Announcement

Collapse
No announcement yet.

Critical % formula

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Critical % formula

    Hey guys,



    I've been all over and have read a lot about % formulas to somehow calculate your % to critical from your current Crit Value Stat.

    Couldn't find anything that i found to be useful as the number they where giving me had me in the 120% crit range.

    So I did my own calculations based off of 10000+ attacks (This is for pve Only) (working on pvp) You can click on link for more info, But the Calculation is.

    V x 0.01047612 = % Critical

    Where V is the critical value stat. that is located on your stats. page.

    http://jackos24-wartune.blogspot.ca/

    THIS IS AT LEVEL 80.
    Last edited by Jackos24; 01-12-2014, 06:16 PM.
    IGN: Jacko
    Server: Kabam S24
    Guild: Genesis
    Main Objective at the moment: Win Class Wars...






  • #2
    gotta say this is a much better accurate description of % crit that ive ever seen.

    Comment


    • #3
      well that seems to match with how my crit proc is currently working o.o...however when i had lower crit i had better crit prof even thou formula shows it quite low,guess we need to find how br affects crit >_>

      Comment


      • #4
        Originally posted by S44-ZEREF View Post
        well that seems to match with how my crit proc is currently working o.o...however when i had lower crit i had better crit prof even thou formula shows it quite low,guess we need to find how br affects crit >_>
        If anything i think its level that effects crit rather then BR. As stated in my blog post i was lvl 80 when I did the testing. If there are people at lower levels that can fill out the info i will be happy to calculate the % crit for each level (based off crit Value).
        IGN: Jacko
        Server: Kabam S24
        Guild: Genesis
        Main Objective at the moment: Win Class Wars...





        Comment


        • #5
          hmmm good point when my crit was lower(2.8k) i was lv 64 and had 70% crit proc in PvE in PvP 40% or so while now lv 66 on almost 3.2k crit only around 30% in both PvE and PvP

          Comment


          • #6
            maybe the level gap of target and attacker, i notice in cata i crit alot on low levels. Worth noticing cuz, theres alot of tries here, each level is 4 attacks, and you see how often it crits really.

            Comment


            • #7
              Originally posted by Jackos24 View Post
              If anything i think its level that effects crit rather then BR. As stated in my blog post i was lvl 80 when I did the testing. If there are people at lower levels that can fill out the info i will be happy to calculate the % crit for each level (based off crit Value).
              It is not really true in my opnion. The crit rate is affected by BR. I have around 9k crit with 130k BR but crit-rate is only around 60-70%. but when I had 8k crit at 100k BR, the crit rate is around 85-90%

              Comment


              • #8
                Originally posted by R27537271 View Post
                It is not really true in my opnion. The crit rate is affected by BR. I have around 9k crit with 130k BR but crit-rate is only around 60-70%. but when I had 8k crit at 100k BR, the crit rate is around 85-90%
                Just currious but how did you calculate your % crit ?
                IGN: Jacko
                Server: Kabam S24
                Guild: Genesis
                Main Objective at the moment: Win Class Wars...





                Comment


                • #9
                  so maybe its really 100 crit for 1%. but people keep saying that br affects critrate. someone would have to repeat this test with different BR/lvl

                  Comment


                  • #10
                    This is a parameter estimation/curve-fitting problem.

                    To be more accurate, one should let z = f(r,s,w,x,y) as a general formulation, where
                    z = critical rate (between 0 to 1)
                    r = level
                    x = critical stat
                    w = BR
                    s = monster or player (binary variable)
                    y = monster/player level

                    Now, it is understood that such parameters estimation is very complicated and intractable in practice, we kept a few variables constant and attempt curve-fitting with the others. You could call this iterative estimation.
                    Given that you kept every other variables fixed, except for critical stat, you have to assume the right form of function for curve-fitting.

                    If I understood correctly, your critical rate is computed by the following formula: z = number of critical hits/total number of hits.
                    This is only true IF the critical formula assumes the form of z = ax, which is a linear function. Then your estimation minimizes the mean-squared error, and leads to your formula. You did not even assume the form of affine function (z=ax+b).

                    If what people said is correct: The critical rate is a diminishing return of the critical stat, then a more appropriate formula to use would be any concave function and do parameter estimation based on that.

                    One can assume z = a log (x) + b, or z = a exp(-bx) to begin with.

                    TL;DR
                    Your formula is most likely to be incorrect.
                    Last edited by CondorHero; 01-12-2014, 06:59 PM.
                    Officially retired from Wartune and Forums.
                    [No longer logging in]

                    Comment


                    • #11
                      Originally posted by CondorHero View Post
                      This is a parameter estimation/curve-fitting problem.

