[4.0pb] Balancing 6 and 4 ele builds against each other

[Cisz]

The problem:

After having played with jolin012 for a double no leak replay, I had the bad feeling that 6-ele builds are overpowered.

Here are my findings so far:

Analysis of 6 ele builds and 4 ele builds for theoretical and actual boni

Towers with relevant boni:

.

.    ldwfne                     Level 1                         Level 2                 Level 3

.# 1 ++     LD Trickery          #######                                

.  2 + +    LW Ice               15% st slow                    30% slow                60% slow

.#11   + +  WN Well              +25% speed                     +50% speed              +100% speed

. 13    ++  FN Flame             15% st slow; +5% dmg           30% slow/+10% dmg       60% slow/+20% dmg

.#14    + + FE Blacksmith        +25% dmg                       +50% dmg                +100% dmg

. 

. 

.    ldwfne

.# 5 + ++   LWF Windstorm        15% aoe slow                   30% aoe slow

.# 7 + +  + LWE Polar            -25% health / +12.5% dmg       

.# 8 +  ++  LFN Nova             15% aoe slow                   30% aoe slow

.#10 +   ++ LNE Enchantment      -5 armour st                   

.#11  +++   DWF Corrosion        -3 armour aoe                  

.#13  ++  + DWE Muck             15% aoe slow                   30% aoe slow

.#14  + ++  DFN Voodoo           +41% dmg                       +87.5% dmg

.#16  +  ++ DNE Roots            15% aoe slow                   30% aoe slow

I am not sure how to calculate voodoo yet. The 41%/87.5% are the higher numbers, 25%/50% is the lower case, but it can even do no effect at all, if the creeps die in under 2.5 seconds after they get the debuff.

Here are the calculations for the highest number:

Maledict level one:

With 100 dmg done in 5 secs, and 50 dmg done in 2.5 secs with you dmg towers, voodoo will add:

50 dmg + 12.5 bonus by voodoo = 62.5 total at 2.5 secs and

62.5 dmg (the dmg at 2.5 secs) + 50 dmg + 28.125 (new bonus) at 5 secs.

Thats a total of 140.625 or +40.625% on level one.

Maledict level two:

with 100 dmg done in 5 secs, and 50 dmg done in 2.5 secs with you dmg towers, voodoo will add:

50 dmg + 25 bonus by voodoo = 75 total at 2.5 secs and

75 dmg (the dmg at 2.5 secs) + 50 dmg + 62.5 (new bonus) at 5 secs.

Thats a total of 187.5 or +87.5% on level two.

Jolin012's build - all elements on level 1

Resulting in theoretical effetcs:

Slow - minimum speed (dunno the number yet)

Armour reduction - -8 = 49% dmg increase

Dmg increase on towers - +25%

Speed - +25%

Dmg increase on creeps - +5% flame; +12.5% polar; +41% voodoo

Armour is calculated like this:1.06^8 is 1.49, so dropping from 1.49 to 1.00 is like having +49% dmg.

The given effects all stack multiplying, so we get a resulting 378% dmg for a strong voodoo, or 335% for a weaker voodoo (only giving 25%).

So an all-elements-on-level-1 build does about 335% or 378% dmg and minimum speed.

Cisz's build - lwfn

Towers with relevant boni:

.    ldwfne                     Level 1                         Level 2                 Level 3

.  2 + +    LW Ice               15% st slow                    30% slow                60% slow

.#11   + +  WN Well              +25% speed                     +50% speed              +100% speed

. 13    ++  FN Flame             15% st slow; +5% dmg           30% slow/+10% dmg       60% slow/+20% dmg

. 

.    ldwfne

.# 5 + ++   LWF Windstorm        15% aoe slow                   30% aoe slow

.# 8 +  ++  LFN Nova             15% aoe slow                   30% aoe slow

We get a resulting 240% dmg and minimum speed.

Results

We get 335% and 240%, which is a difference of 40% (335%/240% = 1.40).

