Physical and Magic Critical Hit Rate Testing
Summary:
(1) No increase in critical rate from attacking from a particular direction (rear, front, etc.)
(2) Critical Rates for magic and physical damage follow similar formulas
(3) It takes roughly +30 "crit rate" to increase your actual crit % by +1% on R45. This return may vary with dLVL, but would likely only decrease with increasing dLVL.
(4) There is a cap on Crit Rate % at 20%. We did not assess for a Crit Rate % floor.
(5) Crit Rate for R45 was roughly 11.2%. ~6% for an R55 mob (limited trials).
Did not assess for crit rate on special attacks such as Steel Cyclone.
**********
Seiken's post on Lodestone official forums...
To expand on what we did. We started by having us both on GLA using R1 Weather Gladius for the low base damage. Seiken would pull and attack from the front then I would attack from the back. We then gathered 3,000 attacks each for a total of 6,000 trials between us. We did the attack from the front/back thing to assess for a change in critical rate; however, the numbers came back almost identical:
Based on this initial data set, we decided that attacking from different directions did not alter the critical hit rate. I then took one of my MATK data sets using R45 Dapper Cadavers in Copperbell Mines to compare physical and magic critical rate for the same dLVL mob.
Based on this comparison, we felt comfortable in saying that magic and physical crit rates follow a similar / same formula. This seemed likely as we had already verified that critical damage followed the same formulas for magic and physical critical damage bonuses.
Finally, we assessed for the strength of "+ critical rate" by using the same R45 Quartz Doblyns only adding "+62 critical rate"...
We chose to forgo more precise statistical measures like confidence intervals etc. at this point because by just eyeballing this result, we already knew critical rate + gear was extremely lackluster. This comes out to about 2.27% increase for +62, or about 0.036% per point (takes roughly +30 crit to add +1%). This was only tested on R45s, but the rate of return would likely be harsher for higher dLVLs if it varied with dLVL. Very sad outcome.
Finally, I would like to add that there appears to be a cap on crit rate at 20% just based on my MATK testing on R1 coblyns. I would also note that through parsing 300+ R55 Feretrar solos on CNJ, the crit rate for R55 is about 6% (53/894).
Sorry for the bad news
Lastly, special thanks as always to Tachikoma for the wonderful parser.