                      To be more accurate, one should let z = f(r,s,w,x,y) as a general formulation, where
                      z = critical rate (between 0 to 1)
                      r = level
                      x = critical stat
                      w = BR
                      s = monster or player (binary variable)
                      y = monster/player level

                      Now, it is understood that such parameters estimation is very complicated and intractable in practice, we kept a few variables constant and attempt curve-fitting with the others. You could call this iterative estimation.
                      Given that you kept every other variables fixed, except for critical stat, you have to assume the right form of function for curve-fitting.

                      If I understood correctly, your critical rate is computed by the following formula: z = number of critical hits/total number of hits.
                      This is only true IF the critical formula assumes the form of z = ax, which is a linear function. Then your estimation minimizes the mean-squared error, and leads to your formula. You did not even assume the form of affine function (z=ax+b).

                      If what people said is correct: The critical rate is a diminishing return of the critical stat, then a more appropriate formula to use would be any concave function and do parameter estimation based on that.

                      One can assume z = a log (x) + b, or z = a exp(-bx) to begin with.

                      TL;DR
                      Your formula is most likely to be incorrect.
                      I have yet to see how

                      w = BR
                      y = monster level

                      Effect my current crittical. I just retested this in DA just now and found to have the same Crit % using the formula. As for BR my br has gone up by roughly 5-6k since i hit 80 with no change in my crit %, but i will continue to be checking daily.

                      As for r = As i stated i'm level 80 and i'm fairly certain that there is a Player Level Variable so this formula will be more accurate at level 80.

                      As for Monster Level's effecting the % to crit. I have and will continue to test, but to date i've tested this on 70 NM dungeons / 70 Campain's / 50's (sylph Atol) and DA. All my findings have been within 1% of each other.
                      IGN: Jacko
                      Server: Kabam S24
                      Guild: Genesis
                      Main Objective at the moment: Win Class Wars...





                      Comment


                      • #12
                        Originally posted by Jackos24 View Post
                        I have yet to see how

                        w = BR
                        y = monster level

                        Effect my current crittical. I just retested this in DA just now and found to have the same Crit % using the formula. As for BR my br has gone up by roughly 5-6k since i hit 80 with no change in my crit %, but i will continue to be checking daily.

                        As for r = As i stated i'm level 80 and i'm fairly certain that there is a Player Level Variable so this formula will be more accurate at level 80.

                        As for Monster Level's effecting the % to crit. I have and will continue to test, but to date i've tested this on 70 NM dungeons / 70 Campain's / 50's (sylph Atol) and DA. All my findings have been within 1% of each other.
                        That's all fine. You can keep the BR variable and level variable constant.
                        However, you need to take into account the correct form of the function you choose to do parameter estimation over.

                        You used a linear function z = ax and you did parameter estimation to find a (here your a is 0.01 something..), but as I said, the form of the function should at least be tested against other forms of function.
                        I suggest using concave functions such as z = a log(x) + b, or z = a exp(-bx) and use LMMSE to determine a and b.
                        Officially retired from Wartune and Forums.
                        [No longer logging in]

                        Comment


                        • #13
                          Originally posted by CondorHero View Post
                          That's all fine. You can keep the BR variable and level variable constant.
                          However, you need to take into account the correct form of the function you choose to do parameter estimation over.

                          You used a linear function z = ax and you did parameter estimation to find a (here your a is 0.01 something..), but as I said, the form of the function should at least be tested against other forms of function.
                          I suggest using concave functions such as z = a log(x) + b, or z = a exp(-bx) and use LMMSE to determine a and b.
                          lol i can give you all the data if you want to plug in the values.
                          IGN: Jacko
                          Server: Kabam S24
                          Guild: Genesis
                          Main Objective at the moment: Win Class Wars...





                          Comment


                          • #14
                            Originally posted by Jackos24 View Post
                            lol i can give you all the data if you want to plug in the values.
                            Send it to me.
                            Officially retired from Wartune and Forums.
                            [No longer logging in]

                            Comment


                            • #15
                              Originally posted by CondorHero View Post
                              TL;DR
                              Your formula is most likely to be incorrect.
                              Hahaha. Oh dear.
                              I like the summation of this post. I approve.
                              People don't notice whether it's winter or summer when they're happy. - A. Chekhov
                              사람들은 자ì‹*이 행복í•* 때면 지금이 겨울인지 여름인지도 모른다.

                              Ayashi, S9 Camp Carlorn

                              Comment

                              Working...
                              X