[Mabinogi KR Test] Renovation Update - Ranged Combat // 12th of Feburary, 2015

• Kaeporo wrote on 2015-03-02 12:47

  Quote from Qnjo;1266800:

  Back to the topic of Throwing Attack, I did a quick test myself and here are the results:
  [SPOILER="My shamefully low stats."]

  [Image: http://i.imgur.com/JV55j3Z.jpg]

  [/SPOILER]
  Math time. With 80 balance my final min should be 130 rounded up and an ideal max is 297. For the sake of simplicity let's settle for 213 average.
  
  I took 50 javelins and used them all on foxes around Filia. The lowest damage was
  [SPOILER="179"]

  [Image: http://i.imgur.com/7bQWtGW.jpg]

  [/SPOILER] and my best non-critical shot has amounted to
  [SPOILER="281"]

  [Image: http://i.imgur.com/7cLpoyr.jpg]

  [/SPOILER]
  And that makes an average of 230.
  
  So... am I missing something with calculations? Because it seems that Throwing Attack works as intended.

  
  
  80% balance would result in an average damage output of 263.
  
  Throw a few hundred javelins and post a spreadsheet detailing your damage output for all non-critical hits. It would take around a thousand samples to really narrow down the exact number...though JP testers already did that for us.
• Oberynn wrote on 2015-03-02 14:37

  Quote from Qnjo;1266800:

  So... am I missing something with calculations? Because it seems that Throwing Attack works as intended.

  
  
  Sorry if I'm misunderstanding what you're trying to say, but balance is not an average between your highest and lowest damage and it does not determine or affect your highest or lowest damage in any way.
  
  Balance is just damage consistency and it's like a bell curve. If you have 1~1000 damage range and 80% balance, you can still hit 1 and you can still hit 1000 but half of the time you will hit above 800 and half of the time you will hit below 800 and the further away from 800 the least likely you are to hit.
  
  What people mean when they say throwing attack has low or no balance, is that who knows what you're gonna hit. It could be 1, it could be 1000, or it might take hours just to hit either.
• Qnjo wrote on 2015-03-02 15:39
  Now I see. I have always thought that balance is a damage modifier (like 0.8 for 80%), but it isn't. I have much to learn...
  
  

  Quote from Kaeporo;1266808:

  It would take around a thousand samples to really narrow down the exact number...though JP testers already did that for us.

  
  Thank you, I'll keep that in mind. Has a test like that been conducted in KR, after range renovation?
• Rydian wrote on 2015-03-02 17:34
  Balance determines where the bell curve is centered (between min and max), it doesn't determine the range from a fixed center.
  
  If we're talking 130 min and 297 max, then with 80 balance the typical hit would be centered around... 263.6
  Again with the 130 min and 297 max, then with 50 balance reported, the typical hit would tend to be... 213.5
  
  So yeah, a test with a lot of samples to graph would be best.
• ironwoman wrote on 2015-03-02 18:19

  Quote from Qnjo;1266800:

  Back to the topic of Throwing Attack, I did a quick test myself and here are the results:
  [SPOILER="My shamefully low stats."]

  [Image: http://i.imgur.com/JV55j3Z.jpg]

  [/SPOILER]
  Math time. With 80 balance my final min should be 130 rounded up and an ideal max is 297. For the sake of simplicity let's settle for 213 average.
  
  I took 50 javelins and used them all on foxes around Filia. The lowest damage was
  [SPOILER="179"]

  [Image: http://i.imgur.com/7bQWtGW.jpg]

  [/SPOILER] and my best non-critical shot has amounted to
  [SPOILER="281"]

  [Image: http://i.imgur.com/7cLpoyr.jpg]

  [/SPOILER]
  And that makes an average of 230.
  
  So... am I missing something with calculations? Because it seems that Throwing Attack works as intended.

  
  
  Yes, quite a lot actually. Your median damage isn't as simple as throwing 50 spears. If you want way more serious samples, you're gonna have to throw at least 100 times that number, literally.
  
  Also, even using your method, we can see the balance is closer to 50% than 80%.
  
  

  [Image: http://oi62.tinypic.com/4sbjev.jpg]

  
  
  Now using your values and 50% balance...
  
  {(297-130)*.5}+130 = 167 * .5 + 130 = 83.5 + 130 = 213.5.
  
  If using 80% balance, {(297-130)*.8}+130 = 167 * .8 + 130 = 133.6 + 130 = 263.6
  
  I'm pretty sure 230 is a closer number to 213.5 than 263.6, even using a tiny sample size.
  
  Now onto more of the update stuff...
  
  Comprehensive list of Alban Knight Dungeon mobs and rewards.
• wallrocha wrote on 2015-03-02 20:44
  Since we are already on this, how does magic balance at 100% work at all? It sure does not hit at 100% damage the whole time (especially on spells like Blaze)
• ironwoman wrote on 2015-03-02 21:05

  Quote from wallrocha;1266831:

  Since we are already on this, how does magic balance at 100% work at all? It sure does not hit at 100% damage the whole time (especially on spells like Blaze)

  
  
  Refer to the picture I posted.
• gundamweed wrote on 2015-03-02 21:20
  All this talk of taking thousands or hundreds of samples is simply silly. I've already taken three classes on this sort of crap at my university (statistics, research methods, and testing & measurements) , and I can tell you one of the basic principles for getting a bellcurve that's pretty close to acceptable would require 30 samples, just to get a rough idea. adding more samples simply refines that bell curve and smooths it out, but you're not gonna change radically. This is for practical purposes, and not getting an exact number, but I beleive if you're testing for either 50 or 80 balance, the answer doesn't have to be exact. It just has to be approximate to see which balance stat your bell curve most resembles, either or.
  
