Ok, there's rumors of a ninja edit to Guard. Decided I needed to test this out. Here's the setup.
Gear:
Black Adargas
Tiphia Sting
Tiger Mask
Guarding Torque
Guarding Earring
Spike Earring
Melee Cyclas
Melee Gloves
Rajas Ring
Woodsman Ring
Melee Cape
Warwolf Belt
Temple Hose
Melee Gaiters
Total:
+41 Guard [Skill total: 200]
-14 Evasion (decided to keep the +HP of Melee Gloves instead of -eva of kotes)
Food: Shallops Tropicale
Total defense w/Prot 3: 403
Target mobs: Mourioche (Boyahda Tree) [Lvl 62-68]
Guard skill to be EM with mobs (assuming target is A+ rating): 212-241
Mnk/Rdm for phalanx, assisted by lvl 59 healer NPC. Phalanx should be -8 damage with 109 enhancing skill (didn't bring torque).
Tracked all connected hit damage and noted guards.
Recieved exp per mob of: 27 (-11 lvls), 29 (-10 lvls), 31 (-9 lvls), 32 (-8 lvls), meaning mobs were levels 64-67. Guard skill EM range for those levels is 222-236.
Had to warp out after 18 mobs because I linked 3 with no chakra and NPC out of MP. Bah.
Summary:
Lvl 64: 56% -- 13 guards / 10 unguarded
Lvl 65: 30%-33% -- 17 guards / 37 unguarded / 2 uncertain
Lvl 66: 24% -- 10 guards / 32 unguarded
Lvl 67: 27% -- 39 guards / 102 unguarded
So as the mob level approaches current skill (assuming no level correction is involved, considering I'm 75), the guard rate goes up markedly. This is a far cry from Jinghles's tests of capped guard vs EM mob yielding about 8% guard rate. It does not seem unlikely that you could reach a 70%-75% guard rate vs EM for your skill. On the other hand, I didn't get a single skillup.
Thoughts on guard formula: It does not appear (to me) to be that guard is the traditionally believed -1.0 pDiff. With 400 defense, the mandys' cRatio should have been below 1.0 (in fact, 1.0 cRatio on steelshells is around 360 or so, from when I was doing some tests on pld, and they're 10 levels higher). If guard was a straight -pDiff, I should have been getting hit for 0 regularly (especially with phalanx on top of that). That was not the case. I was commonly getting hit for 3-5, but only hit 0 a couple times.
If phalanx is the last modification in the damage done, those 3-5 hits would have been 11-13 natively. I was hit by crits twice, 55 and 60. Taking the higher of those and removing the phalanx mod puts it at 68. Taking a quick average of non-guarded, non-crit, non-phalanx damage from that level of mob gives 27. Diff is 40, which should be the base dmg rating for the mob. 27 average on 40 base dmg weapon means a .675 cRatio, which would be an attack of 270 vs my 400 defense. That's reasonable, assuming 236 skill (lvl 67), 8 base, and 52 str (quite possible; not going to look it up).
11-13 out of 40 base would mean .275 to .325 pDiff. .275 would be -0.4 pDiff compared to cRatio. Thus I would posit that what Guard actually does is force the mob's attack to have the minimum pDiff possible for a given cRatio. This means it's not nearly as effective in damage reduction as previously thought (or possibly it was changed along with this edit; however I've seen results while farming that have also made me question the way guard worked, before the update). The possibility that it may now have an activation rate similar to how accuracy is calculated (75% vs EM?) makes it far more valuable, though. More testing is needed.
Note: if this is the case, some of the guards may end up being mistakes on my part in watching the animations, in which case the guard rate is probably lower than reported.
Also of interest - weaponskill damage. Didn't start using weaponskills til later in the run because I felt I was using up too much time and wanted to get the full 25 fight limit for the NPC.
Weaponskills done with no boosting and no gear swapping, just the above gear, usually near or at 100 TP.
Asuran Fists: 561, 462, 475, 509
Dragon Kick: 444, 586, 395
Howling Fist: 691, 682, 686
Howling Fist is doing surprisingly well. I'm curious to experiment with it on merit mobs.