your point that the more ranged units there are, the less effective melee units are agaisnt them is pretty accurate though among most RTSs.
On a side note one time in a game I played my zergling fought a marine and they both killed each other at the same time O.O
Yes....wayyyyy too much time.
On an interesting note, on one of the trials I actually had that happen. One marine was left versus one zergling and they both killed each other at the same time. Hahahh.
|
On May 11 2007 16:14 TheOvermind77 wrote:
2. These tests did not include the two marine upgrades (stimpack and range upgrade). Range would most definitely change the trials of group 1-3 (not by an incredible amount, I believe, as each marine would at most only gain an additional hit) but would not affect the burrowing trial (group 4) because the zerglings are already in range at the start of combat.
What!? Stim pack is THE BEST tactic to use against a flock of lazy lings. i believe that not adding stim you are eliminating your causal game. most terran upgrade stim first for the purpose of killing lings faster.
|
Quite some work you've done there...Yo, pm me and let's do some tests with toss upgrades vs terran upgrades etc! This stuff looks mad neat :O.
|
On May 11 2007 16:45 LySin wrote:Show nested quote +On May 11 2007 16:14 TheOvermind77 wrote:
2. These tests did not include the two marine upgrades (stimpack and range upgrade). Range would most definitely change the trials of group 1-3 (not by an incredible amount, I believe, as each marine would at most only gain an additional hit) but would not affect the burrowing trial (group 4) because the zerglings are already in range at the start of combat.
What!? Stim pack is THE BEST tactic to use against a flock of lazy lings. i believe that not adding stim you are eliminating your causal game. most terran upgrade stim first for the purpose of killing lings faster.
The problem was getting the marines to stim via a computer and triggers (not that easy). Also, when stim pack is upgraded, the Terran most likely has medics and/or firebats so UHOH! That makes things a bit nasty.
Yes, I know it is a VERY narrow, precise experiment, but the complications became massive if I wandered too far out of the box. Perhaps in the future I will have to figure out a way, though...
|
GrandInquisitor
New York City13113 Posts
this is interesting, but in situations where you have naked marines (medic-less) engaging zerglings, it is very unlikely that the zerglings will have adrenal/metabolic upgrades -.-
|
cool test, using a 1:1 ratio sounds about right, because 50 zerg minerals don't exactly equal 50 terran minerals since terran mines much faster with more scvs at least in early game when this info is most likely to be used, (only lings vs only rines).
except for adrenal glands of course.
|
On May 11 2007 17:24 GrandInquisitor wrote:
this is interesting, but in situations where you have naked marines (medic-less) engaging zerglings, it is very unlikely that the zerglings will have adrenal/metabolic upgrades -.-
Go burrow! I swear, it is one of the most underused abilities, despite how many people talk it up.
The problem I find is fitting it into my build order.
|
On May 11 2007 17:24 GrandInquisitor wrote:
this is interesting, but in situations where you have naked marines (medic-less) engaging zerglings, it is very unlikely that the zerglings will have adrenal/metabolic upgrades -.-
Ah but thats why you get fast gas and evolve it quickly. (obviously not for adrenal upgrade) 
Sometimes i mass up 12 speedlings before getting any more drones and if the terran isn't blocking his ramp, or hes only got 3-4 rines blocking it, you basically win the game.
|
|
|
On May 11 2007 17:30 Jonoman92 wrote:
cool test, using a 1:1 ratio sounds about right, because 50 zerg minerals don't exactly equal 50 terran minerals since terran mines much faster with more scvs at least in early game when this info is most likely to be used, (only lings vs only rines).
except for adrenal glands of course.
Ahh I nvr realized it's cheapter marine 50 minerals than zerg's 50 minerals. hmm!
|
I personally find burrow flanks useless against high tier terrans. They just make more firebats than usual and you are then at a position where you have less drones. If you want to pursue the early burrow flank thing, you need to kill his scout scv first and make lings after AND make sure his second scout doesn't show up. it's really hard. But yeah 350% against marines sounds great
|
On May 11 2007 17:37 TheOvermind77 wrote:Show nested quote +On May 11 2007 17:24 GrandInquisitor wrote:
this is interesting, but in situations where you have naked marines (medic-less) engaging zerglings, it is very unlikely that the zerglings will have adrenal/metabolic upgrades -.-
Hi randInquisitor, Go burrow! I swear, it is one of the most underused abilities, despite how many people talk it up. The problem I find is fitting it into my build order.
I used barrow last week zvt on longinus. I also obs a game where XenZei on RoV used it last month. In both situations the Terran 2rax 1faced. I noticed that Xenzei was experienced enough to understand from scouting that Terran was going to break the sunken line with tanks and a M&M mod surrouding the tanks. It is easy to know where to barrow lings because of the range of tanks In both games we both destroyed the contain easy. This suggests that scouting and understanding of timing along with travel paths on each map, helps to determine when to include barrow into a BO. When countering other BOs one can find the timing and location of barrow's usefullness.
