Violent > Swift (?) -- the question and the answer

Discussion in 'General' started by Ittadakkimaas, Dec 11, 2014.

  1. Ittadakkimaas

    Ittadakkimaas Level 38

    Joined:
    Aug 6, 2014
    Messages:
    2,386
    Likes Received:
    715
    The Question
    I've wondered for a while why top tier players prefer Violent over Swift in spite of the fact that Swift runes are so much easier to farm.

    1) Lets say a monster has a speed of 100. Both Swift and Violent give it an expected speed of 125 (calculation for Violent is easy -- expected number of turns = 1 + 0.2 + 0.2^2 + 0.2^3 + ...).

    My understanding was that the primary difference was that Violent was more unpredictable and RNG based which makes it better for AD (harder to predict --> harder to beat), whereas Swift guarantees first turn making it better for manual teams.

    The Answer
    I just realized today that I was being stupid -- Violent actually gives you more turns than Swift in expectation -- the difference comes from the fact that Swift gives you extra turns based on your base speed, while Violent does so based on your post-rune speed.

    Here's an example to illustrate when/why Violent is significantly better than Swift:
    2) Suppose you have 100 base speed and a +50 speed from stats. Now, a Swift set will give you effective speed of 1.25 x 100 + 50 = 175. Violent, on the other hand will give you an effective speed/number of turns of (1 + 0.2 + 0.2^2 + 0.2^3 + ...) x (100 + 50) = 1.25 x 150 = 187.5

    The higher the +spd you plan on getting from substats/slot 2 etc, the greater the difference between Violent and Swift.

    The Verdict
    Unless your team relies on going first (buffers/shielders/cc etc), Violent is in most cases strictly better than Swift. Unfortunately this epiphany has come to me long after I have already dumped a Swift set on my Light Inugami, Light Bearman and Fire Vampire - its too late for me, but it may not be too late for the rest of you - save yourselves! Don't make the same mistakes I did...

    NOTE: Swift is probably still the better choice for Fire Vamp if you are planning to use it as a speed arena buffer. If you are planning to use it for long drawn out dungeon farming, then Violent is better.
     
    Last edited: Dec 11, 2014
    ExUprising, slopixi, Darky and 5 others like this.
  2. BrFrPlayer

    BrFrPlayer Level 37

    Joined:
    Apr 10, 2014
    Messages:
    2,056
    Likes Received:
    661
    Speed will reliably give you more turns if your total speed is more than 200% of the opponents'. Therefore, it only gives you a massive advantage in situations like right after a reset where teams with avg. 130 or so speed can faces ADs with more than 260 speed because their ranks are the same or in PVE because PVE monsters have no runes.

    Violent is also very streaky and you can lose a lot of Arena ranks in a row simply due to bad RNG, just like with Megans and Bernards in AD teams.
     
    ExUprising likes this.
  3. Ittadakkimaas

    Ittadakkimaas Level 38

    Joined:
    Aug 6, 2014
    Messages:
    2,386
    Likes Received:
    715
    I thought it was obvious (both implicitly, and through some of my explicit examples) what kinds of monsters/situations my post was talking about - I have also mentioned that Swift is better in some situation like buffers for PvP etc.

    This is about the expected number of turns a monster gets --- makes it very significant for long/tanky PvP fights, or PvE farming. This is more relevant for PvE and higher level arena fights where everyone has great runes, extremely tanky monsters, and long drawn fights.
    Violent vs Swift is not as big a concern in lower tiers anyway.

    I have also clearly mentioned that Violent is more RNG based which is an advantage for AD, and a disadvantage for short AO fights (still better for long AO fights though).
     
    Last edited: Dec 11, 2014
  4. BrFrPlayer

    BrFrPlayer Level 37

    Joined:
    Apr 10, 2014
    Messages:
    2,056
    Likes Received:
    661
    Point taken.

    However, I've seen the 'RNG is better in AD/auto' argument from players as high level as YDCB (from his vids) and I think this is false. The probabilities of procs don't change but the anticipation of getting good rolls is higher when we manual then when we auto because we're conditioned to regard the AI as being an idiot compared to us plus we don't remember bad Violent rolls that we don't see.

    This is coming from someone who's built two good Violent sets, one with cash. Violent sets are very good in their own right because one good streak is enough to seal a battle especially on units like Ahman or with big AoEs happening 2 or 3 times in a row but I would trade a godly Violent set for a godly Swift set any time.

    EDIT: I know that the coding probably has a p=0.2 for the Violent proc but as far as I can observe it tends to go off more than 20% of the time especially on Ahman.
     
  5. Gangsta

    Gangsta Level 7

    Joined:
    Sep 21, 2014
    Messages:
    51
    Likes Received:
    3
    You are not stupid my friend. I play in arena top 100. The problem with top arena teams is that they all have superb survival units with tones of HP, ressers like wind AA, wind DK, wind Epikon, DMG reduction like Darion, Choles, and other nasty things :)

    The easiest way to beat them is to bring THE RANDOM FACTOR. Additional turn = faster coold down refresh rate and additional dmg.
    In many cases violent proc makes the difference between enemy unit is killed or escaped with 5 % hp.