  Of course, simply taking your lowest and highest number and gettnig an average of that was wrong. What you should have done was record each and every non crit, 30 or 50 times, and see which numbers were the most common.
• ironwoman wrote on 2015-03-02 21:35

  Quote from gundamweed;1266834:

  All this talk of taking thousands or hundreds of samples is simply silly. I've already taken three classes on this sort of crap at my university (statistics, research methods, and testing & measurements) , and I can tell you one of the basic principles for getting a bellcurve that's pretty close to acceptable would require 30 samples, just to get a rough idea. adding more samples simply refines that bell curve and smooths it out, but you're not gonna change radically. This is for practical purposes, and not getting an exact number, but I beleive if you're testing for either 50 or 80 balance, the answer doesn't have to be exact. It just has to be approximate to see which balance stat your bell curve most resembles, either or.
  
  Of course, simply taking your lowest and highest number and gettnig an average of that was wrong. What you should have done was record each and every non crit, 30 or 50 times, and see which numbers were the most common.

  
  
  And I've taken enough Math and stats in my college too, enough to change my major right over to Math and Statistics if I wanted to. While you can get the most basic curves with only 50-80 hits, in a game like Mabinogi, where the damage difference can be as big as 400 between minimum and maximum damage (in my case, 500-900), it is still arguably better to do many trials to smooth it out. And if we didn't know the balance of the throwing attack beforehand, which is why people did this thousands of hits, then you can't really say 50 hits is enough to determine the balance.
• Zekkii wrote on 2015-03-02 21:36
  You guys realize that throwing attack is still weak right? The balance doesn't matter.
• Rydian wrote on 2015-03-02 22:36

  Quote from wallrocha;1266831:

  Since we are already on this, how does magic balance at 100% work at all? It sure does not hit at 100% damage the whole time (especially on spells like Blaze)

  Balance determines where the center of the curve is located (what percentage of the way from your min to your max), it does not determine the width of the curve.
  
  Assuming a standard distribution (bell curve, 40% at center), then here's some rough stat-students-might-mock-this examples if you had 350 min and 650 max damage.
  
  With 50% balance...
  

  [Image: http://i.imgur.com/Qsey3qZ.png]

  
  
  With 80% balance...
  

  [Image: http://i.imgur.com/uzQMqIs.png]

  
  
  The point I'm trying to make is that in both cases the "width" of the curve is the same, balance just determines where the center of the curve is. The curve continues past where I drew a path, just the chances become very low the further away you get from wherever it's centered.
  
  So with 100% balance (using magic), you will still hit less than max damage around half the time.
  
  

  [Image: http://i.imgur.com/cyuasxT.png]
• Oberynn wrote on 2015-03-02 23:41

  Quote from wallrocha;1266831:

  It sure does not hit at 100% damage the whole time (especially on spells like Blaze)

  
  
  Blaze completely disregards balance and just hits 100% always... it's like the only skill that does that, so I'm not sure why you single it out as an example for the opposite. Maybe you're just not fully charging? Or somehow I've blazed 200 bears and gotten the exact same number every time by a freak accident of luck.
  
  

  Quote from Zekkii;1266838:

  You guys realize that throwing attack is still weak right? The balance doesn't matter.

  
  
  But many people think what makes it weak is the balance.
  
  There are people here doing 10k+ non crit before they even rearrange the formula and add upgrades/special upgrades. You can also get effective range 20 on them and they will shoot every 4 seconds after update. None of that makes it seem weak to me. But having a crappy damage range that's all over the place? that sure does.
• Traw wrote on 2015-03-03 00:00

  Quote from Oberynn;1266852:

  There are people here doing 10k+ non crit before they even rearrange the formula and add upgrades/special upgrades. You can also get effective range 20 on them and they will shoot every 4 seconds after update. None of that makes it seem weak to me. But having a crappy damage range that's all over the place? that sure does.

  
  When you realize how end game you have to be for that crap then yea its weak
• gundamweed wrote on 2015-03-03 01:41

  Quote from ironwoman;1266837:

  And I've taken enough Math and stats in my college too, enough to change my major right over to Math and Statistics if I wanted to. While you can get the most basic curves with only 50-80 hits, in a game like Mabinogi, where the damage difference can be as big as 400 between minimum and maximum damage (in my case, 500-900), it is still arguably better to do many trials to smooth it out. And if we didn't know the balance of the throwing attack beforehand, which is why people did this thousands of hits, then you can't really say 50 hits is enough to determine the balance.

  
  
  While there is the Law of Large Numbers, you should know your supposedly numerous instructors told you it's pretty redundant to collect an eccentrically large sample, and even comes with problems when inappropriate. Knowing if our difference is statistically significant is easy, because we're comparing 80 balance to 50 balance. What a huge difference! In such cases, it's cited and stated in your supposedly numerous math and stats text books as fact that we do NOT need a large sample to determine this difference.
  
  
  Even then, no matter how large your sample, there'll always be "sampling error". Since you'll never be exact and the difference being closer to the true balance won't differ much from 90 samples to 9000 samples, it is a waste of time and even intellectually questionable to suggest large sample taking in this case.
• ReFuSeR wrote on 2015-03-03 01:49

  Quote from Traw;1266857:

  When you realize how end game you have to be for that crap then yea its weak

  
  Isn't it just str based?