Curse

anti_z3r0 · Nov 28, 2005, 08:46 AM // 08:46

Well, for my term project in STAT 360, I decided to gather data regarding the block/evade rate on Ward Against Melee and Aegis in Guild Wars. I wanted to do some hypothesis testing to verify that the block/evade rates were indeed 50% as stated in the skill descriptions. (This also seemed to ease my mind as a warrior who gets very frustrated and thinks that the rates are way higher than 50% whenever I try to adrenaline spike) I went to two isolated areas, one being right outside Ice Tooth Cave and one right outside Frontier Gate. I got a single minotaur/grawl since neither of them use any special attacks such as Irresistable Blow, and they have the attack speed of a hammer, which makes it much easier to follow and record data. For WaM I had 199 samples, and for Aegis I had 134 samples. That proved to be enough to accurately judge the evade/block rates.

The sample mean of the block rates for both were as follows:
Aegis: 54.477%
WaM: 48.744%

After performing some hypothesis testing, I found that both Aegis and WaM both did not show enough evidence to reject the null hypothesis of the block/evade rate being 50%. Therefore it can be said that both skills do indeed have a 50% block/evade rate, as the description suggests.

Here's the calculations and process of hypothesis testing for the two if anyone cares to look over it. I used a significance level of 99% for thet testing, for those that care.

Aegis

Ward Against Melee

Hope at least someone found this interesting. Now whenever you feel like WaM or Aegis is broken in either direction, you will know that it really isn't!

Q.E.D.

Allmightybob · Nov 28, 2005, 08:53 AM // 08:53

Was it ever under debate that the block rates were anything but 50%?

Racthoh · Nov 28, 2005, 08:58 AM // 08:58

I know I've always questioned all skills involving a %. When you don't gain any energy during the full duration of a Bonetti's Defense, I know I was contemplating doing something like this.

For testing purposes, I would think the sample data should be the same.