Lords Mobile Today: 0    Total Posts: 1583

Total:#3

Create Thread

[Guide] Analysing Squad Attack Data

[Copy link] 2/342

#1
Posted on 2018-05-14 14:54:01 | Show thread starter's posts only

Recently, I decided to test Squad Attack mechanics for myself. I'd seen a few others try it, but had some questions of my own.

For my first run of tests, I wanted to see single-hero damage profiles in order to try and work out what a few stats were really doing. I've inserted the Data Table below.
Tests were 10,000 vs. 10,000 T1 Ranged, No Leader, Both in Inf. P., No Defender Heroes, All Attacker Heroes Gold.

Null refers to the Control with no heroes used on either side.
Atkr. Damage:Control is the ratio between HERO and Control Net Atkr. Damage.


Atkr. HeroAtkr. CasualtiesDefr. CasualtiesNet Atkr. DamageAtkr. Damage:Control
Null928553246061
Trickster835647756421.23
Songstress752681460621.32
Lore840582049801.08
Rose908580048921.06



With this, we're able to see that in this scenario, Squad Atk. gave a 23% boost to our Net Damage, whereas 20% Army Atk. (Rose), w/o any Squad Atk, only garnered an additional 6% Net Damage compared to Control. 

We are also able to confirm our tests by noticing that Songstress:Lore warrants the same result as Trickster:Null (within margin of error).

---

Now, for some quick theory, we know that in Battle, our troops are split into four equally-sized 'Squads' per troop-type. These Squads are what give rise to Squad Atk. in Heroes: 
Squad Atk. is the Hero unleashing an 'Ultimate' of sorts that will inflict up to three-times the damage of Hero's entire Squad combined onto the Enemy at once.

As shown above, a Gold Squad Atk. Hero, without any troop boosts, is practically akin to having a fifth Squad on the field in terms of Net Damage Inflicted.

Or is it?

---

It's impractical to send only one Hero. My next tests were using a full, five-man, roster.

I was very interested in testing whether or not you were better off using all five /troop-type/ Heroes when attacking in full /troop-type/.

Currently accepted thought dictates only Four such Heroes are ever necessary: One for each Squad.
This means, in theory, we could send a full Range march with Lore Weaver or Berserker as our Fifth Hero, and get the exact same result as if we'd sent Songstress as our Fifth.

Comparing these Five-Hero tests to our previous Single-Hero runs could also give us some insight into how Stats stack within Lords. Are there diminishing returns, or is the curve Linear?

Tests were similar to above, but reduced to 5,000 vs. 10,000 T1 Ranged (otherwise Defender would have no survivors).

Base Heroes: Death Archer/Tracker/Black Crow/Songstress (shortened to A/T/C/S)
[Again: All Heroes Gold]
Atkr. HeroesAtkr. CasualtiesDefr. CasualtiesNet Atkr. CasualtiesAtkr. Damage:Control
Null1144288017361
A/T/C/S688586451762.98
A/T/C/S/Goblin676607654003.11
A/T/C/S/Lore640610454643.15
A/T/C/S/Watcher644610254583.14
A/T/C/S/Snow Queen619605754383.13
A/T/C/S/Trickster688586051722.98
A/T/C/Trickster/Lore688586051722.98
A/T/C/Berserker/Lore699547547762.75
A/T/C/Trickster740561648762.81
A/T/C/Berserker/Lore ** 540515946192.66

** => Spacers (4 Inf & 4 Cav) incl. in Atkr. Lineup


These Tests were not what I was expecting.
They seem to contradict themselves.

The first set of results to catch my eye was that of Watcher and Snow Queen.
Watcher is an Inf. Hero with +20% Army Atk. and +25% Army H.P.
Snow Queen is a Ranged Hero with +30% Ranged Atk. and +50% Ranged  H.P.
These tests are 100% Ranged (unless stated).
Yet, we see Watcher outperformed Queen in terms of Net Damage.
My first reaction is H.P.
We know it makes you lose less but also kill less. 
These results would seem to show 25% H.P. has more sway than 10% Atk.

Next, we can quickly look at 
A/T/C/S/Trickster
and
A/T/C/Trickster/Lore
to verify the 'only need Four /troop-type/ Heroes' when fighting full /troop-type/ assumption.


Now to 
A/T/C/S/Lore
and
A/T/C/Berserker/Lore
Using this, we can see how IGG stacks their troop boosts (somewhat).
Going back to my first dataset, we know Squad Atk. gives an ~+23% Boost when going from Zero-to-One Hero.
Using this new set, we can see the same for going from Three-to-Four Heroes.
Once we Normalise our results and compare, we see that this only gives a ~+15% Boost.


The results start to become a little more unusual once we begin to contrast
A/T/C/S
with
A/T/C/S/Trickster
First, what are the Stat. differences between these lineups?
The Latter has +200% Squad Atk., although we've already proven it's unnecessary.
Therefore, we'd imagine the two teams would Battle equally, right?
Well, IGG doesn't agree.
For some reason, the Trickster team inflicts less damage whilst suffering even more.
Yet oddly enough, the results of this Trickster team and
A/T/C/Trickster/Lore
are identical.
Watching the replay, Trickster is commanding a Squad in all instances; Army heroes are sitting out.
This leads me to believe that either Songstress has a somehow-superior Squad Atk. to Trickster (unlikely), or there's a game mechanic thats trying to normalise for Number of Allied Heroes in Battle - i.e. more stats for diminishing returns, as seen earlier (most likely in my mind).


Finally, we have
A/T/C/Berserker/Lore
and
A/T/C/Berserker/Lore **
I was expecting Spacers to do
1) Absolutely nothing; or
2) Provide marginal gains
Surprisingly, it was blatantly detrimental. More testing will need to be done to understand what's going on. 


If anyone has any questions, feel free to ask! Or if any have more data on why not all runs went to plan, I'd love to see it!

Signature
#2
Posted on 2018-05-15 00:04:04 | Show thread starter's posts only

Extremely interesting,  if you were to have sent only one hero how do you think the results may have varied or lead to an anomaly?

Signature
~Birds of feather stick together
#3
Posted on 2018-05-16 06:26:13 | Show thread starter's posts only

I'm gonna have to re-read this to understand the results when I wake up because for some reason, it's logically not computing properly in my head just out of either sheer tiredness or what... But my mind I'm pretty sure is about to be re-blown when I wake up in a bit.
This was interesting as f', I'm not gonna lie...

Signature