Author Topic: Theorycrafting! (Because I like competitive metagames too much)  (Read 124910 times)

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #275 on: February 25, 2012, 05:47:41 AM »
Ok, tried to get a real read on unit HPs.

Marine- 2 cannon hits
Banshee- 8 cannon hits
Mutalisk- 6 cannon hits
Zealot: 4 cannon hits
Hydralisk: 6 cannon hits?
Roach- 9 cannon hits
Ghost- 5 cannon hits
Tank- 5 cannon hits
Immortal- 15 cannon hits
Colossus - 25ish cannon hits?
Ultralisk - 35ish cannon hits?

I'm fairly certain that Cannon damage is unchanged--which is to say, one cannon hit is 20 damage.  This seems to line up with Storm dealing 100 damage.  So...this puts my best estimates for HP at...

Marine: 30
Zealot: 80
Roach: 180
Hydralisk: 120
Ghost: 100
Mutalisk: 120
Banshee: 160
Tank: 100
Immortal: 300
Colossus: 500
Ultralisk: 700 (doesn't line up with Heal Wave's description, since Heal Wave heals more than half an Ultra's HP.  Possibly there's armor in the mix here?  Possibly my HP estimates are a little high?  Possibly it's due to zerg units having some passive regen--I suspect Mutalisk is actually 100 HP, and just regenerates the 1 HP required to survive a storm/five canon hits)

And yes, Hydralisks, Mutalisks, and Banshees all survive one storm.

Adjustments to this:

Mutalisk: 100 (they sometimes survive storm just due to regen, but not if there are any marines shooting at them).

Immortal--Clearly dies in three storms, suggesting about 250 HP, but takes more than 15 cannon hits to kill.

Roach: HP based on storm and HP based on cannons roughly matches.

Colossus: Storm tests  suggest pretty much exactly 400 HP.

Ultralisk: Actually took 8 storms to kill.  (I'll attribute this to HP regen, which means that yeah, 700 HP is accurate).


So...(all HP numbers approximate...)

Marine: 30
Zealot: 80
Roach: 180
Hydralisk: 120
Ghost: 100
Mutalisk: 100
Banshee: 160
Tank: 100
Immortal: 250
Colossus: 400
Ultralisk: 700

I'd like to do some damage tests too, but this is harder without a testing partner.  I do know that Ultralisks are about...

Ultralisk (dmg): 20 (+60 Armored)

Based on the fact that they 2HKO marines, but high 3HKO Roaches.  (This is all very much an estimate, though).

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #276 on: March 20, 2012, 08:42:21 AM »
Starjeweled, looking at unit statistics in the custom map editor.

Just to remind myself how to get there... (EDIT: thanks to Myself (the user of that name) for pointing me in the right direction)

Map -> Map Status -> look at stuff.

Most of these match my previous testing fairly well, but there's a lot of interesting damage bonuses (and some weird cooldown values that I don't think are right...EDIT: tested Marine and Hydra firing rate a little, they seem to match in-game o_O)

Photon Cannon:

800 HP, 3 armor
Damage: 20
Cooldown 0.8 (faster than in-game huh?)  (25 DPS)
Range: 10.5

Marine:

30 HP, 0 armor
Damage: 4
Cooldown 1.5 (wait, really?  Seems like that would be noticeable) (2.6 DPS)
Range 4

Zealot:
70 HP, 3 armor
Damage: 8 (+2 biological)   ---- (nb: I think this is x2 for the two hits.  I'll assume it is)
Cooldown: 1.2 (13.3 DPS, 16.7 DPS Biological)
Range: 0.1

Roach:
145 HP, 1 armor
Damage: 16 (+6 light)
Cooldown: 2.0 ( 11 DPS light, 8 DPS armored)
Range 4

Hydralisk:
110 HP, 0 armor
Damage: 8 (+2 armored)
Cooldown: 0.4 (WTF?? this can't be right?) (20 DPS light, 25 DPS armored)
Range 5

Ghost: (Side note: ghosts are biological Psionic, but not light; same as in-game, but something I didn't know)
100 HP, 0 armor
Snipe Damage: 40
Damage: 8 (+12 light)
Cooldown: 1.5 (13.3 DPS light, 5.3 DPS armored)
Range 6

Mutalisk:
100 HP, 3 armor
Damage: 30 + 20 + 10 (for each respective bounce hit)
Cooldown 1.5246 (39.4 DPS, or 19.6 DPS singletarget)
Range 5 (Wait, that high??)

Banshee:
125 HP, 3 armor
Damage: 15 (+15 armored)    ------- (nb: I think this is x2 for the two hits.  I'll assume it is.)
Cooldown 1.4 (21.4 DPS light, 42.8 DPS armored)
Range 6

Siege Tank:
100 HP, 1 armor
Damage: 18 (+40 structure)
Cooldown 2.0 (9 DPS, 29 DPS structure)
Range 11

Immortal
250 HP, 3 armor (plus hardened shields--didn't find where the shields were specified, I assume it's about 20 shields?)
Damage: 35 (+55 structure)
Cooldown 1.8 (19.4 DPS, 50 DPS structure)
Range: 5

Colossus
400 HP, 3 armor
Damage: 15 (+25 light)   ---------- (nb: I think this is x2 for the two hits.  I'll assume it is)
Cooldown: 2.2 (13.6 DPS armored, 36 DPS light)
Range 7

Ultralisk
500 HP, 5 armor
Damage: 16 (+52 armored)
Cooldown 0.861 (18.5 DPS light, 79 DPS armored)
Range 1
« Last Edit: April 21, 2013, 05:52:26 PM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #277 on: March 20, 2012, 09:04:50 AM »
And of course, the thing that everyone cares about...fastest killing of cannons for the cost.  For the cost of 1000 energy you can deal....

Marines:
0.7 DPS per marine (but they're free!)

Zealots:
8.3 x20 = 166 DPS

Roaches:
6.5 x13.3 = 87 DPS

Hydralisks:
17.5 x10 = 175 DPS

Banshees:
38 x4 = 154 DPS

Immortals:
48 x3.33 = 161 DPS

Ultralisks:
75 x2 = 151 DPS
« Last Edit: April 21, 2013, 05:54:19 PM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #278 on: March 20, 2012, 09:21:21 AM »
As far as killing air units goes...air units have high defence and aren't armored, so Hydras are held largely in check (only 12.5 DPS) buuut Ghosts kinda lack the raw stats to do better than that regardless (11.3 DPS).  Pretty close at the specialized task of just killing air, though.  One Mutalisk deals 34 DPS if all three bounces hit air (gamebest for the cost), but their AI doesn't prioritize air unlike hydra/ghost, and they are frailer (Espeically with hydra, ghost, and cannon AI prioritizing Mutas).


metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #279 on: March 21, 2012, 04:38:00 AM »
Spells

Most of the spells do exactly what they describe; Time Bomb is 50% time speed, for instance.  Heal Wave is the one interesting one; the stats in the editor are...