    As for "violent turns out to be faster" theory, you are right.
     
  6. Childish

    Childish Level 17

    Joined:
    Feb 8, 2014
    Messages:
    381
    Likes Received:
    98
    man i'm gonna have to re-read this read more than once:(

    "Violent, on the other hand will give you an effective speed/number of turns of (1 + 0.2 + 0.2^2 + 0.2^3 + ...) x (100 + 50) = 1.25 x 150 = 187.5"
    is the .2^2 20% of 20%? and if so how high does that first part of the equation get up.? sorry for the clarification but this is something i think the community can benefit from so trying to understand it 100%

    ~C.
     
  7. Ittadakkimaas

    Ittadakkimaas Level 38

    Joined:
    Aug 6, 2014
    Messages:
    2,386
    Likes Received:
    715
    I realize I have skipped a few steps in between - just didn't want to make the post any longer than it already was lol :)

    The calculation for Swift is obvious. 25% speed bonus with a base speed of 100 gives you an effective speed of 125, or 1.25 times as many turns.

    Now I want to compare this against the expected number of turns that Violent gives.
    On any given turn, with probability 0.2 (20%), Violent gives you an extra turn. Since with probability 0.8 you do not get the extra turn, and with probability 0.2 you get the extra turn, on average (the technical/mathematical term is "in expectation"), you get 0.2 extra turns.

    That brings us to expected 1 + 0.2 turns.

    But it doesn't end there. Suppose you get an extra turn (with 0.2 probability), now you have another 0.2 probability of getting a second extra turn. This happens with probability 0.2 x 0.2 (first 0.2 is for the first extra turn, and second 0.2 is for the second extra turn). This contributes another expected 0.2 x 0.2 = 0.2^2 turns.

    Similarly, if you look at the probability of getting a third extra turn, that contributes an additional 0.2 x 0.2 x 0.2 = 0.2^3 extra turns.

    So the total expected number of turns that Violent gives you is 1 + 0.2 + 0.2^2 + 0.2^3 + 0.2^4 + .... where this sum goes to an infinite number of terms. This is a standard geometric progression, whose sum is 1.25

    Another way to derive the same result is as follows: You get exactly 1 turn with probability 0.8 (probability 0.2 of continuing with additional turns). You get exactly 2 turns with probability 0.2 x 0.8 (0.2 probability to proceed after first turn, 0.8 probability to stop and not get any further extra turns). You get exactly 3 turns with probability 0.2 x 0.2 x 0.8 (0.2 for proceeding with extra turns the first two times, and 0.8 probability to stop at the third), and so on.
    This gives us an expected number of turns = Probability (1 turn) x 1 + Probability (2 turns) x 2 + Prob(3) x 3 + .... infinite terms
    It is easy to see that this also gives you the same geometric progression as my previous analysis.

    ------------------------------------------------------------------

    So, both Swift and Violent appear to be giving an expected number of turns = 1.25 times normal. INCORRECT!
    The crucial difference comes from the fact that Violent procs are on actual turns, where your speed is already boosted by runes. So suppose you have a base speed of 100, and a bonus +50 spd from substats/slot 2 etc. Then, you will be getting turns at an effective speed of 150. Each of those turns, now give you an expected 1.25 times as many turns! So, effectively you will be generating turns at the speed of 150 x 1.25 = 187.5!
    For the same case, Swift on the other hand will give you a 25% bonus speed on your base speed. So, a Swift set with the same stats will generate turns at a speed of (100 x 1.25) + 50 = 175!

    When your speed from stats (and not set bonus) is +50, Violent has an effective speed of generating turns of 187.5 as compared to Swift's 175. This difference is extremely significant for slow/long/tanky/heal-based fights. The difference will only go higher as the bonus speed from stats increases.

    ------------------------------------------------------------------


    Let me know if this is still unclear -- I would be happy to try and explain it better.
     
    AyeJames and Div like this.
  8. Div

    Div Level 35

    Joined:
    Jan 6, 2014
    Messages:
    1,620
    Likes Received:
    708
    Location:
    NYC
    i feel like ive been schooled 10 times over...

    my head is imploding upon itself

    great observations!!!
     
  9. BrFrPlayer

    BrFrPlayer Level 37

    Joined:
    Apr 10, 2014
    Messages:
    2,056
    Likes Received:
    661
    Both set bonuses go hand in hand IMO. Imagine a Violent set with 6 star Speed on 2 and disgusting Speed substats. Hard to imagine a mon that wouldn't dominate with that.
     