---I-Emerge
P.S. Thanks for the study! This is very interesting and helpfull.
|
I appreciate the praise...I hope this helps further confirm some truths about burrow.
|
On May 11 2007 18:54 I-Emerge wrote:Show nested quote +On May 11 2007 17:37 TheOvermind77 wrote:On May 11 2007 17:24 GrandInquisitor wrote:
this is interesting, but in situations where you have naked marines (medic-less) engaging zerglings, it is very unlikely that the zerglings will have adrenal/metabolic upgrades -.-
Hi randInquisitor, Go burrow! I swear, it is one of the most underused abilities, despite how many people talk it up. The problem I find is fitting it into my build order. I used barrow last week zvt on longinus. I also obs a game where XenZei on RoV used it last month. In both situations the Terran 2rax 1faced. I noticed that Xenzei was experienced enough to understand from scouting that Terran was going to break the sunken line with tanks and a M&M mod surrouding the tanks. It is easy to know where to barrow lings because of the range of tanks In both games we both destroyed the contain easy. This suggests that scouting and understanding of timing along with travel paths on each map, helps to determine when to include barrow into a BO. When countering other BOs one can find the timing and location of barrow's usefullness. ---I-Emerge P.S. Thanks for the study! This is very interesting and helpfull.
Thats a nice strategy, but isnt the terran going to be a bit suspicious of the low ling count?
|
Low ling count could just mean drone powering. Or lings could be behind sunks or elsewhere on the map. Lots of possibilities!
|
United States24625 Posts
I found this a fun read (being a man of science) and interesting in general, but didn't provide much new insight. Maybe you can work out some minor kinks in what you did and then do something else using the adapted method. What would be really cool would be if you could reasonably disprove some kind of a mechanical misconception in bw.
|
On May 11 2007 20:48 micronesia wrote:
I found this a fun read (being a man of science) and interesting in general, but didn't provide much new insight. Maybe you can work out some minor kinks in what you did and then do something else using the adapted method. What would be really cool would be if you could reasonably disprove some kind of a mechanical misconception in bw.
Heh, I know, it really only went to show things that many people already knew. Still, I found the numbers interesting to crunch. Finding a 'mechanical misconception' that can be disproved isn't something easy to do...
|
United States24625 Posts
On May 11 2007 16:14 TheOvermind77 wrote:
5. The fourth group, the Burrowing group, did not receive such a significant decrease as the number of marines decreased. The efficiency for this group varied quite widely in a dampering simple harmonic oscillator motion. I have yet to explain this.
I thought about this while I was in the shower just now (because that's what I think about while I shower). Your classification of this behavior has a 'dampering simple harmonic motion' is premature. I don't believe you have enough data to claim that, even though I'll admit the small bit you do have graphed is strangely like a damped sine wave. I have a feeling if you can carefully increase the number of units, the observed effect will decay away as the interactions become less discretized. If you repeat this part of the experiment you also may get a differently shaped graph.
The fact that only the Burrowing group didn't get immediately hurt by an increase in numbers stands to reason, I believe. It is the only of the 4 tests you ran where marine range has little to no benefit, almost regardless of numbers.
|
On May 11 2007 19:16 fusionsdf wrote:Show nested quote +On May 11 2007 18:54 I-Emerge wrote:On May 11 2007 17:37 TheOvermind77 wrote:On May 11 2007 17:24 GrandInquisitor wrote:
this is interesting, but in situations where you have naked marines (medic-less) engaging zerglings, it is very unlikely that the zerglings will have adrenal/metabolic upgrades -.-
Hi randInquisitor, Go burrow! I swear, it is one of the most underused abilities, despite how many people talk it up. The problem I find is fitting it into my build order. I used barrow last week zvt on longinus. I also obs a game where XenZei on RoV used it last month. In both situations the Terran 2rax 1faced. I noticed that Xenzei was experienced enough to understand from scouting that Terran was going to break the sunken line with tanks and a M&M mod surrouding the tanks. It is easy to know where to barrow lings because of the range of tanks In both games we both destroyed the contain easy. This suggests that scouting and understanding of timing along with travel paths on each map, helps to determine when to include barrow into a BO. When countering other BOs one can find the timing and location of barrow's usefullness. ---I-Emerge P.S. Thanks for the study! This is very interesting and helpfull. Thats a nice strategy, but isnt the terran going to be a bit suspicious of the low ling count?
Good observation,
There was post I read here that explains the uses of misinformation. The post focused on hiding units in your main to throw off your opponet. In the game I played using barrow I remembered the post's comments and I reset my rally points for all ling producing Hats to a location not normaly scanned in my main. This helped me take advantage because of the lack of Bats created based on misinformation.
Also...
On May 11 2007 19:18 TheOvermind77 wrote:
Low ling count could just mean drone powering. Or lings could be behind sunks or elsewhere on the map. Lots of possibilities!
---I-Emerge
|
On May 11 2007 21:27 micronesia wrote:Show nested quote +On May 11 2007 16:14 TheOvermind77 wrote:
5. The fourth group, the Burrowing group, did not receive such a significant decrease as the number of marines decreased. The efficiency for this group varied quite widely in a dampering simple harmonic oscillator motion. I have yet to explain this.
