Requesting Data for Parry RATE (not amount mitigated)

[Zfz]

Now to summarize all the data up to this point:

734 Parry is with Drk not WAR.

P% is Parry/Swings
P%-M is Parry/(Swings-Misses)

A few observations:

• Look at Disc's data for 456 parry: the parry rates differ from 4.79% at 376 samples to 6.77% at 561 samples, then to 8.21% at 4005 samples. Seeing also how others have also commented that the parry rates don't fluctuate much after about 1000 samples, I think this is evidence enough that anything less than 1000 samples are not reliable and should not be considered for analysis.
• Look at the miss rate, there is a big discrepancy. For example Disc's results on Tsanahale showed 7.85% miss out of 6532 swings, as well as 9.34% miss out of 4005 swings. Unless there are some known reason for this, it may suggest that we should be looking at sample sizes of at least 10,000 swings.
• Habien's DRK parry is consistently higher than WAR and PLD results. With 354 parry giving about 10% (double that of Disc's shieldless PLD) and 468 parry giving 13.02% (more than Hundred's WAR at 514 parry giving 11.72%). One plausible explanation is that Habien is doing the tests on a Lv58 Goblin Tinker, and Hundred is testing on a Lv60 Chimera.

In conclusion, I suggest that:

• All future test must have at least 1,000 swings.
• All future tests done on Lv60 targets
• Be more specific about the data, in particular the crits and exact enemy name+level