Even worse, jolin012 did not use more than 6 picks, so he could have gotten much more bang for his build. In addition to that, my build gave me a hard time to actually build the towers for those cool boni, as I didn't have the money to do so.

I suggest two changes:

I.) Buff level 2 and 3 duals to let the 4 element builds ctach up

II.) Decrease the cost and dmg of all multi element towers or at maybe just for those of higher level, but keep the dmg/gold ratio as it is now. But don't change the support abilities, so the new and cheaper corrosion would still do -3 armour. This will result in more towers with the same effective dmg, but easier access to the support effects.

I can't tell you proper numbers yet, but I will provide them soon enough. I expect that the waterfall/foundry level will have to go up to about +400% and that a maxed dual should cost only about 4.5k, while having less damage accordingly, but those are just my first guesses. MagicalHacker is working on the problem too, and will hopefully come up with suggetions for better boost numbers.

List of all tower for reference:

.    ldwfne 

.# 1 ++     LD Trickery          Duplicates towers

.  2 + +    LW Ice               Single target slow and dot (like poison in 3.0)

.  3 +  +   LF Electricity       Bouncing spell dmg (a bit like metal in 3.0)

.  4 +   +  LN Life              Gains life with kills (but slower life gain than in 3.0)

.  5 +    + LE Quantum           Can teleport itself

.  6  ++    DW Poison            Dmg in a line (like sun in 3.0)

.  7  + +   DF Magic             ???

.  8  +  +  DN Death             Does more dmg if the target is on full health

.  9  +   + DE Gunpowder         Long range tower

. 10   ++   WF Kindle            Aoe dot (a bit like steam in 3.0)

.#11   + +  WN Well              Speeds up other towers

. 12   +  + WE Hydro             Sometimes throws a unit in the air, dealing splash when it lands

. 13    ++  FN Flame             Single target slow and increases dmg taken

.#14    + + FE Blacksmith        Gives your towers a boost on dmg

. 15     ++ NE Rage              Can cast rage on itself for some exta fire power

. 

. 

.    ldwfne

.  1 +++    LDW Hail             Attacks some random area near it with splash

.  2 ++ +   LDF Jinx             Warps creeps back on the path and damages them

.  3 ++  +  LDN Oblivion         Summons skeletons

.  4 ++   + LDE Laser            (Placeholder till prism tower is coded)

.# 5 + ++   LWF Windstorm        Creates moving aoe slowing and dot hurricane

.  6 + + +  LWN Tidal            Charges an aoe attack for every 10 attacks, can be discharged by the player

.# 7 + +  + LWE Polar            Slows  target and causes extra dmg

.# 8 +  ++  LFN Nova             Dmg and slow to all creeps in range

.  9 +  + + LFE Gold             Gives bonus gold based on inflicted dmg

.#10 +   ++ LNE Enchantment      Decreases creeps armour, dot

.#11  +++   DWF Corrosion        Dot and armour reduction debuff, aplied with a splash

. 12  ++ +  DWN Drowning         Chance for instant kill based on targets health

.#13  ++  + DWE Muck             Splash attack and aoe slow with that (much like ice in 3.0)

.#14  + ++  DFN Voodoo           Units get extra dmg based on dmg taken since cursed

. 15  + + + DFE Flamethrower     Rapid fire splash tower with incinerate

.#16  +  ++ DNE Roots            Slows all creeps in a line

. 17   +++  WFN Impulse          Does more damage the further away the target is

. 18   ++ + WFE Zeal             Gets faster and faster the longer it fires w/o pause

. 19   + ++ WNE Beef Golem       Grows and grows while dealing dmg

. 20    +++ FNE Quaker           Has a chance to do splash dmg

. 

. 

.# - One of the stronger support towers

[Twilice]

Yes, using all 6 elements seems much stronger then using 1-5 Elements. I have watched all 3 new replays for the ronald hunt and everyone of them used all 6 elements. I also noticed that there is not much duals used exept supporters (well, blacksmith, trickery.)

Mostly it were duals at the very early begining, and were later uppgraded to tripple.

