Which stats affect true damage?

[scott-5496]

Yeah I am with you on this Blade - I try to level out stats where I can and that includes DEF and MR - though it is a long -term goal of mine for all characters - to have them all fully tallied and all having an even distribution amongst that stats.....pipe drem most likely but it is something!

[Blade]

Yeah I am with you on this Blade - I try to level out stats where I can and that includes DEF and MR - though it is a long -term goal of mine for all characters - to have them all fully tallied and all having an even distribution amongst that stats.....pipe drem most likely but it is something!

You're my man! Connected to this, it is always a "problem" for me that talli sets that are given as rewards all have the same subtype usually! This makes it hard to impossible to level out stats. Most of the times the devs take care and rotate between subtypes. But not always - the sustaining tallis from the last Gauntlet are an example for this, as it was again overwhelming while it "should" have been ward (or mystical) following the rotation approach.

[Mizrael]

actually you don't. If you deal a random hit, and each dmg type is equally likely, i.e. 1/3 atk, 1/3 mag, 1/3 true and if you have the choice of boosting mag/atk by x/4 each compared to boosting atk by 2x/4 and mag by 0, you'll deal the same damage in expectation:
case 1:
x/(4*3) + x/(4*3) + 2x/(4*3) = 4x/(4*3) = x
case 2:
2x/(4*3) + 0/3 + 2x/(4*3) = 4x/(4*3) = x

when your choice is x/4 on mag x/4 on atk but y/4 > 2x/4 on atk (or mag) you should do the latter since in expectation, your damage will be greater:
case 1:
x/(4*3) + x/(4*3) + 2x/(4*3) = 4x/(4*3) = x
case 2:
2y/(4*3) + 0/3 + 2y/(4*3) = 4y/(4*3) > 4x/(4*3) = x

Therefore random damage is an equally good type of damage as true, since it doesn't really matter what you boost, same as for true.

Although my school years are long gone and my only mathematical riddle now is $$$=food and clothing i can tell you one thing: This Sir is one mathematical nonsense!

An “x” can NEVER equal an “x divided and multiplied by something.

The only case this would be true is:
a) if you multiply and then divide by the same number. fe. Xx2/2=x. pretty unnecessary...

b) x=0 and we, super mathematicians, know that multiplying or dividing by 0 is impossible and strictly forbidden...

Also “4y/(4*3) > 4x/(4*3) = x” doesn’t make any sense...
You can’t use “is bigger than” and “equals” in the same mathematical sequence, because > is not a certain value. it is more a scale from X till infinity and that can’t equal ONE number.

I would keep it down to earth and take the side of Scott and Blade: it is best to even Atk and Magic out for both, true and random damage. especially for random damage, where the damage type gets chosen at random.

Lets make it simple (no multiplications or brackets): Atk=3 and Magic=1
The character deals 3 hits of random damage...If you hit 3x Physical damage you deal 9 damage. If you hit 3x Magical damage you deal 3 damage.

So you would be better off having both values 2 and deal both times 6 damage. as simple as that!

The exception is the rare case which Yup mentioned, where a character deals “true damage” PLUS physical or magical with another skill. then you naturally lean towards that type of damage.

@IronBrain: Next time open the in-game help section in the settings (top right corner). Your answer is there and none of us can tell you more about it. The “scaling” remains till these days a mystery...even for the Devs themselves!
You can test it yourself by equipping talismans with Atk, go do a battle in the cosmos (make sure there are no other buffs or debuffs present) and see how much true damage you deal. Then equip a set with the same amount of Magic and do the same battle again...

[Liu]

You're right with one observation, The result should be x/3 in each instance rather than x. This is because I wanted the output to be x rather than the input to be x, however I forgot to scale back up - the proper input should've been 3x/4 as then the output is x each time. It might have been more clearer had I used x as input - will repair later tonite and tag your post as reason for edit since you saw it, thanks.

b) x=0 and we, super mathematicians, know that multiplying or dividing by 0 is impossible and strictly forbidden...