Heal Target 1: 4800
Heal Target 2: 2880
Heal Target 3: 1728
Heal Target 4: 1036
Heal Target 5: 622

Now, obviously these aren't the HP values healed.  My theory is that maybe this is the amount the effect would heal if it lasted a full second (or something) but each bounce of the Heal Wave is some increment of that, like 1/16 of a second.  Going by the 1/16 theory, that would make it...

Target 1: 300 HP
Target 2: 180 HP
Target 3: 108 HP
Target 4: 65 HP
Target 5: 39 HP

This has some interesting consequences.  Say you have five Hydras, and they get stormed.  You heal wave them.  Only three of them will be restored to full HP--the other two will actually be at risk of dying to the next storm.

The other interesting thing to note is the maximum range on grabbing next target: range 5.  (So an easy way to remember it is about the same range as the attack range for most of the units in the game).  Each of the four jumps can go an additional 5.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #280 on: March 21, 2012, 03:55:44 PM »
So...confirmed thorugh testing: all the units I thought did double attacks (Zealot, Colossus, etc) do do double attacks.

Also, Movement speeds:

Photon Cannon: 0.0
Marine: 2.25
Zealot: 2.25 + 0.5 (plus charge ability) (EDITED)
Roach: 2.25
Hydralisk: 2.25
Ghost: 2.25
Mutalisk 3.75
Banshee: 2.75
Siege Tank: 2.25
Immortal: 2.25
Colossus: 2.25
Ultralisk: 2.25


Hmm...just some general musings on these stats...

If there were no spells, Ultras would actually be quite counterable, even with a pure ground army.  Mass Hydra with a trickle of Zealots to draw aggro (so that the Ultras attack a zealot rather than splashing a hydra).  That, however, could be countered by mass colossus, because colossus OHKO Zealots and high 2HKO hydras.  But Colossus get countered by...Ultras!

I'm thinking the problem with starjeweled balance is less the units and more the spells; particularly Storm shredding everything that counters ultras.

More interesting notes, though: these stats give a better explanation of why Provoker-style 1 colossus, 2 banshee, 3 hydra is so hard to deal with.  Colossus melts light units, and Banshees melt armor units.  Compositions like that make a strong case for using Ghosts, however--Ghosts aren't light, so don't melt when hit once by a Colossus.  (Not sure what to use for the rest of the wave; Ultras still melt to the double banshee.  Maybe just Colossus Banshee Ghost to counter Colossus Banshee Hydra?  Melt the Hydras, ghosts live long enough to gain a Banshee advantage...).

Defence stat: So...the defence stat adds an interesting dimension I was not previously aware of.  Let's say your opponent has a ton of marines out; what do you build?  High defence units can almost ignore marines.  For instance, it takes longer for marines to kill a Zealot than to kill a Roach, despite Roaches having double the HP.  Note: Ultras having 5 defence to Marine's 4 attack doesn't make them immune--in SC2 when you have >= defence than the opponent's attack, you take 0.5 damage per hit.  So...Ultras go down in 1000 marine shots....well ok more than that because they regen too, but marines can kill them eventually.
« Last Edit: March 22, 2012, 06:20:32 AM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #281 on: March 22, 2012, 06:44:13 AM »
Did a bit of testing, and Mirror is correct: Zealots are faster.  They seem to just have the Charge upgrade applied to them, Which in addition to the charge ability, adds a passive speed boost of 0.5 (bringing them up to 2.75, exactly equal to Banshees--which matches my in-game testing).  All other ground units do seem to move at identical speed, though.

Makes me wonder what other units have upgrades on them.  Marines obviously don't have stim or combat shield.  Roaches don't have speed.  What else is there...?  Hydra Range, Colossus range, Ghost energy, Ultralisk Defence, Cloak, Siege Mode.  I'm going to guess "no" on all of these.

Oh yeah, one extra set of stats:

Base (Armored Mechanical Structure):
400 HP, 1 defence, 0.398 regeneration rate (slightly higher than Zerg regeneration rate, which is 0.273).


Not too many surprises, except for the really low defence, which means it's vulnerable to everything (20 marines will kill it in about 10 seconds--you need to spam units).  And also the regeneration rate, which means that if you manage to pull off a decence you aren't screwed the next time they come anywhere close to your base.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #282 on: March 22, 2012, 07:38:14 AM »
More stuff:

Immortal Shields: 2.  Immortal Shield regen: 0.  (I assume it's a modified version of Hardnened shields that reduces all damage to 1 instead of reducing all damage to 10)

But yes, immortals will never regenerate their shields, and this also explains why their shields seem to disappear instantly when stormed.  Also, even a marine will take off half their shields.




Speed at killing Ultras, for 1000 energy worth of units:

Zealots: 166 DPS
Roaches: 75 DPS
Hydralisks: 125 DPS
Ghost: 20 DPS (not counting snipe)
Muta (vs 3 targets): 147 DPS
Muta (vs 1 target): 82 DPS
Banshee: 142 DPS
Siege Tank: 22 DPS
Immortal: 55 DPS
Colossus: 18 DPS (assumes singletarget)
Ultralisk: 146 DPS (assumes singletarget)

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #283 on: March 23, 2012, 11:00:37 AM »
So...who actually wins in a fight, assuming pure concentrations of each unit, and all units are able to instantly get into range?

Since I know I have a couple of new readers, I'll do a quick review of some ancient RTS formulae

Let's say there's a 2v1 unit situation, where two smaller units fight one bigger unit.  Let's say the only difference between the units is HP.  The way to get this balanced is for the "big" unit to have 3x the HP of the "small" units.  It will lose 2/3 of its HP before the first unit dies, and 1/3 of its HP fighting the second unit.  In a 3v1 you need 6, and in general to balance things you need

Ratio = (n+1)*n/2

Which we can solve with the quadratic formula, so...

n = 1/2*(-1+sqrt(1+8*Ratio))

For any isolated unit matchup, we can get the relative durability ratio pretty easily, which lets us compute the numerical difference at which they'd be balanced.