My conclusion:

6 Elements need nerf

Duals need buff

?somehow make it more balanced?

The question is how to balance these things.

Make tripple slows less effect the more of them you have? (of diferent type, not same type)

[Cisz]

Good news: I have found issues.

Bad news: There are issues.

This is the second version, my first attemp had errors in properly representing the armour reduction (counted in to strong) and ranked polar to low. The armour reduction is now corect, and polar is counted as haviong 25%/50% dmg increase, alltough that's a simplification.

Version 1.1

I enlarged my build explorer worksheet to a huge mass of zeroes and ones and finally got the first section done.

I may have made mistakes, so if any build is misrepresented, please correct me.

My results assume that trickery would work as a well or blacksmith, which is not yet the case. If anyone can provide me with a properly calculated damage increase equivalent for the three levels of trickery, I'll use that numbers instead. I also assumed the lower damage increase for voodoo (25%/50%)

I don't differ between single target damage increasers (like enchantments armour reduction or flames sunburn) and aoe damage increasers (like corrosions spew acid or voodoos maledict), which might be problematic. In most cases I would expect that two single target debuffers can more or less replace one aoe debuffer, so that sounds reasonable. But I do differ between st and aoe slow, as they are very different in their clumping.

I havn't taken into consideration if the towers needed to actually gain the boni for your game are affordable for the player. I just pretend that "available" equals "affordable". I expect this to be very untrue for the higher level towers/effects, and will check this problem later on. Until them these results are only "theoretical rankings".

Theoretical build ranking for 4 ele builds, 4.0pb (With trickery faked and weaker voodoo.) Version 1.1

DmgInc = Damage Increase (given as a factor, 2.79 = +179% dmg, or 279% total damage)


AOEslw = Area of effect (splash) slowing power, given in % speed reduction.


STslw = Single target slowing power, given in % speed reduction.



DmgInc  AOEslw  STslw      l d w f n e    


2,79    30      60    ldwf 3 3 3 2       1


3       0       60    ldwn 3 3 3   2     2

3       0       60         3 2 3   3     


3       30      60    ldwe 3 3 3     2   3


3,3     30      30    ldfn 3 3   3 2     4

3,3     30      30         3 3   2 3     

2,7     30      60         3 2   3 3     

2,7     30      60         2 3   3 3     


3       0       0     ldfe 3 3   3   2   5

3       0       0          3 3   2   3   

3       0       0          3 2   3   3   

3       0       0          2 3   3   3   


3,8     30      0     ldne 3 3     3 2   6

3,8     30      0          3 3     2 3   


2,4     60      66    lwfn 2   3 3 3     7


2,25    30      60    lwfe 3   3 3   2   8

2,25    30      60         3   3 2   3   

3       30      30         3   2 3   3   

3       30      30         2   3 3   3   


5,7     0       60    lwne 3   3   3 2   9


4,56    30      60    lfne 2     3 3 3   10


5,03    0       60    dwfn   2 3 3 3     11


2,79    30      0     dwfe   3 2 3   3   12

2,79    30      0            2 3 3   3   


2       60      0     dwne   3 3   3 2   13

2       60      0            2 3   3 3   


3,6     30      60    dfne   2   3 3 3   14


3,6     0       60    wfne     3 3 3 2   15

3,6     0       60             2 3 3 3

Here are the losers:

Build 5 - ldfe: Only 300% dmg, no slow whatsoever. It only has blacksmith and trickery.

Build 2 - ldwn: Only 300% dmg, 60% single target slow.

Build 4 - ldfn: 30% aoe slow and either 270% dmg, 60% st slow or 330% dmg but only 30% st slow.

Build 8 - lwfe: 30% aoe slow and either 225% dmg, 60% st slow or 300% dmg but only 30% st slow.

Build 12 - dwfe: 279% dmg, 30% aoe slow.

Build 13 - dwne: 60% aoe slow, but only 200% dmg.

And here are the winners:

Build 9 - lwne: 570% dmg, 60% single target slow.

