Well, I got the same results as you did during your "quick test" so I did some more test modding to figure out what was up.
Turns out the manual lies about two things. Firstly, unlike the manual says,
the defender loses a tied parry roll. Actually I had noted this in my first battery of tests but failed to re-read the spreadsheet when I crossposted it here (it was a month ago).
Secondly, and much more importantly,
a shield's -def/enc modifier is subtracted from the parry value before the parry value is put into the manual formula. So the correct defender's formula is actually:
2 + DRN + ((shield parry - shield enc) * 2) - (fatigue/20)
My previous tests had all been done using 0 -def/enc shields, which did not pick up this discrepancy.
Actually, the modding manual gives a hint that this happens when it says:
Quote:
|
Final parry value is always the #def value plus #enc value and the #enc value is subtracted from the full parry value to give effective #def value.
|
So, modding a #def 2 #enc 5 shield got me a shield with 7 stated parry, -5 def, 5 enc, and it performed identically to my 2 parry, 0 def, 0 enc shield in projectile parry tests.
Xbow or Normal made no difference to the parry rate in any scenario.
Kind of interesting here to note that this means the Bucker performs exactly as it says it should with an effective parry of 2, Shield/Hide Shield/Turtle Shell Shield has an effective parry of 3, Kite Shield an effective parry of 4, and Tower Shield/Ice Aegis an effective parry of 5.