The Lowdown on Chance to Hit

[Nyo]

I spent quite a while testing different amounts of accuracy and evasion to come to these results. Far, far too long. In any case, here are the results!

• You get a +10% boost for being at the side of the target
• You get a +20% boost for being behind
• Ex. - you have a 50% chance normally. +20% of 50% is 60%.
• You cannot have higher than 98% chance to hit
• You cannot have lower than 27.5% chance to hit
• Day/night cycle has no effect on chance to hit unless of course you have Shadow Sense.
• Height relative to target has an effect on chance to hit. This is not tested as of yet.

To find your chance of hit, you take your accuracy minus the target's evasion. There is a bit of math to that, but I'm keeping it simple and to the point. So...
[Accuracy] - [Evasion]
Take the result and use this:
+10 or higher - 98%
+9 - 97.5%
+8 - 97%
+7 - 96.5%
+6 - 96%
+5 - 95.5%
+4 - 94.5%
+3 - 93.5%
+2 - 92.5%
+1 - 91.5%
0 - 90.5%
-1 - 89%
-2 - 87.5%
-3 - 86%
-4 - 84.5%
-5 - 83%
-6 - 81.5%
-7 - 80%
-8 - 78.5%
-9 - 77%
-10 - 75.5%
-11 - 73.5%
-12 - 71.5%
-13 - 69.5%
-14 - 67.5%
-15 - 65.5%
-16 - 62.5%
-17 - 59.5%
-18 - 56.5%
-19 - 53.5%
-20 - 50.5%
-21 - 45.5%
-22 - 40.5%
-23 - 35.5%
-24 - 30.5%
-25 or lower - 27.5%

There are actually a lot of theories I found on how accuracy versus evasion works. Some were pretty darn close to the testing results I have. The accuracy on the low end was usually a bit too low or high though so I simply devised my own theory. You're looking at it.

I'll be using this setup in L2Calc when the next version is released. If you believe there is something wrong, major problem or not, now is the time to speak up!

And lastly here are the tests I performed, uglified in a fat /code box.

accuracy-evasion-position-hits-tries-ratio-difference
109.594   71   front   392   400   0.98   38.594       
101.844   83   front   393   400   0.9825   18.844   
69.393   68.3109   front   371   400   0.9275   1.0821       
65.643   68.3109   front   348   400   0.87   -2.6679       
65.643   66   front   372   400   0.93   -0.357       
69.393   66   front   379   401   0.945137   3.393       
69.393   74.311   front   337   411   0.819951   -4.918       
65.643   74.311   front   320   400   0.8   -8.668       
69.393   60.3109   front   391   400   0.9775   9.0821       
65.643   60.3109   front   388   400   0.97   5.3321       
66.144   61.775   front, night   382   401   0.952618   4.369       
78.496   74.311   front   426   450   0.946667   4.185       
69.393   68.3109   front   380   400   0.95   1.0821       
75.496   74.311   front   422   461   0.915401   1.185       
74.746   74.311   front   382   418   0.913876   0.435       
78.496   79.495   front   389   433   0.898383   -0.999       
65.643   68.3109   front   354   404   0.876238   -2.6679       
71.746   74.311   front   351   413   0.849879   -2.565       
75.496   79.495   front   370   432   0.856481   -3.999       
69.393   74.311   behind   401   411   0.975669   -4.918       
69.393   74.311   front   337   401   0.840399   -4.918       
65.643   74.311   behind   401   405   0.990123   -8.668       
65.643   74.311   front   305   400   0.7625   -8.668       
63   74.311   front   341   448   0.761161   -11.311       
71.746   83.495   front   328   425   0.771765   -11.749       
87.489   101.59   front   876   1264   0.693038   -14.101       
87.489   102.59   front   271   403   0.672457   -15.101       
87.489   104.394   front   230   404   0.569307   -16.905       
87.489   105.59   front   228   402   0.567164   -18.101       
83.489   104.394   front, night   181   405   0.446914   -20.905       
83.489   105.59   front   148   400   0.37   -22.101       
83.489   105.59   front, day & night   188   404   0.465347   -22.101       
87.489   111.59   front   118   402   0.293532   -24.101       
87.489   111.59   front, day   123   404   0.304455   -24.101       
87.489   115.59   front   135   478   0.282427   -28.101       
87.489   115.59   front   104   399   0.260652   -28.101       
83.489   115.59   front   117   474   0.246835   -32.101       
83.489   115.59   front, night   125   404   0.309406   -32.101       
87.489   119.59   front   114   399   0.285714   -32.101       
87.489   119.59   front, night   111   404   0.274752   -32.101       
83.489   119.59   front   127   477   0.266247   -36.101   
48.311  119.59  front, night  112  418  0.267943  -71.279         
48.311   119.59   behind, day+night   130   420   0.309524   -71.279       
83.489   119.59   behind   129   404   0.319307   -36.101

[ttekert]

I was thinking, if those results are product of "a lot of manual testing"
Maybe the following formula is the real one:
(sry i mean, is hard to do a formula with those results, for example, -7 is 20% to miss, and -8 22.5% and -9 23%, is a erratic curve)
Base 10% evasion when acu=eva, then 10% difference in chance to evade per point, minimum and maximum 3%-75% miss)
(%) to miss = 10 * 1,1^([evasion] - [accuracy])
I Think ^^.
Bye.

[Nyo]

It curves a bit low when the difference between accuracy and evasion is big. I've decided that the accuracy vs evasion probably isn't a simple formula like that. There is precedence for this, for example with HP/MP/CP regen where it uses a tier system for every 10 levels of the character. At every 10th level, there is a large jump in regen rate. I applied a similar concept when I decided on how chance to hit probably works.

[ttekert]

I understand, but for example, one of my concerns is DPS with and without shields using your table. A mystic lvl 70+ vs mystics lvl 60+- without shield they hit you 41.8dps, and With shield 44.2, i know.. it makes some sense to hit you more, but 5% more? mmm, using my own formula for example, the results are, 43.002 without shield and 42,9 with shield.. that seems a little more logic, if not, whats the point in shield sometimes, someting needs a little more research (i think).
Is only a point of view, i know, it doesn't say nothing, but it draws me to think the other way around (+10% every point etc).

[Nyo]

Why does a level 70+ mystic need to take less damage with a shield than without against a character who has much lower accuracy?

[ttekert]

logic? ^^, not less, at least equal, but no higher. Yeah i know logic doesn't work here hehe.

edit: bah, i didn't say anything ^^, your words makes sense from another perspective i have now.
These days i was measuring if a shield on a mystic really helps overall. (i was thinking to go duals)