Zealot
Vs Marine: 10:1
Vs Roach: 1:1.17 (cost effective, but barely)
Vs Hydra: 1:1.11 (cost effective)
Vs Ghost: 1.15:1 (cost effective--well...assuming we ignore Snipe...)
Vs Siege Tank: 1.07:1 (cost effective.  In fact, one zealot beats one Siege Tank...sad)
Vs Immortal: 1:2.9 (cost effective, but not by as much as you'd think)
Vs Colossus assuming no splash: 1:6 (cost effective but barely
Vs Colossus assuming maximum splash: 1:22 (lol)
Vs Ultra assuming no splash: 1:4 (cost effective)
Vs Ultra assuming maximum splash: 1:12 (roughly cost-even)

Roach
Vs Marine: 5.6:1
Vs Zealot: 1.17:1 (mildly cost inefficient)
Vs Hydra: 1:1.28 (roughly cost-even)
Vs Ghost: 1.67:1 (very cost efficient)
vs Siege Tank: 1.18:1 (Does a single Siege Tank beat ANYTHING?  Seriously, lol)
vs Immortal: 1:2.5 (mildly cost efficient)
vs Colossus assuming no splash: 1:2.7 (cost efficient)
vs Colossus assuming maximal splash: 1:5 (still kinda cost efficient)
vs Ultra assuming no splash: 1:8.0 (mildly cost inefficient)
vs Ultra assuming maximum splash: 1:36 (lol)

Hydralisk
vs Marines: 6.8:1
vs Zealots: 1.11:1 (cost inefficient)
vs Roaches: 1.28:1
vs Ghost: 1.42:1 (note: doesn't account for Snipe--that should even things up)
vs Muta: 1:1.9 (about cost-equal...)
vs Banshee: 1:1.45 (mildly cost-efficient)
vs Siege Tank: 2.0:1 (very cost-efficient)
vs Immortal: 1:1.5 (cost-efficient)
vs Colossus without splash: 1:3.3 (mildly cost-efficient)
vs Colossus with splash: 1:7 (mildly cost inefficient)
vs Ultra without splash: 1:3 (mildly cost efficient)
vs Ultra with splash: 1:6.6

Siege Tank
vs Marines: 4.5:1 (Ok, this is quite sad.  The 10 marines produced by building up to 300 energy would kick the crap out of the siege tank you could buy with that energy)
vs Zealots: 1:1.07
vs Roaches: 1:1.17
vs Hydras: 1:1.97
vs Ghosts: 1.5:1
vs Immortal: 1:3
vs Colossus assuming no splash: 1:3
vs Colossus assuming maximal splash: 1:6
vs Ultra assuming no splash: 1:9
vs Ultra assuming maximal splash: 1:45

Ladies and gentlemen, the least cost efficient unit in the game, where every matchup is cost inefficient, and the only question is how MUCH cost inefficiency there is.  At least Tanks don't lose to a solo Ghost?

Immortal:
vs Marine: 20:1
vs Zealot: 2.9:1 (cost inefficient)
vs Roach: 2.7:1
vs Hydra: 1.6:1 (cost inefficient)
vs Ghost: 4.9:1 (cost efficient)
vs Siege Tank: 3.3:1 (cost efficient)
vs Colossus: 1.15:1 (assuming no splash)
vs Ultra assuming no splash: 1:3.0 (cost inefficient)
vs Ultra assuming maximal splash: 1:6.0 (cost inefficient)

Colossus (no splash / Maximal splash)
vs Marines: 32:1 / 545:1
vs Zealots: 6:1 / 22:1
vs Roach: 2.7:1 / 5:1
vs Hydra: 3:1 / 7:1
vs Ghost: 5:1 / 15:1
vs Siege Tank: 3:1 / 6:1
vs Immortal: 1.01:1
vs Ultralisk: 1:3.8 / 1:9 (if the Ultra splashes maximally)

Ultralisk (no splash / Maximal splash)
vs Marines: 41:1 / 871:1
vs Zealots: 4.3:1 / 11.6:1
vs Roaches: 8:1 / 36:1
vs Hydra: 3:1 / 6.6:1
vs Ghosts: 8:1 / 41:1
vs Siege Tanks: 9:1 / 45:1
vs Immortals: 3:1 / 6:1
vs Colossus: 4:1 / 9:1

Hmm...I'll edit the rest in later.
« Last Edit: March 23, 2012, 05:33:13 PM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #284 on: March 29, 2012, 03:28:58 AM »
Pobes vs Zeelot

This is a terrible, unbalanced SC2 map.  There are several blatantly obvious ways to just break the map provided everyone is coordinated, and the only way the map functions at all is by taking advantage of idiots.  I don't actually recommend playing it necessarily.

And yet, it has a shiny appeal to me.  Possibly I just like exponents.


The core mechanic involves upgrading your income building, and each time you upgrade, your income doubles.  Now, exponents are powerful: there's a very obvious way to exploit this, by players pooling all their money and having one person upgrade first, and then refund everyone who helped them upgrade once they have higher income.  Just how good is this strategy?

Actually...not as overpowered as you might expect.  Let's math.

Two players pool their resources, and one player upgrades in half the time.
Then that player sends resources to the first player, until they can upgrade.

If x is your income, the first player gets 2x income, and then the second player gets 3x income.  Which means overall the time spent is...

1/2t + 1/3t = 5/6t

So...83% of the time it would take acting independently, or 20% faster.  What happens if three people coordinate?

1/3t + 1/4t + 1/5t = 47/60t

So...78% of the time it would take acting independently, or 28% faster.

Well...what if you had a ton of people working together, like 100, how good would it be?

Sum(1/x) where x = n .. (2n-1)

But we can rewrite this as

1/n * Sum( n / (n+x) ) = 1/n * Sum( 1/(1+x/n) )
for x = 0... n-1

But this should look familiar--this is just the construction of the integral.  If we take the limit as n goes to infinity, this would give us the integral of 1/x evaluated between (1,2), which is...

ln(2) - ln(1) = ln(2) = 0.69

In other words, with infinitely many people it would reduce the time to 69% of the original, or 44% faster.

Which is to say, in a game of horrible imbalance, sharing resources...really isn't mind-blowingly imbalanced, no matter how many people do it.


Also: whee, a use for calculus.  Can't say I use it very often these days.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #285 on: March 30, 2012, 08:58:42 AM »
Pobes vs Zeelot

So...there's this interesting mechanic with the miners.  The first four miners are relatively straightforward:

cost: effect

512: mines 1 mineral once every 8 seconds
1024: mines 1 mineral once every 4 seconds
2048: mines 1 mineral once every 2 seconds
4096: mines 1 mineral once every second

I think everyone can see these are about equal.  But then we get to...

20480: mines 6 minerals every second

For the cost, this is more efficient than anything that came before it.  It's the same cost as five L4 miners, but it mines as much as six L4 miners.  And up the chain it goes.

81920: mines 36 minerals every second
262144: mines 216 minerals every second
1000000: mines 1296 minerals every second

The L6 is even more cost-efficient, the L7 is yet more cost-efficient, and the L8 is more cost-efficient than that.

