WAR Dex and Parry testing!

Quote:

---

Originally Posted by Hachiko

The only reason it looks higher is because you cant simultaneously parry and block. The game, in essence, Checks for misses first, then for block, then checks for parry.

If you cut off the misses and blocks from both the WAR and PLD in that referenced post, both parses sit at ~20% of the parryable hits being parried, with the WAR taking slightly more critical hits.
In this thread, and others, it backs up the idea of it being a "multi roll" table. Since the mobs have a high miss rate, the parry rate reported by APP is extremely low. This is because you can't parry a miss. But if you remove misses, it leads to numbers that seem pretty reasonable for fighting lower level mobs.

I don't know how much more testing you really need, to be honest. The poster on reddit was just starting his testing, but he posted enough data to draw conclusions from, which is really what matters.

---

I would like to see without a shield, and an increase in stats to see the difference at a higher level. It's a little early in testing to draw conclusions just yet... At 193 DEX and 475 parry is one thing, but lets say DEX has a 'hidden modifier' for WARs, the difference would grow at a higher DEX rating. If he were to perform the same test with say 215 DEX and 475 parry, I would be more inclined to make conclusions at that time. A reliable parser would be greatly helpful as well.

Well I think an independent study of dex and parry would be required to draw a legitimate theory as parry could be capped under certian dex ranges. So it would be a two person event. So a War and or Mrd with a healer against lets say a ogre at Mor Dhona. There would need to be a control (Gathering gear) and then replace with dd crafted gear as it does not have parry. But the problem with this is you would have to still need to get a huge pool of dex to work with. So the 30 stat points would have to be on the table as well and then reset to do parry (or parry done first then dex so only on reset is required ). And even then we would need a large number of trials and I am not sure how many auto attacks would occur before an ogre falls over but it would have to be a joint raw data session possibly between servers to get the best results to give a base of numbers our local mathmaticians can work with so we can get a definate equation hammered out.

Quote:

---

Originally Posted by Sakasa

Well I think an independent study of dex and parry would be required to draw a legitimate theory as parry could be capped under certian dex ranges.

---

He's got a control test for Dex with the first one. It's Dex with no gear. He increased dex and parry on his second test with no increase in parry rate.

Quote:

---

Originally Posted by Coramac

He's got a control test for Dex with the first one. It's Dex with no gear. He increased dex and parry on his second test with no increase in parry rate.

---

I assumed it was and i also assumed many more then 200 trials for each of the data listed. Just laying out the experiment i suppose.

Quote:

---

Originally Posted by Sakasa

I assumed it was and i also assumed many more then 200 trials for each of the data listed. Just laying out the experiment i suppose.

---

He states he did 5000 per test.

Quote:

---

Originally Posted by Sakasa

I assumed it was and i also assumed many more then 200 trials for each of the data listed. Just laying out the experiment i suppose.

---

You mean like 200 hits per parse? That gets you no where. 200 trials of 5k hits is way overkill... that's like 1400 hours of parsing per group. That's like... literally a year of parsing.

Quote:

---

Originally Posted by Coramac

You mean like 200 hits per parse? That gets you no where. 200 trials of 5k hits is way overkill... that's like 1400 hours of parsing per group. That's like... literally a year of parsing.

---

Yeah... Not doing that.

Quote:

---

Originally Posted by Coramac

You mean like 200 hits per parse? That gets you no where. 200 trials of 5k hits is way overkill... that's like 1400 hours of parsing per group. That's like... literally a year of parsing.

---

Mis used trial it would be a data set (set per occurance hits/miss/parry). Meaning I do not doubt there was enough rounds to lead to an appriopriate Nth degree estimation similar to I do not doubt there was a control. While I am not a fan of negative control it was just say I do not doubt the experiement was done correctly.

Quote:

---

Originally Posted by Sakasa

Mis used trial it would be a data set (set per occurance hits/miss/parry). Meaning I do not doubt there was enough rounds to lead to an appriopriate Nth degree estimation similar to I do not doubt there was a control. While I am not a fan of negative control it was just say I do not doubt the experiement was done correctly.

---

An appropriate Nth degree? I don't understand. We aren't dealing with a population. What's appropriate? This subjective comment should be quantified as it can be calculated.

Additionally, this isn't science. This isn't an experiment. This is modeling and statistics.