Zanshin proc rate and accuracy bonus

[Motenten]

This is done based on the same technique as the crit rate testing. I've been curious about the exact numbers for a while anyway.

The below test was completed in roughly 20 hours against a level 72 Ul'hpemde. I set up as Sam/Blm, no Brutal, no accuracy/dex gear and -acc rings, in order to minimize hit rate and ensure no other source of double attack. The results are from the Extra Attacks plugin of KParser (including an extra bit of collated info that I wrote up and will be included in the next release). I removed the DA/TA/Kicks sections as they aren't relevant to this test.

Basic Data

Player               # Melee Attacks    # Melee Rounds    Attacks/Round    # Extra Attacks
Motenten                       31100             25410                1               5690

Player               # +1 Rounds   # +2 Rounds   # +3 Rounds   # +4 Rounds    # >+4 Rounds
Motenten                    5685             1             1             0               0

Player               # MultiAttack Rounds    MultiAttack %     Kills w/Min Attacks    Kills w/<Min Attacks
Motenten                             5687          22.38 %                       0                       0


Treat As:

Multi-attacks per attack (Zanshin):

Player               # Missed First Attacks    # DA w/Missed First   Possible Zanshin %
Motenten                              12636                   5680              44.95 %

Player               Acc. Rate of First Attacks    Acc. Rate of Second Attacks
Motenten                                50.27 %                        66.67 %

25,410 melee rounds. 12,636 of those missed the first attack, so 12,774 hit, giving a 50.27% hit rate on the first round with a 0.61% confidence interval. Predicted hit rate based on acc vs evasion should be 50.0%.

There were 5680 rounds that had a DA after the first attack in the round missed (12,636) -- that is, Zanshin procs. Zanshin rate is thus 44.95% with a 0.87% confidence interval.

Of the 5680 Zanshin procs, 3787 connected (not shown, but extrapolated from the acc rate on second attacks). This gives a 66.67% hit rate with a 1.23% confidence interval.

Accuracy bonus from Zanshin is thus between 15.44% and 17.90% hit rate, or between 31 and 35 acc. 35 seems the most probable 'round' number in that range.


You'll note that there are reported to be 5690 extra attacks, whereas there were only 5680 Zanshin proc rounds. In the second row you'll see that there was one +2 round and one +3 round (ie: rounds with 3 and 4 attacks). This is obviously a parser error. The Zanshin numbers are restricted to only rounds with 2 attacks, so those mistaken multi-attack rounds won't affect the Zanshin outcome.

There are also five +1 rounds (two attacks) where the first attack was not missed. Again, parser error. The second attacks of those rounds will also have no impact on the Zanshin results because the first hit of those rounds wasn't missed.


Summary: Zanshin proc rate at level 75 is 45%, and provides a +35 acc bonus to the Zanshin hit (subject to correction based on additional sampling).

I also plotted out the effect of the extra attacks and accuracy bonus in Excel across all base hit rates. Excluding the impact of Double Attack traits and gear, Zanshin is effectively the same as a War's Double Attack trait (+10%) when base hit rate is 81%. It's the same as a +20% DA rate when base hit rate is 65%. At 90% base hit rate, Zanshin is the equivalent of +4.75% DA, and at capped accuracy it's +2.25% DA.

It is better to treat the effect of Zanshin as if it were +DA rather than +Acc, in terms of impact on damage. Zanshin does not allow your hit rate to go above the 95% hard cap, for example, although it does raise your average hit rate in any instance when you are not capped. Instead, it gives you more chances to swing, and a better chance to hit on those swings.

[Onorgul]

Nice information. I'd been meaning to do something like this but hadn't found the time. Thank you.

[Shamaya]

Deadgye's method wins again ^^ Pretty nice that you could figure this out so easily.

[Thunder]

Do you have hachiryu gear to try out the enhance zanshin effect?

[Byrthnoth]

Wow, this proves again how bad humans are at guessing percentages of low frequency events. I was convinced Zanshin was a 10% native proc rate and had no idea it got an accuracy bonus.