Now, my first instinct was to go for the biggest bang for the buck.  Save up for a L8 very early.  The problem is that this is an RTS--when you spend money on miners, that means you're going to collect additional money and be able to buy more stuff later.  So...what's the optimal route here?



The answer depends on additional variables.  It's a double feedback loop.  You can use gas to buy miners who get you minerals, and you can use minerals to generate more gas.  Turning minerals back into gas is fairly expensive, but everyone gets one (and only one) gas generator that generates 512.  Outside of this generator, however, it's reasonably expensive to produce gas.  With 1024 minerals, you can create a gas generator that produces 32 gas per second (increasing your gas generation by 6%!)

I'd love to say that I pulled out a clever solution here, but in the end I just plugged these numbers into a spreadsheet, and the results that I found were...

1. Don't get a L5 miner (the ones that mine 6 minerals per second).  At least not after your gas generator has been upgraded to produce 512 a second.  Extra gas generation is just so expensive compared to L5 miners, that they don't really create a worthwhile feedback loop, so you might as well just save up for the bargain.

2a. When solo, get two L6s, then move on.  (Or one and move on, but two seems to be slightly ahead).  There's enough of a feedback loop that this will double up your gas generation (From 512 to 1100 or so) before you're doing L7s, which my spreadsheet suggests is worth it.

2b. When teaming up and sharing resources, skip directly to L7s.  Two people together have 1024 gas income, which is too fast to get much of a boost out of a L6.

3. Get 3 L7s.  Regardless of how many people are teaming up; just...get 3 L7s.  (Getting two also works, and is almost tied but slightly worse).  This was a bit of a surprise to me; I'd expect it to at least vary by number of people sharing resources, but according to the spreadsheet, even if you're four people, you still want three L7s for the group pool before moving on to L8s.  Part of this is that the kind of gas income you generate off of L7s while you build up to L8s dwarfs the default gas production.  (Usually builds up to 7000 gas per second, roughly, before you get your first L8).

4. Max out on L8s, deleting as needed.


Interesting dynamic, anyhow--wasn't remotely obvious what the best way was to go pure econ.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #286 on: March 31, 2012, 04:10:12 AM »
Pobes vs Zeelot

So...new build tricks I hadn't thought of before.



1. Die.  If you're doing a multi-person team-up, have one person salvage all their stuff and die, then come back as a booster to boost mining.  According to my spreadsheet calculations, this is literally worth doing (and a huge advantage) as soon as you get the first L7 miner out, and worth doing even if you have only two people in your team-up.  There are downsides, of course--eventually, 20 minutes later, your teammate will hit the population cap.

EDIT: ok, I made a mistake in my theorycrafting: the Probe helpers actually only boost mining speed by 10%, instead of 50%, which is a lot less sexy.  Certainly you don't want to become a helper after just one L7--the loss of 512 gas income is not worth it.  Possibly by the time you perfect it's worth doing, though, as now you're losing maybe 5% of your team's gas income, and the mineral income is more important anyway.  But...it's only 10%; hard to get excited about.

2. Salvage your generator for a gas boost to get big miner milestones.  It's worth 25550, so...worth roughly the gas it would produce in 60 seconds.  This will generally only get you your first L7 miner 30 seconds earlier, granted, and you lose a bunch of gas production and will want to re-buy the generator, which will slow you down later, so you end up maaaaybe 10 seconds ahead, but you do seem to end up slightly ahead.
« Last Edit: March 31, 2012, 11:34:30 AM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #287 on: April 10, 2012, 04:46:38 AM »
Pobes vs Zeelot

Man, I'm playing this way too much.

Interesting strategies:

Entrapment: get a wall on both sides of the zealot, and then build turrets.  The zealot has a teleport, but it has a 3 minute cooldown, so you can either do this twice (once to force the teleport, and once to kill) or watch the zealot and see if it teleports.  This only really works earlygame.

Turning off auto-repair:  Auto-repair just takes away your control in a whole lot of situations, and can get you killed if there's nearby objects to be repaired outside of your wall.

Salvage and run: The zealot only econs by dealing damage.  Probes have blink and a lockdown effect, and can salvage all their buildings and hide somewhere else.  Very effective at denying econ to the zealot, probably as good or better than turrets.

Avoiding reveal: I think (but am not sure) that if you salvage your nexus and build a new one immediately, that you won't be without a nexus for long enough for the Zealot to get a minimap view of where the new nexus is building.  Would be good to test this....

On the Zealot end....

Stacking:
HP, Damage, and Regen all stack.  Armor and attack speed do not.

Zealot damage:
Obviously you care a lot about DPS, as that's your econ.  L1 damage is a 20% boost.  L1 weapon speed is a 20% boost.  The L2s are 40%.  The L3s are 80%.  But then we diverge--L4 speed is a 100% boost, and L4 damage is a 160% boost.  After that we go back to doubling both price and effect (so L5 damage is 320%, L5 speed is 200%).  Which means...most of the time we have the optimization problem of "max out x*(T-x)" where the goal is to make them both the same.  So...at what point does it become optimal to upgrade to L5 speed over L3 speed?  Is it ever optimal to get L4 speed?
The answer seems to be: L3 and L5 are break-even when you have 2900 minerals, so 1300 minerals on weapon damage, 1600 minerals on weapon speed.  L4 is never, ever optimal.  Basically, because L4 is so bad, you want to make the jump to L5 -early-, spending more on speed than on damage.
Well...what about the jump from L5 to L6?  Surely that will be symmetrical spending with weapon upgrades, right?  Actually....no.  The break-even point for those two is at 5100 minerals, which means 3200 spent on weapon speed, and 1900 spent on damage.  Two different factors at work here.  First, it's not a 1:1 ratio of money to effect--it's a 1:1 ratio of money to percentage bonus.  The ratio between a 200% bonus and a 400% bonus is actually 5:3.  Then again, the ratio between a 320% bonus and a 640% bonus is actually 1.76:1 (slightly better for the same cost jump).  But the big factor is that...because damage bonuses stack, damage is relatively continuous, whereas weapon speed is very quantized.

Zealot durability:
You have one unit.  You cannot lose that one unit or the entire team loses (not like probes; half the time when a probe dies I think "thank goodness").  Your choices for durability are HP, Regen, and damage reduction.  HP lets you survive the biggest beating.  Regen lets you survive indefinitely against some defences.  Damage reduction supports both the regen strategies and the HP strategies.  My general opinion for the first 50% of the game is "Screw regen, HP all the way, with maybe some token damage reduction for quadratic effects."  The thing is, you don't want to stay forever; you want to grind out some damage, and then return to home base and buy better gear (and you get to heal to full at home base).

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #288 on: April 16, 2012, 03:17:38 AM »
Magic the Gathering

So...every once in a while I do this mental exercise to figure out how much the power 9 would cost if reprinted today just with different mana costs.  Specifically, they must be tournament-viable at the new costs, but not overpowered to the point of banning.

