Eternium
Eternium

Announcement

Collapse
No announcement yet.

Optimizing your Hero -- the "XP Grind"

Collapse
This topic is closed.
X
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Optimizing your Hero -- the "XP Grind"

    This short guide is intended to help you compare your "XP Build" to your "Trial Build" (or to compare any two builds or scenarios), as far as how much XP you earn, and how fast, vs. which Trial you run.

    You need the following information for both builds:
    • XP% -- found in the "Attributes Window" accessed by tapping the "..." icon on the main screen (tap the Viking Helmet to get that main screen). This is total XP% you earn. If you have no boosters active, you have a base level of 100%. Example: use CL points in the Utility tab to boost XP by 50%, you will then have 150%.
      ~~~~~~~~~~~~~~~~~~~~~~~
    • Trial Level #s you run with each build. This can be the highest level that build can complete, or it can be a lower level that you can run faster, if you wish.
      ~~~~~~~~~~~~~~~~~~~~~~~
    • Trial Completion Times, in seconds
      • the time at which you finish the mobs and the boss is summoned (this will be better for XP under almost all circumstances)
      • the total trial completion time → not necessary if you focus on the mobs only​​​​​​
    Then use these formulas to calculate XP per second, for each build or any scenario you wish:
    .
    Mobs only
    XP per second = (Trial_Pts_Mobs) * (XP_Monster_Basic) * [ (1 + Scalar_XP)^(Trial_Level - 7) ] * XP_Percent / (Trial_Time_Mobs)
    +
    XP per second = 200 * 210 * [ 1.05^(TL - 7) ] * XP% / Timemobs
    • Trial points for just the Mobs = 200 --> credit to heyjude1971 read here and adjacentengels read here
    • XP earned for one basic monster, Normal Mode, level 70 = 210 --> read the same link above for this
    • XP scalar per Trial level, above Trial level 7 = 5.0% = 0.05 --> read the same link above for this

    Whole Trial including Boss
    XP per second = (Trial_Pts_Total) * (XP_Monster_Basic) * [ (1 + Scalar_XP)^(Trial_Level - 7) ] * XP_Percent / (Trial_Time_Total)
    • Trial points for the whole trial = 225 --> read the same link above for this
    -
    ========================================
    Originally posted by adjacentengels View Post
    ...you can rearrange the formula to compare one trial to another to see if beating a lower trial faster or being able to wear more XP gear would be better than a higher trial that gives more XP per mob. This would tell you how quickly you have to finish a Trial for it to have a higher XP/sec than another trial for which you know your completion time.
    -
    Trial Level (TL)x = lower trial of comparison
    Trial Level (TL)y = higher trial of comparison
    -
    Timey = [ (1.05)^(TLy - TLx) ] * Timex * (XP%y / XP%x)-

    ========================================

    Which level is my Build's 'personal' optimum level?

    Here is a fairly quick way to find your optimum level:
    • from your max Trial Level that you have completed, drop down 21 levels (or the number of your choice)
    • repeatedly run that Trial Level, say 5 times, to get an average time
    • there is variability so you do need to do the repeats. 10 repeats would be better.
    • Don't sanitize the data -- if in the middle of the run you get killed, well that's part of the deal, right? Only exception is if you abandon (exit) at the very beginning because, for example, it's too dark or you're in a cave. Obviously don't count those.
    • keep things consistent -- use the same Build -- don't utilize any gained CL points, or change any gear, during this testing
    • again, keeping things consistent, stop at the same point (i.e. right when the Boss is summoned -- but don't forget to note the time)
    • then, move up 3 trial levels (or move up the number of your choice; 3 makes it the same Boss IF you are killing the boss)
    • repeat the next level for 5 to 10 runs, collecting the same completion-time data from the same Build
    • keep doing this until you get back to your max trial level
    • or you can calculate the ratio (below) along the way and see whether you can stop because it hits a maximum and then starts declining
    • expect some variability, so don't stop testing immediately upon getting a single lower ratio -- verify it by moving up again, and ensuring the ratio goes down again
    • using the above formulas, you can rearrange them to calculate the "relative XP/time" from each level, in comparison to the first level
    • example:
      • for the first / lowest level, "assign" it an XP gained of "1.00"
      • from your data you have the average completion time for that level = "Tx" seconds
      • so the "XP gained per second" from the first trial level is = 1.00 / Tx
      • then move up 3 levels. The relative XP from that one will be = 1.00 * (1.05)^3 = 1.157625
      • from your data you have average completion time = "Ty" seconds
      • so the XP gained per second from that second trial level = 1.157625 / Ty
      • then just compare them
        ~~~
    • Or you can compare any two with a simple ratio -- {y:x} , where Trial Level "y" is better if the ratio is >1.0 :