I thought about this while I was in the shower just now (because that's what I think about while I shower). Your classification of this behavior has a 'dampering simple harmonic motion' is premature. I don't believe you have enough data to claim that, even though I'll admit the small bit you do have graphed is strangely like a damped sine wave. I have a feeling if you can carefully increase the number of units, the observed effect will decay away as the interactions become less discretized. If you repeat this part of the experiment you also may get a differently shaped graph. The fact that only the Burrowing group didn't get immediately hurt by an increase in numbers stands to reason, I believe. It is the only of the 4 tests you ran where marine range has little to no benefit, almost regardless of numbers.
Ok, it is late at night and your post bugged me. I kept thinking about what that graph would look like as it was extended. Go check out the "Extended Study" section at the end of the updated experiment.
|
On May 11 2007 16:47 ilovezil wrote:
Quite some work you\'ve done there...Yo, pm me and let\'s do some tests with toss upgrades vs terran upgrades etc! This stuff looks mad neat :O.
seconded 
|
United States24625 Posts
A few months ago I did something similar regarding pvz. I had armies of zealots fight armies of lings, ultras, and ultra ling, while varying all the upgrades. I didn't publish my results or display my data in a pretty way but it seemed to me to be somewhat useful. Maybe I or someone else should go back through that study again like the way this one was done.
IIRC my conclusion was that zealots>ultras if they have reasonably comparable upgrades, lings>zealots with reasonably comparable upgrades, and ultra+ling >> zealots. When I say comparable upgrades, I mean if zealots don't have at least +2 yet, the ultras don't have any special upgrades at all yet, etc.
|
On May 12 2007 10:48 micronesia wrote:
A few months ago I did something similar regarding pvz. I had armies of zealots fight armies of lings, ultras, and ultra ling, while varying all the upgrades. I didn't publish my results or display my data in a pretty way but it seemed to me to be somewhat useful. Maybe I or someone else should go back through that study again like the way this one was done.
IIRC my conclusion was that zealots>ultras if they have reasonably comparable upgrades, lings>zealots with reasonably comparable upgrades, and ultra+ling >> zealots. When I say comparable upgrades, I mean if zealots don't have at least +2 yet, the ultras don't have any special upgrades at all yet, etc.
You should get some stats on that.
I'm considering taking my spreadsheets and making another experiment...I have a few ideas that may be useful.
|
United States24625 Posts
On May 12 2007 13:17 TheOvermind77 wrote:Show nested quote +On May 12 2007 10:48 micronesia wrote:
A few months ago I did something similar regarding pvz. I had armies of zealots fight armies of lings, ultras, and ultra ling, while varying all the upgrades. I didn't publish my results or display my data in a pretty way but it seemed to me to be somewhat useful. Maybe I or someone else should go back through that study again like the way this one was done.
IIRC my conclusion was that zealots>ultras if they have reasonably comparable upgrades, lings>zealots with reasonably comparable upgrades, and ultra+ling >> zealots. When I say comparable upgrades, I mean if zealots don't have at least +2 yet, the ultras don't have any special upgrades at all yet, etc. You should get some stats on that.
It will be difficult to perform effectively. The goal would be to see what units are more effective given equal money, and a lot of this, in actuality, comes down to the micro. Also, how many zealots have the same cost as an ultralisk? I can count gas and minerals the same (4 zeolots = 1 ultralisk), but that's only partially fair. The way the computer controls the units is key.
|
Many of the problems that you are talking about is the same reason my study was so limited in scale. The problem is that you can only do so much under a controlled environment...once you start adding different units and "micro" it turns into a "o shit" scenario. That is why I really couldn't study certain things.
|
You too can just work together on it so the comp ai doesn't matter. 
|
On May 12 2007 13:50 Slayer91 wrote:You too can just work together on it so the comp ai doesn't matter.
The only problem is that everything must be done the same every time or else the experiment isn't credible...human error might cause our micro, etc to warp the results. But it is an idea :D.
|
On May 11 2007 18:29 evanthebouncy~ wrote:Show nested quote +On May 11 2007 17:30 Jonoman92 wrote:
cool test, using a 1:1 ratio sounds about right, because 50 zerg minerals don't exactly equal 50 terran minerals since terran mines much faster with more scvs at least in early game when this info is most likely to be used, (only lings vs only rines).
except for adrenal glands of course. Ahh I nvr realized it's cheapter marine 50 minerals than zerg's 50 minerals. hmm!
It's not. Due to the gas usage of Zerg, they usually have more bases mining, but the gas usage is comparable to the mineral usage. But seeing as you have less gas, you have minerals left over after gas is spent, leaving lots of money to be spent on lings.
|
|
|
|
|
|
EPT![[image loading]](http://i173.photobucket.com/albums/w56/flyingninja77/Chart.jpg)
![[image loading]](http://i173.photobucket.com/albums/w56/flyingninja77/Graph.jpg)
![[image loading]](http://i173.photobucket.com/albums/w56/flyingninja77/graph2.jpg)