Ancestral Recall (+4 mana)

This one is fairly easy, because it has basically already been done.  sup.  There was definitely a time when I wondered if 5 mana would be too powerful--because they were shying away from making instant-speed card drawing at all, and cards like Inspiration had been very powerful in the past to the point that they won't reprint them.  But...then Jace's Ingenuity got printed, and was...not bad, but not dominant.

Time Walk (+3 mana)

Done.  Not a whole lot to say about this one.  It saw tournament play, did some decent things, but was not dominant.  Add even one more mana (see Walk of the Aeons) and it's trash.

Moxes (+2 mana)

Ok, there's no exact parallel for these (which is a little surprising) but there is...

Mind Stone (Trades coloured mana for the ability to draw a card later when you don't need extra mana).
Diamond cycle (Same idea, except comes into play tapped, and was not considered all that good).
Prismatic Lens (Same idea, except can produce mana of any colour, but at an extra cost)
Talisman Cycle (Same idea, except it's an ice age pain land instead of a basic land)
Signet cycle (Same idea, except can't immediately tap for mana on turn 2, but can immediately tap for mana any turn after that...and produces two colours).

So...yeah, for all that they've never explicitly been done (which is weird), 2 mana seems like it would be correct.

Black Lotus (+3 mana)

Ok, so this one is kinda tricky.  The current status of ritual type effects is that...they're not doing them.  They belong to red now, but red gets like...Pyretic Ritual, which is actually too weak to get played (and thus would not pass my qualification test).  But at the same time, an artifact that did the same thing but significantly better for 2 mana would not be kosher.  Ok, so 2 mana's out.

So...what about 3 mana?  You put 3 mana in one turn, and get 3 mana out the next turn?  Well...there are some parallels for that.  Pentad Prism is a similar idea, but with 2 mana stored, and 2 mana received.  Coalition Relic is a 3 mana card that can produce 2 mana of any colour the turn after you play it (and keep on producing mana after that).  So...I -think- 3 would be acceptable.  Another strong argument for 3 is Lotus Bloom, which if you suspend it on the first turn, can be cracked for mana on the fourth turn.  Note that 3 mana would probably make it decently powerful (Pentad Prism is pretty powerful too).  But that's ok, we want it to be a tournament-playable card, right?

As for higher costs...there really aren't good parallels at 4 mana (maybe Everflowing Chalice kicked twice?)  At 5 mana Gilded Lotus exists (and was pretty much ignored).

I have one voice in my head suggesting it might be overpowered at 3 mana, and another voice saying that at 3 mana it's actually almost certainly worse than Lotus Bloom, the same way Ancestral Vision turned out to be better than Concentrate.  Shows up on about the same turn, but you end up paying less mana for it.  So...probably balanced at 3.

Timetwister (+0 mana???)

So...............................this one is going to be the hardest, and is the real reason I started thinking about these again.  A couple of years ago, they printed Time Reversal, which theoretically should answer my question for me, right?  Well...wrong, turns out Time Reversal sucks.  In two years of being legal, it hasn't shown up in a single tournament top 8 that I can find.

So...here's the thing about Time Reversal.  There's this little card called Tidings which is...not necessarily the best 5-mana draw spell ever printed, but it's serviceable, and tends to show up in tournaments.  So...for a 5 mana sorcery, that's the water mark you want to hit: equivalent of draw 4 cards.  Or: end up +3 cards from where you started.  In order to do this with Time Reversal, you need one of the following scenarios:

Your Hand: 1 card  ......... Opponent's Hand: 4 cards
Your Hand: 2 cards  ....... Opponent's Hand: 5 cards
Your Hand: 3 cards  ....... Opponent's Hand: 6 cards
Your Hand: 4 cards  ....... Opponent's Hand: 7 cards

So...you're relying on your opponent to have 4+ cards left in hand when you hit five mana (which, for a typical deck, is probably somewhere in the range of turn 5-7).  Which means if they play a land every turn, then they've cast an average of one spell every two turns.  Most decks just don't play like that.  And the few that do are probably going to counterspell you anyway.

And to add a little bit more fuel to the fire, they just revealed this card: Reforge the Soul which is a wheel of fortune that costs 5 most of the time, but can occasionally cost 2.

Now, let's disregard for the moment the part where "Timetwister is power 9 and Wheel of Fortune is not because Timetwister is blue and blue is the best colour!"  That doesn't really hold water anymore.  These days, the colours are pretty similar power, which means a card that does better when you can empty your hand is better in a colour that likes to empty its hand quickly (see: red).  Furthermore, the average blue deck runs cards like Ponder and Impulse which means that...even if said blue deck has the exact same number of cards in hand as its opponent, the average card quality in the blue hand would be better, so making both players toss their hand and redraw would generally be disadvantageous to the blue mage.

But can I really argue for 4 with a straight face?  I mean, isn't this the lesson taught to us by Diminishing Returns that draw 7s are just plain powerful?  I mean, that card still gets played in Legacy (type 1.5 for the old school in the audience).  But...here's the thing--that card gets powered off of dark ritual type effects; where if you draw 7, you can turn those 7 into mana right away.  That's not really possible anymore.

Ok, so 4 mana comparisons, I guess the card to look at would be Concentrate.  You'd need two fewer cards in hand, say, 2 cards in your hand, 4 cards in your opponents hand, but that sounds doable on turn 4.  Doable, but doesn't sound like "Pfft, blue decks would do that all the time", so you'd need a specialized deck.  This card still definitely won't be touched by the traditional blue control deck.  Maybe a combo or aggro deck with blue in it (and there's a fair few of those lately).

But I'm still not thrilled by it; I'm struggling to think of an existing deck that actually runs it (part of the problem there being that flashback is back, and quite important to blue right now, so blue doesn't want to shuffle its own graveyard away).  Well...what about 3 mana then?  Yes, I know that's the original cost; bear with me.  The relevant comparison here would be....I guess Divination?  Div's definitely on the weak end.  Thing is, to match div's +1 card advantage, you'd really only need one less card in hand than your opponent, which is very easy to achieve.  You can often pull off -1 or -2 cards in hand just by being first player; never mind the other variables.  And if you pull off -2 or more, then it's a huge swing for the cost (+2 total card advantage).

Ok, 3 CMC is maybe too inexpensive I guess.  Back to 4 CMC I suppose.  Would 4 CMC get played at all?  Hmm...I could maybe see it in certain kinds of decks; nothing current because of how much blue wants to keep its graveyard for flashback, but in a year or so sure.  Alternatively, would 3 CMC be completely busted and banworthy?  Well...Compulsive Research was never anywhere close to getting banned, and would be pretty close to "draw 3 cards" sometimes, and went into every kind of deck.  Jace Beleren is basically draw 3 cards, never even close to being banned.  Jace's Archivist is literally a timetwister style ability, repeatable, on a 3 mana creature, in standard, and has seen absolutely no play, and is dirt cheap.