      Ratio = [ (1.05)^(TLy - TLx) ] * (Timex / Timey)
    \
    /

    See also Dev notes about Trials here

    Also, from Red on Discord: the XP multiplier for higher trial levels stops at TL100:



    \
    /

    HOW XP STATS AND BOOSTERS WORK:

    Outside an Event:
    • During 'normal' non-event play, the XP stat %'s (except for the purchased/rewarded boosters; see below) are additive. So, you start off "naked" at a base level of 100%. That is, you earn all the XP from your kills. 100% of it.
    • Let's say you then add the boost from Champion Level Points, which at max is 50%. Now you have 100% (base) plus (additive) 50% more from CL points = 150% total. So you are gaining "1.5x" versus base.
    • Now let's say you add some gear or jewelry. Let's go with an example of 90% on one piece of jewelry and 40% each on two pieces of gear. Together that's 90% + 40% + 40% = 170% and you add this to the above 150% to get - - in this made up example - - 320%. Now you have "3.2x" versus base.
    • Alternatively one might state it this way: in the above example you are getting "220% more than the base of 100%". The game does NOT use this nomenclature for XP stats on gear/jewelry, but it does for purchased/rewarded boosters and for gold. Don't get confused between XP stats vs. gold vs. XP boosters.
      • Example #1: %-Gold from max CL points is +50% and if you look in your attributes window (with no other boosts active) you will see it displays only "50%" (as in, 50% more than the base of 100%) for gold. This is different from Experience, where from CL points (only), it will display "150%" which again is simply that: 150% or 1.5x base of 100%. You're getting a similar +50% but they are displayed differently.
      • Example #2: an XP booster is displayed as +Y%. That is "+Y% more than the base before the booster was activated". In other words, the "Factor" =
        = {1+Y/100}*{Base} -- see below.
    • In summary, the stat boosts you get are additive, and the sum including the 100% base you start off with, is displayed in the attributes window.

    • Switching gears from stat boosts on gear/jewelry/CL points, to a purchased or rewarded XP booster... those are multiplicative.
    • The options are
      • level 1, so to speak, which costs 30 gems = displayed as +50% ... again this is "+50% more than the base before the booster was activated" so it multiplies whatever you already have by the "Factor" = {1+50/100} = 1.5
      • level 2, costs 75 gems = displayed as +100% ...the multiple is 2
      • level 3, costs 200 gems = displayed as +200% ... the multiple is 3
      • Side note: obviously the most "gem-efficient" option is therefore the "level 1" = 50% option. (1.67% per gem spent, vs. 1.33 or 1.00); But the most "time efficient" option is of course the max.
    • You can see the result of the booster while running the booster in that character's attributes window. Yes it is character specific. Only the character that starts the booster runs the booster.
    * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

    During an Event:
    • the boost during the event is for the Event Hero only
    • it is multiplicative: whatever you see in the Attributes Window, behind the scenes, it is doubled
    • you do not see the effect in the Attributes Window; we must trust that it is happening.
    • It's like running a level 2 booster (+100% = 2X) all the time
    • If you then also do run a purchased / rewarded booster, that is also still multiplicative.
      • Example #1, during event, the event hero uses a level 3 XP booster, its XP is then 2x3 = "6x" -- 3x of it is displayed in the Attributes Window but the other 2x from the event is behind the scenes
      • Example #2, you remove all gear but leave CL points alone: if you have max CL points for XP, it displays "only" 150% and if you then run a level 3 booster (+200% = 3x) it displays "only" 450% -- in other words, not 300% and 900% respectively, because the doubling during the event is not displayed in the Attributes Window
    \
    /
    Last edited by Coda; 05-26-2018, 04:22 AM.
    VUFO ZEBE ZAQI 1381
    ~ Gethi, mage, TL109 mobile & TL121 PC
    ~ Kisheli, XP mage, TL112

    ~ ANB-beta, TL100 (#1); ANB-1, TL116 (top 10); ANB-2, TL110 (top 10)

  • #2
    Travis, I vote to move this to the guides section.

