Board Thread:Game Discussion/@comment-84.110.72.169-20130907124036/@comment-11490363-20130907142445

I thought (hoped?) that defence was a straight soak, but that is not the case. My cursory testing reveals that.

Here are my results for a single battle with Autobot Jazz level 5

Def: 22K - Damage taken: 21K

Def: 15K - Damage taken: 27K

Def: 10K - Damage taken: 30K

As you can see, doubling your defence does not hale the damage taken, I want to gather a bunch of data points and see what the pattern is.

It could still be linear, but not 100%, i.e. for each 10 points in defence you decrease damage taken by 5 points (I thought it was 1:1 initially). Or it could be exponential with a power of less than 1. The range in the data I have is too small to tell which it is yet.

I will do more testing whey I have time. The problem is that each attack is a random variable so you need a lot of data to see the pattern properly. Like 50 attacks or something.

Part B though, the 200K damage is clearly superior based on this testing, the difference is just too huge.

You also have to consider the unit's HP. That will tell you how many attacks the unit would get through the course of the attack, and you can calculate the damage totals for both scenarios and that will show you which is better overall. In this case though 200K is better in all cases.

As an example though, if you had said 200K / 10K and 180K / 22K that would be much less obvious. Assume the unit's HP are 200K (The choice is actually arbitrary, you can use any number for HP and get basically the same results, though the larger the value you test with, the closer to reality it will be, choosing HP equal to the base attack of the Mob wold result in the most error, so HP of 10-20 times the mob base attack (my 20K - 30K above) is a good rule of thumb). Using my numbers above, you see that:

Number of attacks for 200K / 10K = 6 (200 / 30, rounded up), Damage = 6 * 200K = 1200K.

Number of attacks for 180K / 22K = 10 (200 / 21, rounded up), Damage = 10 * 180K = 1800K.

So the 20K drop in damage is more than compensated for by the more than doubling of defence since it allows the unit to have 66% more attacks!

Also, you can see that if you had dropped the unit's attack to 120K with the same difference in defence (raised to 22K), then the 2 case would be exactly the same (10 attacks of 120K is 1200K total damage), so there is a break even point that you want to stay above when dropping attack in exchange for defence. It looks like the greatest gains are in the low teens, so going from 30K-35K defence is not as useful as going from 10K-15K.

Again, once I get more data, I will have better rules to follow. I might even code up a simulator and deck recommender if the pattern is easy and obvious. Then we could maximize damage and not just attack power overall.