Yes, I know. I was also just wondering that If I only told 30% decrease it would have been misleading, that’s why I showed the calculations just in case. But yeah, it’s a coincidence that both are 30% in this case.
I spent 20 minutes convinced you were wrong. But you were correct.
Math is weird.
And that is not even getting into the INCREASED damage if the increase in defense stat results in a reduction of the nerf from the attack stat soft cap. This is one of the reasons it is important how the attack stat soft cap is calculated.
True. In fact, after helping with numerous posts on damage calculations, I was pleasantly surprised how much difference even a simple attack buff or defense debuff totally plays a big part. In fact, the basic fundamental principle here at play is that:
Ratio greater than 1, increases more when raised to positive power greater than one.
Ratio less than 1, decreases more when raised to same power.
Once players can grasp this, the damage output can be increased even without doing much calculations.
Typical example being reversing the Wilbur shared damage with a pulverizer, and then attacking the least defense hero with snipers so that max damage is shared. Simple, but wonderful things.
Yeah, the math is fine. Using the given formula it would yield a variation ( θ) of about 0,55 and that seem unlikely. There’s probably an attack soft cap in effect so the actual damage is much lower.
Edit: I tried adding the modifiers directly to the base stat and it would result in 900, which can be fine if we assume θ can be around 0,84.
Y’all need to quit pretending you know what that formula means… I’ve read it 5 times now and I still don’t know what damage is. If I have a hero and it says he does 240% damage… 240% of what?? A bunch of y’all saying thanks knowing you’re as lost, if not even more now, on what that % is… Now, if someone could put it where your average everyday working shlub could understand it that’s different…
Yes, it could be stated a whole lot more clearly – this “damage” is not the actual health loss the target suffers. Just using a different word might have spared many of us a whole lot of confusion.
But that’s the word SGG used … nothing we can do about it.
I don’t believe this is accurate. The number crunching I’ve been doing implies that they do actually mean “Damage.”
Consider the scenario presented by markpessan just above, and consider three formulae
The attack/defense modifiers being a coefficient on the base calculation, as markpessan ran them. This results in a significant overestimation of expected damage (80% higher than actual), with a fairly high extreme Theta (0.55).
The attack/defense modifiers being added together and applied directly to the stats in question, as Liliac ran them. This results in a smaller exaggeration, but still off: 18% high, with a Theta of 0.88.
The numbers being exactly what the text says they are (other than the hidden 20% boost to the AI’s defense). Those numbers would be as follows:
Richard: A = 813+234 = 1047 Attack
Thorne: D = 935-412 = 523 Defense
AI Defense Boost: D = 523 * 1.2 = 627
Special: S = 415% Damage
That’s an estimation of 8.96% higher than actual, for a Theta of 0.94.
So, on top of being a much smaller error, and a much smaller Theta (which we would expect to be more likely from a Gaussian distribution), this formula also trusts that what they’re telling us is accurate.
SGG is generally very precise in their descriptions, so why would they have some specials that say “attack” when they mean “attack” but others that say “damage” when they mean “attack”?
I am curious to know if what you stated might be able to arrive at a similar solution.
PS: For the Richard - Thorne case I just assumed it was one of the outlier cases in the Gaussian distribution with random parameter = 0.65 (which is still within the bounds of 0.606 and 1.649)
With Special As Damage Multiplier it would be
100 × 415% × (1.612 × 869 / (982×1.2))^1.35 = 461 (random parameter = 1.612).
That, admittedly, is more of an outlier than the “special modifies attack” calculation, but it is still within the e^±0.5 range
NB: I made a typo in the Richard/Thorne case, and propagated that error through; the attack bonus is 243 rather than 234, for a total of 1056 attack, and a Theta of 0.64. Not that that’s a huge difference, and still within the expected bounds, but I wanted to correct my error.
So for either of the formulas, one of these two cases is becoming an outlier. So we can’t be sure which one is the correct one, unless we do a few more. I will try it when I have some free time. Will probably look into 10 different cases and try to come up with the damage calculation using the corresponding formulas and we can tally the results here after a few days I guess.
That’s been on my backburner for a while, now, because I’ve been talking about this with Bloodfenraptor on the subreddit for the game. So far, I have 39 data points logged in my spreadsheet, including the Leo/Kun-chan and Richard/Thorne above. Most are attacks from Kageburado, Gravemaker, Rana, Seshat, or Joon, because those are who I raid with.
…I really should do more with Joon, given that he has the highest special modifier of my heroes, and should therefore display the difference most clearly.
The following is a histogram of results of the Ln(θ), where Blue is “Special Modifies Damage” and Red is “Special Modifies Attack.” You will note that while neither (yet) makes a smooth bell curve, the Red series (“Modifies Attack”) has a strong skew towards low θs, while the Blue series (“Modifies Damage”) is much more balanced.
This looks like a very interesting project. Cool, let me know if you want me to supply data points too. We really need a large number of data to find any kind of statistical significance.