Huh, ok.  I'm leaning back into the "maybe 3CMC is ok" camp.  Let's just quickly rule out 2CMC.  Night's Whisper is a very, very good card used in just about every deck type.  Matching the card advantage from it is pretty trivial (have 1 less card in hand than your opponent) and significantly outperforming it is common.  Yep, definitely no 2CMC.

But 3CMC I'm not sure; would be curious to see.  Probably make it cost more specific mana (1UU or UUU) just because of the whole "Doesn't benefit typical blue decks" (but would significantly benefit, say, a red burn deck that's just splashing for a quick hand refill).

Dark Holy Elf

  • DL
  • Denizen
  • *
  • Posts: 8135
  • Well-behaved women seldom make history
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #289 on: April 16, 2012, 03:33:01 AM »
I'm curious to see what you'd think of recosting other early edition low-cost, restricted cards such as Sol Ring, Fork, Demonic Tutor, Balance, and Regrowth. (Note that I'm pretty out of the loop and it's possible that some of those are reasonable enough to be brought back as is, depending on what the metagame looks like these days.)

Erwin Schrödinger will kill you like a cat in a box.
Maybe.

Yoshiken

  • Denizen
  • *
  • Posts: 2068
  • Yay!
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #290 on: April 16, 2012, 03:56:16 AM »
Fork effectively is back in two separate forms. Reverberate and Increasing Vengeance. Neither are particularly playable at the moment.

Balance is still banned in Legacy and restricted in Vintage, so that one would be interesting to look at.

Sol Ring.. Yeah, they've tried that a few times, so I suppose it could be good to look at? The closest available in Standard at the moment is Palladium Myr - a 3 mana creature.

Demonic Tutor is now Diabolic Tutor which has been in the last 2-3 core sets, at least, and has seen practically no play. At 4 mana, it's just too expensive, but at 3 mana, it would be too good.

Regrowth's an interesting one. Reclaim and a better version of it, Noxious Revival, are both available at the moment, which are effectively Regrowth for 1 mana instead of 2, but skip your next draw. And, yet again, they're completely unplayed, even as sideboard cards.


The one that amuses me, though, is looking at Time Walk. Avacyn Restored brings us a beautiful little card called Temporal Mastery which.. I can't see -not- being in a blue deck, especially given the mass of Delver of Secrets decks running around, and Ponder being back in Standard. Was Time Warp too weak for them? Maybe, maybe they just want something big to show off in Avacyn Restored, maybe they want to steal some of the power away from White at the moment. I'm interested to see how this one plays out, though.

Cmdr_King

  • Strong and Full of Love
  • DL
  • Denizen
  • *
  • Posts: 5567
  • Is Gay
    • View Profile
    • CK Blog
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #291 on: April 16, 2012, 04:02:22 AM »
Strikes me that the simplest thing to do to Balance is split it into different cards.  Though I wonder if they'd see use.
CK: She is the female you
Snow: Speaking of Sluts!

<NotMiki> I mean, we're talking life vs. liberty, with the pursuit of happiness providing color commentary.

Dark Holy Elf

  • DL
  • Denizen
  • *
  • Posts: 8135
  • Well-behaved women seldom make history
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #292 on: April 16, 2012, 04:24:07 AM »
Quote
Regrowth's an interesting one. Reclaim and a better version of it, Noxious Revival, are both available at the moment, which are effectively Regrowth for 1 mana instead of 2, but skip your next draw. And, yet again, they're completely unplayed, even as sideboard cards.

To be fair, not only is "skip your next draw" a pretty bad drawback, it's actually underselling the problem here... you don't get the card until your next draw action, which unless you have some way to get cards out of your library quickly means you (a) can't use the card you regrew this turn, and (b) telegraph a fair deal about your next turn.

EDIT: Oh, they're instants. Never mind about the telegraphing part, and the (a) drawback is reduced to a one-turn delay after acquiring Reclaim/etc.

Erwin Schrödinger will kill you like a cat in a box.
Maybe.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #293 on: April 16, 2012, 04:52:01 AM »
I'm curious to see what you'd think of recosting other early edition low-cost, restricted cards such as Sol Ring, Fork, Demonic Tutor, Balance, and Regrowth. (Note that I'm pretty out of the loop and it's possible that some of those are reasonable enough to be brought back as is, depending on what the metagame looks like these days.)

Sol Ring (+3 mana?)

Ok, first things first, let's take a moment to note that mana burn no longer exists.  Gone from the game.  Sol Ring got better due to that little rule change.


Now....this one's a pain in the ass, because it's right in between 3 and 4.

At 3 we get it with drawbacks or super conditionally on cards nobody plays like...

Palladium Myr
Myr Resevoir
Worn Powerstone (which...I don't even know, this might be considered too powerful now even though it comes into play tapped?  It's an Urza card after all.  EDIT: yeah, Worn Powerstone actually saw first-place finish tournament play in the Urza era in the same deck as stuff like Grim Monolith.  An outright upgrade is probably not kosher).

And at 3 mana, people will play the "add 1 mana with marginal upside" cards.  Take a look at Pristine Talisman--yeah, that card's been showing up in tournaments lately.  Yes really.

But at 4 mana...Everflowing Chalice makes it look really bad.  I guess Chalice was a top tier card, and this might be a marginal card at 4 mana, so that would be ok?



Fork (+0 mana)

So uhh...yeah reverberate, or if you just want to copy your own stuff, Increasing Vengeance has flashback.  I think Reverberate is used occasionally as a sideboard against counterspell decks?  It's definitely around, but not dominant.

Demonic Tutor (+1 mana)

So well, there's Diabolic Tutor but it's atrocious--it's been legal pretty much constantly for the past 10 years, and was only relevant about 10 years ago (in the Cabal Coffers deck which has approximately infinite mana anyhow).

Beseech the Queen actually gets played in normal decks, though.  And you can cast it for BBB.  Well...here's the thing, mind you.  Tutor effects for 3 are good.  Really quite good.  People used Dimir House Guard.  But nobody plays Diabolic tutor.  I don't know that a flat out format-defining tutor at 3 would be anything banworthy, though.  Good, and a bit format warping, sure, but I don't think it'd be banned.

Balance (+4?? ?? mana)

Uhhh.....well.....

Restore Balance didn't get played, but warning your opponent six turns in advance about this kind of effect...yeah, let's them avoid the worst of it.  And they tried to do Balancing Act at 4, but that was also kind-of overpowered.