Build 10 - lfne: 456% dmg, 30% aoe slow, 60% single target slow

Build 7 - lwfn: 240% dmg, 60% aoe slow, 66% single target slow

It's hard for me to ballance slowing against damage increase, but even for the same slowing power there are good and bad builds. If you go dwne instead of lwfn, you drop from 240% to 200% damage (and loose your st slow). If you go dwfe instead of lfne you drop from 456% dmg to 279% and loose your st slow. Playing ldfe is simply stupid, or brave, or both.

Note that all builds with flame and polar are doing really, as give a st slow and a dmg increase. (Polar always contains ice.) Only one build with either of them is in the loser category: ldfn. That build highly depends on a (hopefully working) trickery, and you can't max trickery and flame (They share no element). All winner builds have one of the two.

Note also that the three builds that have no strong triples (but two of the strong duals) do especially bad. The ldfe and ldwn are in the looser category and wfne is only saved by flame, without its damage increase it would be as bad as ldwn. The strong duals are not strong enough. MagicalHacker predicted this.

As far as I can say, we are one single target slow short, and it should be in ldfe. Since fn and lw are st slows, de would be a logical choice, so they share no elements.

Once I am done with my table, we can enter the dmg increase, armour reduction, and slowing power for every tower and get a power ranking for all builds in the same second. Once I managed to learn the proper commands, I will probably even be able to let open office calc do the searching for overpowered and underpowered builds.

Coming up: 5 and 6 ele builds.

[Cisz]

5 Element builds

Same rules as before.

DmgInc = Damage Increase (given as a factor, 2.79 = +179% dmg, or 279% total damage)


AOEslw = Area of effect (splash) slowing power, given in % speed reduction.


STslw = Single target slowing power, given in % speed reduction.



DmgInc  AOEslow STslow  Build  l d w f n e    Nr.

6,29    60      66      ldwfn  1 1 3 3 3      1

6,92    60      60             2 2 3 1 3      


5,24    60      60      ldwfe  3 3 3 1   1    2

6,29    60      30             3 3 1 2   2       


8,55    60      30      ldwne  2 2 3   3 1    3

8,55    60      30             3 3 2   2 1       


8,55    60      60      ldfne  1 1   3 3 3    4

9,41    60      30             2 2   3 1 3       


9,41    60      60      lwfne  1   3 2 3 2    5


6,92    60      30      dwfne    1 2 3 2 3    6

6,92    60      30               1 3 2 3 2

All have 60% aoe slow and at least 30% st slow.

Worst is no light, only 692% damage, 30% st slow.

Best is no dark, 941% with 60% st slow, closely followed by no water, 941% too, but only 30% st slow.

Even the worst 5 element build is painfully good when compared to the 4 element builds.

Edit

6 Element builds

DmgInc  AOEslow STslow  l d w f n e

14,93   66      60      1 1 3 1 3 2

14,93   66      60      2 1 3 1 3 1

Now 6 ele builds even go up to 1493%, full slow ahead.

[0rb3r]

At the moment I've just found 3 ideas to balance:

1) Increasing the rate at which duals gain side effects from x2 to x3

2) Limiting the capability of towers to stack

3) Making all multielement requiring towers more expensive (+66%) but reducing their cost after owning already one on the field (iex: 1st 166%, 2nd 120% 3rd 80%, 4th 66%, 50% for all remaining).

+) Nerfing Periodic Element tower

I will edit the post if once more ideas come up.

[jolin012]

just a random fact - builds containing both trickery and blacksmith or well are weaker, because they don't stack well as trickery is imbalanced atm. so for for example 6ele builds better use trickery, (except for lamer ppl like me who do just well+smith). another reason to make the new trickery. Waagh..

[Cisz]

Update

After removing some mayor errors in my table, here are the values for eletd 4.0 again:

4 Element builds - 30.63 dmg/gold/sec +/- 9.65

5 Element builds - 53.81 dmg/gold/sec +/- 15.82

6 Element builds - 132.86 dmg/gold/sec (54.67 +/- 18.22)

The +/- bit is the standard derivation (and not the total derivation). I have taken the best build for each element combo and gave the average of those best builds for each number of elements. That is a drastic reduction of possibilities in the case of 6 elements.