Each x is in the numerator. Set it to 0 and the result is 0. You can always multiply with 0, only division by 0 is undefined.

Also “4y/(4*3) > 4x/(4*3) = x” doesn’t make any sense...

I'm assuming above that y/4 > 2x/4 which implies y > x since y/4 > 2x/4 <=> y > 2x > x.

You can’t use “is bigger than” and “equals” in the same mathematical sequence, because > is not a certain value. it is more a scale from X till infinity and that can’t equal ONE number.

">" is not a value, it's a relation. both "=" and ">" are relations. Of course you can use different relations in a sequence as long as you don't produce a contradiction:
5 + 2 = 7 > 5 > 2
works fine.
however:
5 + 2 = 7 > 2 > 5
is clearly a contradiction.
Also the term "uncertain" relates to the realization of a random variable, while in this case both x and y are definite variables. The random variable here is the realization of the type of damage for which I'm calculating the expected value.

Lets make it simple (no multiplications or brackets): Atk=3 and Magic=1
The character deals 3 hits of random damage...If you hit 3x Physical damage you deal 9 damage. If you hit 3x Magical damage you deal 3 damage.

So you would be better off having both values 2 and deal both times 6 damage. as simple as that!

This is clearly false. You can use my equations to show that (though mind the correction as mentioned above)
atk = 3
mag = 1
expected damage:
3/3 + 1/3 + 4/3 = 8/3
atk = 2
mag = 2
expected damage:
2/3 + 2/3 + 4/3 = 8/3

In both instances you produce the same damage in expectation. I admit I fucked up by not scaling x properly and possibly confused the heck out of everyone who reads this. It doesn't change the fact that (unless your OCD forces you) it doesn't matter which stat you scale as the expected values won't change. If you have a choice to boost, say atk by 5 and mag by 0, or atk and mag by 2 each, you should do the former since this would lead to a higher expected damage. You should deal with random damage the same was as you deal with true damage: you need to maximize the sum of atk and mag.

I hope I could clear all misunderstandings ?

[Mizrael]

You're right with one observation, The result should be x/3 in each instance rather than x. This is because I wanted the output to be x rather than the input to be x, however I forgot to scale back up - the proper input should've been 3x/4 as then the output is x each time. It might have been more clearer had I used x as input - will repair later tonite and tag your post as reason for edit since you saw it, thanks.



Each x is in the numerator. Set it to 0 and the result is 0. You can always multiply with 0, only division by 0 is undefined.



I'm assuming above that y/4 > 2x/4 which implies y > x since y/4 > 2x/4 <=> y > 2x > x.




">" is not a value, it's a relation. both "=" and ">" are relations. Of course you can use different relations in a sequence as long as you don't produce a contradiction:
5 + 2 = 7 > 5 > 2
works fine.
however:
5 + 2 = 7 > 2 > 5
is clearly a contradiction.
Also the term "uncertain" relates to the realization of a random variable, while in this case both x and y are definite variables. The random variable here is the realization of the type of damage for which I'm calculating the expected value.




This is clearly false. You can use my equations to show that (though mind the correction as mentioned above)
atk = 3
mag = 1
expected damage:
3/3 + 1/3 + 4/3 = 8/3
atk = 2
mag = 2
expected damage:
2/3 + 2/3 + 4/3 = 8/3

In both instances you produce the same damage in expectation. I admit I fucked up by not scaling x properly and possibly confused the heck out of everyone who reads this. It doesn't change the fact that (unless your OCD forces you) it doesn't matter which stat you scale as the expected values won't change. If you have a choice to boost, say atk by 5 and mag by 0, or atk and mag by 2 each, you should do the former since this would lead to a higher expected damage. You should deal with random damage the same was as you deal with true damage: you need to maximize the sum of atk and mag.

I hope I could clear all misunderstandings ?

I think the biggest problem with application of equations in LotB is that you forgot to implement one important factor- the RNGsus.