Balance in particular is "cards in hand, creatures, and lands".  You can play around these.  Replace lands with artifacts (the Borderpost cycle pretty blatantly lets you swap out, in fact).  Play with flashback cards or reanimation cards so that you get advantages from discarding.  Use planeswalkers instead of creatures.

I'm going to say 6 mana, because people do play wrath of god/day of judgement variants at 6 mana, so it feels like it'd find a home at 6 mana.  But I'm not entirely sure if 6 mana is high enough.  Like I said, they've never successfully brought back anything quite like a newly costed Balance; I don't really have datapoints I can work with, other than "they failed at 4, even though the 4 mana one was harder to abuse".

Regrowth (+0 mana)

Note that there is an argument for +1.  Recollect got tournament play...although arguably only because of the cards around it; I don't think it would go anywhere in the current format.  Nature's Spiral exists and is weaker.  On the other hand, Nature's Spiral sees very minimal play, and Eternal Witness exists.  (And, honestly, in most decks Eternal Witness is a better card than Regrowth.  Yes, in Legacy/Vintage Type 1/Type 1.5 formats regrowth tends to be banned/restricted because of the Storm mechanic and dark ritual variants, but I don't think they're ever bringing the storm mechanic back in a new set, and dark ritual type spells are all but nonexistant these days).
« Last Edit: April 16, 2012, 05:05:00 AM by metroid composite »

SnowFire

  • DL
  • Denizen
  • *
  • Posts: 4938
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #294 on: April 17, 2012, 01:56:41 AM »
Revive exists, too, as far as Regrowth variants. ( http://gatherer.wizards.com/Pages/Card/Details.aspx?multiverseid=45433 ).  Different drawback than Nature's Spiral, still a nerf.  Eternal Witness is above the curve and saw play *everywhere*, inclined to think a Regrowth reprint would be similar - not format dominating, but a staple in Green decks.

Believe it or not, Pyretic Ritual sees play in Modern storm decks since Rite of Flame got banhammered.  2 Storm decks made Top 8 of hte most recent Modern Pro Tour, and had 7 Pyretic Rituals among them: http://www.wizards.com/Magic/Magazine/Article.aspx?x=mtg/daily/eventcoverage/gptur12/welcome#1

But yeah, Pyretic sees play strictly because Storm is one of those mechanics that becomes more and more insane with a large card pool, and they're just playing spells for the sake of playing spells.  Since Storm is never coming back, its use anywhere else is limited.  (And I'd honestly be pretty cool with banning more Storm cards, though I'm not even certain where to begin.  I don't play competitively but Storm is a classic example of a fishbowl deck.  There are hosers, yes, but you don't really interact with Storm, and Storm doesn't interact back.  Maybe just go to the source and ban Empty the Warrens, Grapeshot, and Dragonstorm?!)

Timetwister for 3 mana sounds right, although I suspect they'd want to cost it as UU1 now.  Timetwister is best when you can just dump your hand incredibly quickly, but don't care about the graveyard, so something like a Rakdos aggro strategy - a UU1 costing would make it harder for psychotically aggressive red decks to toss it in as a cheap way to refuel.

Balance is super-hard to balance because leaving out artifacts (and now, Planeswalkers) is just asking for players to cheat in deck construction.  It can easily be "6 mana: you win the game," but thanks to power creep, maybe WotC is okay with that now.  I'd gutcheck 6 mana might be okay in an environment where Balance was very hard to exploit, and 7 mana otherwise.  Of course, thanks to Planeswalkers generally being at least borderline playable and often even better, I suspect 7 mana is the safer choice.  Balancing Act is a better reprint, because it's much harder to "cheese"; Balancing Act decks had to play all Invasion sac-lands to guarantee hitting 0 permanents, which is a scary proposition - "sac all my lands, cast Balancing Act, oops you countered it" is kind of game over.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #295 on: April 17, 2012, 06:47:07 AM »
Eternal Witness is above the curve and saw play *everywhere*,

But it has also been banned nowhere.  (It's a Mirrodin card too--when they ban they like to ban a bunch of cards at the same time, which they did for Mirrodin; Witness was not one of them).  Even post-banning, I'm looking at, oh, US Nationals decklists from 2005 where only one deck is running Eternal Witness.  Well...maybe that's not fair, since in 2005 Japan was completely dominant.  Looking at Japan Nationals, there's three Tooth decks running Eternal Witness, but NONE of them are running it as a 4-of (2-of in two cases, 3-of in another).

Honestly, witness wasn't even one of those cards that was painful and format-warping, where it wasn't banned but everyone still breathed a sigh of relief when it left the format.  I'm looking at you, TarmogoyfTooth and Nail.  Just to name a couple of green ones.  I'll agree with calling it "above the curve" but depending on how you define "the curve", you tend to need to be above it to be tournament-relevant.

Quote
inclined to think a Regrowth reprint would be similar - not format dominating, but a staple in Green decks.

I'm skeptical, actually; If Revive and Nature's Spiral were mostly fringe, Regrowth probably would be as well.  I'm looking at Ghoulcaller's Chant as well here (note that Zombies are a currently viable deck; this card is still unplayed).  Your graveyard just isn't that great of a place to find stuff.

Quote
Believe it or not, Pyretic Ritual sees play in Modern storm decks since Rite of Flame got banhammered.  2 Storm decks made Top 8 of hte most recent Modern Pro Tour, and had 7 Pyretic Rituals among them: http://www.wizards.com/Magic/Magazine/Article.aspx?x=mtg/daily/eventcoverage/gptur12/welcome#1

But yeah, Pyretic sees play strictly because Storm is one of those mechanics that becomes more and more insane with a large card pool, and they're just playing spells for the sake of playing spells.  Since Storm is never coming back, its use anywhere else is limited.  (And I'd honestly be pretty cool with banning more Storm cards, though I'm not even certain where to begin.  I don't play competitively but Storm is a classic example of a fishbowl deck.  There are hosers, yes, but you don't really interact with Storm, and Storm doesn't interact back.  Maybe just go to the source and ban Empty the Warrens, Grapeshot, and Dragonstorm?!)

Yeah, I'm aware; just like Lava Spike is good in Legacy burn...but in Standard was a terrible card.

I'm really evaluating purely the power level in Standard here (otherwise Timetwister would not be 3 CMC, because yeah, that card is good in storm + ritual decks).

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #296 on: May 07, 2012, 06:40:53 PM »
Pobes vs Zeelot

Gold mineral patches:

Here's how they work (or so I've been told; seems to line up with observation, though).  When your miner mines, if it's on gold, it will get an extra +1 minerals.  This is regardless of the quantity that they mine (so perfect miners are 1296+1) but does depend on the frequency in which they mine.  What this means is that the first four miners are doubled in their output, and all higher miners are largely unaffected.  But this changes the cost-efficiency of the miners.