I am working on a new distribution of the dmg/gold and dmg increase. Here are some very unfinished numbers:

Rough draft, anything but finished - handle with caution

Those values are damage/gold/second, in a unit called "u", and thats the dmg/gold/sec of an amplified single element tower.

Amplified single - 1u

Focused single - 1.5u

Refined single - 2u

Duals - 1.5u/2u/4u

Triples - 1.5u/2u

Dmg inc for the three strong duals: 24/51/275%

Dmg inc for the eight strong triples (pretending the net effect of the aoe slows was comparable to a dmg inc): 10/35%

Result:

4 Element builds - 26.71u dmg/gold/sec (27.34 or 22.65)

5 Element builds - 27.23u dmg/gold/sec +/- 0u

6 Element builds - 27.5u dmg/gold/sec (18.14u +/- 4.12u)

Those values are really close, and the derivation for 5 ele builds is zero. But 4 ele builds have 22.65 for builds with two strong duals or 27.34 for one strong dual, and thats still a big difference (factor 1.2).

What I believe to be true so far: It's a bad idea to make any tower better at dmg than duals, as that greatly improves 5/6 ele builds. If I make boosting triples stronger, boosting duals get really low at lv one and really strong when maxed.

If you feel bad about the suggested numbers, let me know what you'd like to be changed, and I'll try to make that work, or at least tell you why I failed with that.

[0rb3r]

So, basically what you did on:

After removing some mayor errors in my table, here are the values for eletd 4.0 again:

4 Element builds - 30.63 dmg/gold/sec +/- 9.65

5 Element builds - 53.81 dmg/gold/sec +/- 15.82

6 Element builds - 132.86 dmg/gold/sec (54.67 +/- 18.22)

Is an average and standart derivation of all possible builds of "u"(dam/sec//gold) using known damage output and cost. Am I right?

What I don't get is how you manage to know the dam/sec, are you ignoring range and location?

Just being horribly curious about it.

[Cisz]

So, basically what you did on:

After removing some mayor errors in my table, here are the values for eletd 4.0 again:

4 Element builds - 30.63 dmg/gold/sec +/- 9.65

5 Element builds - 53.81 dmg/gold/sec +/- 15.82

6 Element builds - 132.86 dmg/gold/sec (54.67 +/- 18.22)

Is an average and standart derivation of all possible builds of "u"(dam/sec//gold) using known damage output and cost. Am I right?

No. Those are absolute values, not in "u", and they are representing averages of the best sollutions for each element combo. I calculated the highest dmg/gold/sec (that is the best available boostable dmg tower, boosted by all boosts, and hitting a maximum debuffed creep) for all possible ways to spend 11 lumber (not counting interest or essence). I divided those builds into groups, based upon what elements they contain. For each such group I used the maximum dmg/gold/sec (that is I chose the variation with the highest resulting dmg/gold/sec), assuming that a player going lfne will know how to properly play that. Next I took those maximums and calculated the average maximum dmg/gold/sec for 4, 5 and 6 element builds. (In case of 6 elements, there is only one best build, and no average best build can be calculated, in brackets I gave the average of all 6 ele builds instead). Oh, and I ignored pures.

What I don't get is how you manage to know the dam/sec, are you ignoring range and location?

Again I assumed that a player knows how to properly place and micro a tower. I totally ignored the range issue. My reasoning for this is that personally I don't see a future for long range in this map. And the values are given on the homepage, I just had to type a bit to get them all per second. Splash is a problem here. I made up some factors to make it look like I properly factored in splash. Most special abilities can be transformed into dmg/gold/sec. And I only looked at boostable dmg, so I didnt even type in the dmg values for well or blacksmith.

There are a lot of errors and assumptions in this, but all improvements I could possibly make won't change anything: The ballance as it is sucks.