  10. Rotohomer

    Rotohomer Level 44

    Joined:
    Jan 12, 2014
    Messages:
    2,942
    Likes Received:
    1,865
    Location:
    California
    Medals:
    B-b-but what if the guy with swift goes first and kills your guy with violent? Then wouldn't swift be 1 turn vs violent's 0 turns :p ;)
     
  11. BrFrPlayer

    BrFrPlayer Level 37

    Joined:
    Apr 10, 2014
    Messages:
    2,056
    Likes Received:
    661
    I have never seen a guy with Swift able to kill anything in one turn. The only danger is if the Swift guy is an ATB booster that allows the non-Swift teammates to go right afterward with actual damage.
     
    venearas likes this.
  12. Ittadakkimaas

    Ittadakkimaas Level 38

    Joined:
    Aug 6, 2014
    Messages:
    2,386
    Likes Received:
    715
    Well, if you're using Violent, then its by definition *not* both set bonuses going in hand in hand :) this goes back to my statement -
    Lol, true :D

    Disclaimer: Swift has its place with teams that rely on first turn nuking/CC strategies. This is more relevant for tanky, heal spam arena fights, and for most PvE situations. Violent will give you more turns in expectation as compared to swift, but matches need to go on long enough for that difference to show.
     
  13. BrFrPlayer

    BrFrPlayer Level 37

    Joined:
    Apr 10, 2014
    Messages:
    2,056
    Likes Received:
    661
    I wonder if there's a limit to how much burst damage a team can put out and if it's lower than the cap for tankiness. If the latter is true, then that would be the true decider that Violent is better than Swift.
     
  14. AyeJames

    AyeJames Level 16

    Joined:
    Apr 13, 2014
    Messages:
    296
    Likes Received:
    89
    sooo should i start farming for 4-5* violent runes or keep the 6* runes i have on my perna and arnolds?? this should be moved to the guides section since its VERY informative and a good read too.
     
  15. Ittadakkimaas

    Ittadakkimaas Level 38

    Joined:
    Aug 6, 2014
    Messages:
    2,386
    Likes Received:
    715
    I don't know of a principled way, but for my own purposes, I like to use 20-25k HP as a threshold - seems to work for me. With 25k HP you can tank most nukes even if you get outsped, and heal back up. Even if your damage is not high with this much HP, if you have semi-decent heals, you should be able to keep the fight going and slowly chip away at the enemy.

    The arena is dominated by defense ignore nukers making HP more valuable. The threshold for PvE (Giants especially) is probably higher -- but there a mix of high HP and high Defense is better than pure HP. Having a mix of HP and Def makes you tanky enough to survive one-shots and helps your HP go a lot further.

    You can probably do the exact calculations to see your expected damage / heal by computing (expected number of turns) x (heal/damage) per turn, but it will be very painful and not fun at all to do so lol :) Personally I think just choosing 6* runes is better, at least for slots 2/4/6. You can slowly move to Violent as you find suitable runes in the shop / farm them from Dragons.

    Thanks for the vote of confidence :)
     
    Last edited: Dec 14, 2014
    AyeJames likes this.
  16. Darky

    Darky Level 15

    Joined:
    Sep 7, 2014
    Messages:
    313
    Likes Received:
    67
    Really Good job! I was thinking about all the question in game which you can get an answer through math and Statistics, like this swift vs violent or Rage vs Fatal, or at which crit rate it's efficient to replace Atk/Atk/Atk with Atk/Crit.Dmg/Atk or at what Damage it's best to replace Atk/Atk/Atk with Atk/Crit.C/Atk (For lushen, tyron or other sylphs for example). You can get the exact threshold with a 2nd or 3rd degree calculation which your variables are Atk, Crit.C and Crit.Dmg..
     
  17. Childish

    Childish Level 17

    Joined:
    Feb 8, 2014
    Messages:
    381
    Likes Received:
    98
    yeah my head officially blew up! i might have to summerize the summarized info lol! i Just play games and lift weights.
    ~C.
     
  18. Div

    Div Level 35

    Joined:
    Jan 6, 2014
    Messages:
    1,620
    Likes Received:
    708
    Location:
    NYC
    pffft...i just play games...stop being so productive
     
    Meefster likes this.
  19. Meefster

    Meefster Level 13

    Joined:
    Jul 7, 2014
    Messages:
    203
    Likes Received:
    42
    Location:
    UK
    but everything, when you get right down to it, is maths
     
  20. Ittadakkimaas

    Ittadakkimaas Level 38

    Joined:
    Aug 6, 2014
    Messages:
    2,386
    Likes Received:
    715
    To summarize:
    1) Swift is good for monsters that rely on getting the first turn in the arena, for example, shielders like Chloe, buffers like Wind Griff, CC monster like Water Sylph/Phantom Thief (although Despair is often a better choice for the last two)

    2) But, for long drawn fights, PvE, healers etc, what matters is the number of turns you get, and Violent beats Swift in that department. If a monster has a base speed of 'a' and gets speed 'b' from non-set bonus stats from a Swift or Violent set, keeping everything else the same:
    Swift will generate turns at a rate of (1.25 x a) + b
    Violent will generate turns at a rate of 1.25 x (a + b) which is 0.25 x b higher than Swift.
    So the higher your speed bonus from stats, the better Violent is as compared to Swift.

    3) Violent has a further advantage in AD that comes from its unpredictability - having random streaks makes it hard to predict and manage cooldowns, skill timings etc, making it a lot riskier to go up against Violent teams. With Swift teams you can always look at the attack bars and try to judge when to time your shields, other skills etc, but with Violent you never know what's going to happen.
     
    slopixi and AyeJames like this.

Share This Page