    Comment


    • #3
      +1
      VOWO TOSO QEME 4656

      Comment


      • #4
        How quickly you can kill a given mob is a function of your damage, their health, their armor, and any special abilities (sisters' healing, if ghosts' drain is vampiric, etc). Without healing ghosts and sisters definitely have less health and/or armor than the skeleton warriors, for example. If armor scales with trial level that would need to be incorporated.

        Also, when considering elite vs normal mobs, it seem the time it takes to kill is a higher scaling factor than the experience you get. So just like with the boss, even though you get more experience your XP/sec will be lower. The optimum XP setup would be all normal mobs.

        There is enough variability in mob health/armor, map layout, number of elites, number of healers, etc that you can't calculate an accurate "perfect" time to beat a trial. Even with perfect clicks and gameplay you will have slightly different completion times if you run a trial over and over. But every trial level has equivalent variables, so all of that variation negates when comparing one trial level to another as opposed to trying to find the perfect time of one trial.

        But using the information to decide which trial level to farm can be very helpful. To that end, you could rearrange the formula to compare one trial to another to see if beating a lower trial faster or being able to wear more XP gear would be better than a higher trial that gives more XP per mob.

        Trial X = lower trial of comparison
        Trial Y = higher trial of comparison

        Trial_Time​​​Y = [(1 + Scalar_XP)^(Trial_LevelY - Trial_LevelX)] * Trial_TimeX * (XP%Y / XP%X)

        This would tell you how quickly you have to finish a Trial for it to have a higher XP/sec than another trial for which you know your completion time.

        Comment


        • #5
          Originally posted by Coda
          Nicely done! Would like to incorporate this into OP and give credit ... Ok?
          Sure, go ahead.

          Comment


          • #6
            Wow, there are some serious math geeks here, didn't expect that for a mobile game . Good job!

            Comment


            • #7
              Originally posted by Coda
              XP/sec ratio of TL2 vs. TL1= (1.05/1.08) ^ (TL2 - TL1) = (0.9722) ^ (TL2 - TL1)
              I agree with this result.

              Originally posted by Coda
              t = (1.05/1.08)^(TL2 - TL1) * [ 1 + HL*1.035^(TL1 - 7) ] / [ 1 + HL*1.035^(TL2 - 7) ]
              I got a different result for this. I tried a few times, making mistakes the first few iterations, so it's possible one of those mistakes carried forward. Just to make sure, I'm showing all my work. It made it easier to visualize using Word to format the formulas; see the attached images for each step.

              Base formula
              Click image for larger version  Name:	Formula01.PNG Views:	1 Size:	1.6 KB ID:	77395

              Simplify formula: All fields that multiply through the whole term that are not trial level dependent will end up canceling out
              Click image for larger version  Name:	Formula02.PNG Views:	1 Size:	1.4 KB ID:	77396

              Now start to expand on the time to kill mobs on a given trial level (using your formula). W is trial-dependent so named it WTL
              Click image for larger version  Name:	Formula03.PNG Views:	1 Size:	1.4 KB ID:	77397

              Now to expand WTL
              Click image for larger version  Name:	Formula04.PNG Views:	1 Size:	1.2 KB ID:	77398

              Substitue WTL back into the TimeMobs formula
              Click image for larger version  Name:	Formula05.PNG Views:	1 Size:	1.8 KB ID:	77399

              Assume M=0; pull H out since it is included in both (I could have pulled DPS out and save a couple steps)
              Click image for larger version  Name:	Formula06.PNG Views:	1 Size:	2.2 KB ID:	77400

              Now consolidating over the common denominator, DPS
              Click image for larger version  Name:	Formula07.PNG Views:	1 Size:	1.9 KB ID:	77401

              Plug in Scale factors (Heal=0.035, Health=0.08), rewrite as numerator/denominator
              Click image for larger version