L1-L4: costs 2048 gas for every 1 min/sec
L5: costs 2925 gas for every 1 min/sec
L6: costs 2214 gas for every 1 min/sec
L7: costs 1208 gas for every 1 min/sec
L8: costs 771 gas for every 1 min/sec

Which is interesting; implies that it might be worth spamming some L4s after reaching max gen (but only when on gold) as they're both more granular (allowing the exponential feedback loop to kick in quicker) and more cost-efficient.

There is, however, the little issue of the population cap.  You can make at most 15 L4s (and realistically fewer; don't want to go over 13 if you plan on building mines, and if you have to blow up a lower level miner to get them, their efficiency drops, so maybe 10 at most).  10 Prof miners is...20 minerals per second income, which is about half that of one L6 (Ultra miners mine 36 minerals per second).

Still...seems probably worth-it, particularly since the whole point of getting ultras at all is to kickstart your gas income (as two ultras give you double to triple gas income by the time you get Legendary, which means you don't get Legendary that much slower, and get your second Legend faster as your gas is already flowing).  Advanced spam will let you start the gas flow a little earlier.  The downside, though, is that golds are often undefendable and taken as a ninja, so you could lose all that income.



Anyhow, so some more thoughts on gold miners:

2048 gas buys you ~1 min/sec income.
At the market, you can generally sell 10 minerals for ~200 gas (although it varies between 0 and 400).

Ergo, in an average market, it takes about 100 seconds for a miner to pay for itself.  100 seconds, coincidentally, tends to be roughly the time between most gen upgrades.  So...you end up in the same place for gas income, but with extra mineral income; a clear advantage.  There is an argument for buying pretty much all of your miners all at once, probably on the gen that takes the longest to upgrade (which I believe is gen 7, which requires getting both mega wall 1, and the wall before mega wall 1).
« Last Edit: May 07, 2012, 06:45:33 PM by metroid composite »

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #297 on: May 08, 2012, 07:11:03 PM »
I just realized that I made an assumption that wasn't true in prior zealot calculations.

I assumed that the first three levels of damage up bonus would be the same percentage bonus as the first three levels of attack speed bonus.  I.e. 100 minerals gets you a 20% bonus.  This is not the case; 100 minerals gets you a 40% bonus.  This changes the optimal damage setups significantly.

0-300 total minerals: no speed
400-500 total minerals: L1 speed
600-800 total minerals: L2 speed
900-3100 total minerals: L3 speed
3200-5300 total minerals: L5 speed
5400+ total minerals: L6 speed

Put another way...

L1 is optimal when you have 300-400 money sunk into just attack power
L2 is optimal when you have 400-600 money sunk into just attack power
L3 is optimal when you have 500-2700 money sunk into just attack power
L4 is never optimal
L5 is optimal when you have 1600-3700 money sunk into just attack power
L6 is optimal when you have 2400+ money sunk into just attack power

I also think it's quite reasonable to just skip the first couple weapon speed upgrades.  Get 500 money in attack, and then save for Wep 3.  At 500 money with no attack speed upgrade, the ratio is 15:15.6, so it's not that far off from optimal (4% below, but eliminating the buying and selling time honestly probably matters more).

On the high end...there's a nice overlap between the three, where you can build up to 2400 invested in just attack power, and then...swap out gloves, farm some more, swap out gloves, done.

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #298 on: May 10, 2012, 12:38:14 AM »
Pobes vs Zeelot

Wanted to get a sense of timing; this is just from a replay, where I went mostly pure eco, including mining from gold and salvaging wall; stopped for a couple of turrets and walls at various points, and had my gold miners killed, buying/selling at the market wasn't all that profitable, so it's not the most extreme eco game, but it gives a rough sense of things:

0:20: probes spawn
1:00: Zealot spawns
2:10: Gen 3
3:05: Gen 4
4:20: Gen 5
6:00: Gen 6
8:30: Gen 7
11:00: Gen 8
13:40: Gen 9
15:30: Gen 10
17:15: Ultra Miner
19:50: Ultra Miner x2
24:05: Legendary Miner
27:15: Legendary Miner x2
29:55: Legendary Miner x3
32:05: Perfect Miner
34:00: Perfect Miner x2
35:20: Perfect Miner x3
36:05: Perfect Miner x4
36:40: Perfect Miner x5
(etc, etc.  Not going to list all 15 perfects)
39:20: Final Wall (probably could have rushed it slightly earlier)

metroid composite

  • m_ACac
  • Administrator
  • Denizen
  • *
  • Posts: 4375
    • View Profile
Re: Theorycrafting! (Because I like competitive metagames too much)
« Reply #299 on: May 10, 2012, 08:42:42 PM »
Pobes vs Zeelot

Rough cost efficiency estimates.

For these, I assume that 10 minerals is worth 150 gas.  Obviously the relative value fluctuates throughout the game.

Turrets:
T1-T6: 16 gas = 1 damage
T7: 19.8 gas = 1 damage
T8: 21.6 gas = 1 damage
T9: 14.4 gas = 1 damage
T10: 16.8 gas = 1 damage
T11: 4.0 gas = 1 damage
T12: 5.8 gas = 1 damage
T13: 6.5 gas = 1 damage

So...interesting thought here: imagine you're only getting one T9.  If you instead spent all that money on T6s...the gas part of the T9 gets you 8 T6s, the cost of the Global Market gets you 2 more T6s, and selling the minerals for the T9 on the market gets you maybe 3-4 more T6s, so you're looking at 14 T6s for the same total resources.  Not immediately intuitive, but neat.  (In practice you'll often just have the minerals for a T9 anyway, though, and not have space for 14 T6s).

Among other things, though, I'd like to note that there need not be such a jump between T12 and T13--just build a crazy number of T12s.

Income:
automines: 1 gas/sec per 480 gas investment.
L1-L4 miners: 1 gas/sec per 273 gas investment.
L1-L4 miners on Gold: 1 gas/sec per 136 gas investment
L6 miner: 1 gas/sec per 152 gas investment
L7 miner: 1 gas/sec per 81 gas investment
L8 miner: 1 gas/sec per 51 gas investment
Generator upgrades....these are kind-of weird, because they have their actual cost, and then the prerequisite cost, but the prerequisite can be sold.
Generator with prerequisite cost: 1 gas/sec per ~160 gas investment
Generator without prerequisite cost: 1 gas/sec per 80 gas investment

Of course, all of this is dependent on the assumption that the gas to mineral conversion is 150:10.  It tends to drop lategame.  Automines and ultra miners tend to be tied if the market is around 85:10.  Automines and legendary miners tend to be tied if the market is around 60:10