[Lilswitche]

Interesting data, and thanks for the time you put forth testing this!

Just wondering. Are you planning to do the same thing to test /sam zanshin?

[Motenten]

Unfortunately I don't have any Hachiryu gear, so can't test that.

And yes, I'm planning to test /sam as well. However it was suggested that I should run another test at a different hit rate point to verify that it's a flat addition and not a scaled one. As it's a reasonable concern (JP tests are suggesting a scaled acc bonus), I'm running that first.

[Kirschy]

Wow 45% activation rate, and a large acc bonus on the extra hit. Thats crazy and I'm impressed. I wouldn't imagine you want to re-run the test /w /SAM Big thanks for the testing!

[Yugl]

Wow 45% activation rate, and a large acc bonus on the extra hit. Thats crazy and I'm impressed. I wouldn't imagine you want to re-run the test /w /SAM Big thanks for the testing!

The testing is actually simple since you can afk during the entire process. Check the Critical Hits thread for details on why that is the case.

[Kirschy]

Yea I saw the test in the crit thread. It's a great test, and I think it has a lot of applications~

[Masamune]

Thank you deeply for that test Motenten. i reread it 2 times, found 0 flaws(yu found them yurself), if i was a teacher @ school surveying yur copy=20/20 w.

Except maybe one thing: yu saying Zanshin can't allow to get above the 95% accuracy hard cap. when i included Zanshin (with a base proc rate=10%) into my DoT formula in my excel tool, alongwith DA/TA and accuracy probabilities, it puts total theoric accuracy(hence hit rate) @ 97.14% @ capped conditions ! So yu can guess my surprise when i read this topic and see that :

and at capped accuracy it's +2.25% DA.

That makes gross 95% accu for the additional 2.25% DAd hits added by zanshin correct ? Check posts below.

[pahnphoenix]

masamune,

95% is the hard cap. no amount of further swings could cause you to get > 95% ACC.

i (and others) have said for a long time that you should think of a zanshin as "an accuracy bonus" (i would say "irrelevant ACC bonus" but that doesn't matter), but we're speaking metaphorically.

say you have 95% ACC. you swing 10,000 times, miss 500 times. accordingly, you zanshin 225 times, and land 213.75 of them (don't be upset about the fraction).

how many total swings? 10,225. how many hits? 9,713.75. hit rate? 95%.

of course, if you're under 95% ACC, you will get an ACC bonus, but only on the zanshin hits as illustrated in motenten's test (and the patch notes when they implemented the zanshin ACC bonus). the additional hits are, of course, never an ACC bonus on the non-zanshin hits, no matter how you spin or think of it. when people say "think of zanshin like an ACC bonus," they either mean: a) (me version) don't worry about it; it won't change your gear choices, it just gives you a few more hits, or b) (zanshin aficionado version) zanshin influences what can or can't be a tolerable hit rate for X, Y, or Z (say, wearing askar korazin).

edit: if anyone cares, the "me version" above is supposed to carry the sense of "it's *just* an ACC bonus, not a reason to wear X,Y,Z or merit zanshin or blablabla. it's a nice little bonus of hits." that may or may not be true in all cases, but it's where i was coming from.

[Masamune]

erm actually i was gonna reedit my post above after checking my tool results under capped/non capped hitrate conditions, and i get always 95% indeed XD i've worked this dot formula at like beginning of this year with CDF so i don't remember very well, forget my above post. Applied hard accuracy cap AFTER DA/TA/Zanshin total #hits have been calculated.
Thank you anyways for yur post Pahn, illustrates nicely the misunderstanding.

EDIT: actually there is still a difference: i assumed Zanshin=10% as a base (not ~45%), no matter level or /sam, exactly like DA for war, with its own merits as SAM75.

Do you want me to test as PLD/sam Mot ?

[test123]

can you do a 2nd test to make sure Zenshin is not a static 2/3 connect chance instead of +35 acc?