(I’m gonna use the simple language, which i find is easier to understand)

We need to split the conversation into two things:

True damage:
In this case it’s true that Atk=3 and Magic=1 provides the same results as Atk=2 and Magic=2. Even though you could “dissect” this into smaller, individual cases, where a certain buff/debuff could prove this wrong. but let’s keep it simple...

Random damage:
This is where RNG plays a big role in the right or wrong...
Let’s use a simple example: you flip the coin 100 times.
In case A (3+1) you only win against B below if you get the 3 more than 50 times. the extreme being 300 vs 100.
In case B (2+2) you always get 200.

And that’s where i choose to stick to the old phrase from back home: “A sparrow in the fist is more that a pigeon on the roof”...

Of course, this type of damage is even more sensitive to specific, individual scenarios, where the target wears a certain damage type shield or is immune to one kind of damage.only then you would choose the go with the 3+1 alternative.

And to your: “If you have a choice to boost, say atk by 5 and mag by 0, or atk and mag by 2 each, you should do the former since this would lead to a higher expected damage.”
I’ll let you do the math here, since you are so good at it (no equations needed, just simple straight forward reality):
you have 6 talisman slots.
the strongest single stat bosting talisman is a red fierce/safeguard with a value of 453.
A purple talisman boosts 2 stats by a combined value of 684.
How can i fill all the slots with ONLY red talismans to get a bigger value than 6 purples?

[Liu]

I think the biggest problem with application of equations in LotB is that you forgot to implement one important factor- the RNGsus.

Hence my above stated assumption of a uniform distribution. It may be that true damage is selected with a different probability than atk or mag, but it's VERY likely that both mag and atk are chosen with identical probability which renders my statement regarding expected values still true.

Random damage:
This is where RNG plays a big role in the right or wrong...
Let’s use a simple example: you flip the coin 100 times.
In case A (3+1) you only win against B below if you get the 3 more than 50 times. the extreme being 300 vs 100.
In case B (2+2) you always get 200.

If the coin is fair you will get 200 in expectation in both cases

And to your: “If you have a choice to boost, say atk by 5 and mag by 0, or atk and mag by 2 each, you should do the former since this would lead to a higher expected damage."

IF ... THEN ...


I say it again, it doesn't matter what you're boosting as in expectation it's the same. I'm not saying you should only boost attack, I'm also not saying you should even it out. I'm merely saying that evening out is not necessarily the best strategy, maximizing the sum of both atk and mag is the best strategy which I've said already:

You should deal with random damage the same was as you deal with true damage: you need to maximize the sum of atk and mag.

Whether you do that by evening the stats or by boosting one stat is irrelevant and I have never made a remark as to that. Of course maximizing the sum is only the best strategy if your objective is to maximize damage.

/e: I feel like this is going overboard. Nobody is gonna read this except you and me anyway. Thank you for hinting at a mistake, I've corrected this by now. However, if you would like to continue this discussion I suggest we move to either pm's or you join the lotb discord

[Mizrael]

I feel like this is going overboard. Nobody is gonna read this except you and me anyway. Thank you for hinting at a mistake, I've corrected this by now. However, if you would like to continue this discussion I suggest we move to either pm's or you join the lotb discord

I’m glad you said that mate! i was asking myself: “Hang on! what are we actually arguing about?”
I think we are both talking about the same, just in a different language...
So let’s close it with: The OP question has been answered! GL HF

PS: i’m not doing any FB, twitter, instagram (i have no idea what discord is, but i guess it’s something similar?).

[Liu]

I’m glad you said that mate! i was asking myself: “Hang on! what are we actually arguing about?”
I think we are both talking about the same, just in a different language...
So let’s close it with: The OP question has been answered! GL HF

PS: i’m not doing any FB, twitter, instagram (i have no idea what discord is, but i guess it’s something similar?).

Haha agreed

discord seems, at least to me, like a fusion of irc and teamspeak (if you know these) - a small and private chat room so to say. Just to give the topic a bump: here