Name:	Formula08.PNG
Views:	1
Size:	1.7 KB
ID:	77410

              Pull out common terms not dependent on trial
              Click image for larger version

Name:	Formula09.PNG
Views:	1
Size:	1.6 KB
ID:	77411

              Plug TimeMobs back into the simplified XPS formula in step #2
              Click image for larger version

Name:	Formula10.PNG
Views:	1
Size:	1.7 KB
ID:	77412

              Flip and pull out common terms
              Click image for larger version

Name:	Formula11.PNG
Views:	1
Size:	1.7 KB
ID:	77413

              Build formula of XPSTL2 / XPSTL1; notice terms not dependent upon trial cancel out, but HL remains (this was the mistake I had to fix after my original posting)
              Click image for larger version

Name:	Formula12.PNG
Views:	1
Size:	2.4 KB
ID:	77414

              Flip denominator and rearrange so like terms are in similar order
              Click image for larger version

Name:	Formula13.PNG
Views:	1
Size:	1.7 KB
ID:	77415

              This is where I think we diverge.
              1. Pull out and consolidate 1.05X terms, as we've done in other simplifications
              2. In the numerator, pull out 1.08TL1-7 factor from both terms in the parentheses; the first term needs to become (1.035/1.08)TL1-7 and I think this wasn't done in your calculation
              3. Do a similar operation to the denominator pulling out 1.08TL2-7
              Click image for larger version

Name:	Formula14.PNG
Views:	1
Size:	2.4 KB
ID:	77416

              Last step is to merge the 1.05 and 1.08 using the common power term (TL2 - TL1) as you did when you assumed W=0.
              Click image for larger version

Name:	Formula15.PNG
Views:	1
Size:	2.0 KB
ID:	77417


              EDIT: Had a mistake in step 8 where I accidentally pulled HL out as a common term with H and DPS.
              Last edited by adjacentengels; 12-11-2017, 08:20 PM.

              Comment


              • #8
                There are a few other sources of uncertainty that may or may not be reasonable to model as scaling with trial level.

                Wasted Movement:
                At higher trial levels you have to pay more attention when pulling groups (especially as a Mage) because all the flying attacks have a greater chance of killing you; especially multi-strike or multiple ghosts draining you. So your movement becomes less efficient as you play more conservatively.

                Wasted Damage:
                At lower trial levels you may be over-killing mobs (hit a mob for 1M damage when he only has 600k health), which would become wasted damage. This would make it seem like you could clear a low level trial more quickly than reality.

                Mob Armor:
                I don't know if mobs have armor in the same way that players do, but if they do, and especially if it also scales with trial level, not only do mobs have more health in higher trials, but they also absorb more health at higher trials.

                All of these are rather open-ended, so trying to incorporate them into the formula would get even uglier, but they are sources of variation that would need to be considered in some way.

                Comment


                • #9
                  Serious math work. Impressive.
                  Last edited by Rytrik; 12-12-2017, 04:27 AM.

                  Comment


                  • #10
                    RiseOfAbyss quick maffs
                    22nd September 2017

                    Comment


                    • #11
                      Holy fk yea LOL
                      Originally posted by CountBelzebul;n96592
                      when i start playing eternium last year...my warrior had 3 maggie as companion...when i heard maggie being nerfed...i got frustrated...then i just delete my warrior...
                      YOLO

                      Comment


                      • #12
                        Is there any reason this is better than just checking how many bars of xp you get per second? My question is will I gain something by doing the math.

                        Comment


                        • #13
                          How long does a optimal run with 800% exp gear takes? 2 min? 3 min? If i can Just get this number i can run close to It and bem close to best witgout the math...

                          Comment


                          • #14
                            Originally posted by Seilmore View Post
                            How long does a optimal run with 800% exp gear takes? 2 min? 3 min? If i can Just get this number i can run close to It and bem close to best witgout the math...
                            Sorry, doesn't work that way.

                            Comment


                            • #15
                              A point would be if D is established by Trial stats, that is skewed by the boss fight and the possible single vs mob damage differences.

                              However, if they implement stats for each run instead of best run, we may be able to make some estimations?
                              Last edited by orion_134; 12-24-2017, 11:52 PM.

                              Comment

                              Working...
                              X