When calculating new values for a better ballance, I replaced all support towers with one of two versions (for two and three elements), simply giving a dmg increase. I even did that to the aoe slows. And likewise I replaced all single, dual and triple dmg towers with placeholder singles/duals/triples that have easy stats. So once we decide on a new set of basic numbers, the next task is to get all towers equal to their dummy representation. Like making sure that polar and corrosion give the same net effect, or making earth and light beeing equally strong.

[jolin012]

Also - to make 4ele stronger than it allready is, the duals tier 3 needs to be worth to use more. Here is a problem. (Although with pures as plan - you have to rebuild at some point for a 25% loss. or begin with singleele as dmg.)Pures are almost as strong as dual tier 3 and the networth is usually so low that all cash on dmg towers is on pures. I mean the 5ele build "33221"(2 eles at tier 3. 2 at tier 2 and one at tier 1.) is as strong in damage as "3332". because they can still spend all cash after supporters on damage towers with >10dmg/gold/sec even without a third tier 3 elemental. There are several ways to fix this. Make networth higher. Make tower cost lower. Make fewer/none pure essences for free. Make pures weaker. make other damage towers stronger. etc, etc.

[Karawasa]

I have been playing around with the table, and present my first attempt:

Round 1 - Hagle Time

Amplified single - 1u

Focused single - 1.5u

Refined single - 2u

Duals - 1.5u/3u/6u

Triples - 2u/3u

Damage increase for the three super duals: 25/50/200%

Damage increase for the eight super triples: 10/25%

Result:

4 Element builds - 26.97u damage/gold/second +/- 0.4u

5 Element builds - 32.94u damage/gold/second +/- 0u

6 Element builds - 30.15u damage/gold/second +/- 5.25u

Analysis:

This distribution of power reduces the explosiveness of support growth (i.e. 275% and 35%). It also makes triples more powerful versus single element towers, while at the same time maintaining the relationship between duals and triples (for the most part). Finally, it significantly reduces the deviation for 4 Elements, in return for marginally increasing the deviation for 6 Elements (thus stabilizing the standard build).

The downside is that the builds are not as balanced. 4 Element is still the weakest, followed by 6 Element (interestingly enough). Compared to the values that were produced for the current state of things, this minor imbalance seems acceptable.

[Cisz]

The main problem here is that you basically removed the suport triples. Hm.

[jolin012]

yeah, the risk is that they are not worth to build becayse they also cost alot compared to their %. including that they (sup triples) don't get boosted by duals.

[Karawasa]

I think the jump from 10 to 35 is simply too large (i.e. it doesn't feel right). The numbers do seem low for the triples, but I don't see a way around that. We all know that there are 8 triples compared to 3 duals, so the ability to stack them with a 5 or 6 Element build needs to not be overpowered.

Edit: Raising level 2 triple support from 25% to 30% doesn't hurt things much.

[0rb3r]

I think the jump from 10 to 35 is simply too large (i.e. it doesn't feel right).

I do feel that a maxed out single, dual or triple tower, should be something special, at least, you do require or 6 picks and/or a pure essence.

[Karawasa]

I feel 10%/30% works the best. It is a good compromise between 25% and 35%. Also, it makes more sense, in an odd way perhaps. Duals go like 2x/4x, so it seems to follow that triples should go 3x.

[0rb3r]

Yeah, sounds good, but it will probably mean that the step between lvl2 and lvl3 duals will have to be more then x4. Well, we will see how it turns out.

edited right away, duals, not triples. Thanx Cisz.

[Cisz]

There are no lv 3 triples.

[Karawasa]

For all those who don't have it, here is the table.

[ImmolatusBurn]

Yay a giant table of cryptic data... <3

[m0rph1ing]

I'm wondering, is this table balanced for all the different builds yet?

[Kaini]

Nice table.

Ever tryed 'Conditiornam Formatting'? Would make it much easier to read. (And stick the tower names). But the AOE/ST-Slows columns don't work. Now I understand the usage of the formular elplorer in OOo.