[Motenten]

I'm currently running a test with +30 acc compared to the first test. However I can verify right now that the hit rate is not simply 67%, since I ran tests a few months ago with near min accuracy (using an elemental staff in campaign), and the Zanshin hit rate on those was around 40%-45%.

Also, a quick check of the numbers in my current test has my Zanshin hit rate at 81.75%, roughly 15% higher than the first test, as would be expected from the +30 acc in gear.


And Masamune: Yes, please, if you have the time that means I don't have to do it And you can collect the data now, and then analyze the hit rate once I release the new parser version (I'm holding it til after the patch since I'll have to update it for that anyway). I'm using a ceremonial dagger and turban and it's having no real trouble counting the DAs, if you want a reference for what speed you can work at.

[Deadgye]

Deadgye's method wins again ^^ Pretty nice that you could figure this out so easily.

fufufufu I've always said being incredibly lazy was one of my most awesome traits.

[Taint]

Sweet test, TY

[Masamune]

And Masamune: Yes, please, if you have the time that means I don't have to do it And you can collect the data now, and then analyze the hit rate once I release the new parser version (I'm holding it til after the patch since I'll have to update it for that anyway). I'm using a ceremonial dagger and turban and it's having no real trouble counting the DAs, if you want a reference for what speed you can work at.

Ok im gonna run a test as PLD75/sam37 with IbushiShinai (im 8sword merited) on Om'Hpemde + Turban/Homam ^^

... Actually i can't -.-, explanation :

- Om'Hpemde lvl76:
VIT=80
AGI=72
DEF=314
EVA=302

- Om'Hpemde lvl77:
VIT=80
AGI=72
DEF=319
EVA=307

requires 66 max str for 0 dmg. My elvaan PLD/sam naked has STR80 base... no way i can find gears dropping my str for -14... Sorry Mot

[Motenten]

Preliminary +30 acc data:

Basic Data

Player               # Melee Attacks    # Melee Rounds    Attacks/Round    # Extra Attacks
Motenten                       25772             22236                1               3536

Player               # +1 Rounds   # +2 Rounds   # +3 Rounds   # +4 Rounds    # >+4 Rounds
Motenten                    3534             1             0             0               0

Player               # MultiAttack Rounds    MultiAttack %     Kills w/Min Attacks    Kills w/<Min Attacks
Motenten                             3535          15.90 %                       0                       0


Treat As:

Multi-attacks per attack (Zanshin):

Player               # Missed First Attacks    # DA w/Missed First   Possible Zanshin %
Motenten                               7823                   3525              45.06 %

Player               Acc. Rate of First Attacks    Acc. Rate of Second Attacks
Motenten                                64.82 %                        82.72 %

65% hit rate on the first attack of each round, compared to 50% previously. +15% hit rate matches +30 accuracy.

82.72% hit rate on Zanshin attacks, compared to 66.67% previously. 16% higher roughly matches +30 accuracy.

Margin of error for Zanshin accuracy bonus is 1.25%. That gives the current test's accuracy bonus a range of +33 to +37. Combined with the previous test's +31 to +35, that puts the accuracy bonus range between +33 and +35.


I had intended to run this longer to reduce the margin of error, but I think this may be sufficient.

Conclusions:

1) Zanshin's accuracy bonus does not scale with base accuracy. It's a constant flat amount. (While there's a slight variance in the margin of error, it's not large enough to be credible as having scaled with the accuracy unless it's something like 30 + (hit rate / 10), which is barely worth the trouble of identifying.)
2) Zanshin's accuracy bonus range has been narrowed, and +35 still looks like the best number.

[Motenten]

Starting the sam subjob test:

Pld/sam

Ceremonial dagger + Turban + 2x Merman's ring

Dagger skill is 20 points higher than sam's, so +18 acc.

Target level 75 Om'hpemde (297 eva), so +17 eva.

Dex of 62 rather than 68, so -3 acc. Should end up with 49% hit rate.