can you explain more in depth?
I get 13% just like OP using 6.602×100/7.580 which comes out to 87.098 in other words, 6.602 is 87% of 7.580 (6.602 being a 13% increase in the speed at which the swordsman attacks?)
what exactly is wrong with that?
Ok, I’ll use this hypothetical scenario to show you the problem. Imagine that without a trait, a phantasm attacks once per 10 seconds. With a trait, it attacks once per 1 seconds.
Under frifox’s calculations, this would be:
( 1 second per attack ) / ( 10 seconds per attack ) * 100% = 90%
A 90% increase is clearly wrong. The traited phantasm attacks once per second, meaning that for every time the untraited phantasm attacks, the traited one will have attacked 10 times. Therefore we should expect our result to be 1000%, indicating that the traited phantasm is ten times faster.
So in order to calculate correctly, we do
( 10 seconds per attack ) / (1 second per attack ) * 100% = 1000%.
If this seems backwards, it’s because we’re calculating in units of “seconds per attack” rather than “attacks per second”. If you convert each of the numbers into attacks per second instead, the calculation becomes what you expected.
( 10 seconds / attack ) is equivalent to ( 1 attack / 10 seconds) = 0.1 (attacks / second)
( 1 second / attack ) is equivalent to ( 1 attack / 1 second ) = 1 ( attacks/second )
( 1 attack / second ) / ( 0.1 attacks / second ) * 100% = 1000%
And we get the same result.
Correct.
Furthermore, if we assume phantasm CD really did go down by 20%, then we can calculate the duration of the animation and idle.
Don’t confuse delay reduction with attack speed increase.
If you have x attacks per y seconds, and you reduce y by 20%, you should expect:
x / ( (1 – 0.2) * y ) = 1.25 x / y
A 25% attack speed increase.
I have a set of ascended zerker gear in my invisible bag for pinging. And no one notices the difference.