[X] My Assistant

I've never noticed any of my attacks being parried in 1H style with no overpower, but I'm sure a few must be and I don't notice, especially now that I've mostly transitioned into the "sweeping the screen" spread style. In DW rapier+waki with no overpower it is a bitch.

----------------------------------------------------------------------------
Wait, now that I look at Research for 1H again, the overpower section data looks suspicious. Why is the unstunned parry rate with no overpower only 9.2518%...it should be at least 10% with the PFUDOR bonus alone, right? And in fact we might expect most monsters to have a PFUDOR parry rate of 19%...

Parry = 1 - (1 - min(10 , (DEX/ 100) , (DEX - Level) / 75) / 100) * (1 - chaos_interception_rank * 0.5%)

scaled_stats = int(0.01 * base_stat * monster_level + (monster_level ^ 1.076675) * 0.3325)

Assuming PL 1000~2250 monster's base dexterity ≈ 200 the scaled DEX at level 456 as in the Research thread would be about 1150 which suggests most monsters have 9~10% parry before the PFUDOR bonus.

Also, overpower 5 reduced enemy parries from 9.2518% to 6.0557% which is a 6.0557/9.2518=0.6545, so his overpower 5 granted him 35% counter-parry?! It's supposed to be 20% counter-parry...
----------------------------------------------------------------------------

Nevermind, I'm an idiot. Overwhelming Strikes also provides counter-parry, it all makes sense.

O5 increases overall hit rate by about 1.3%. the numbers seem to be in clear favour of butcher and fatality, but rather than looking at the raw numbers, consider their actual effectiveness: how many times is a 2.3% increase in damage really going to make a monster take one less hit to kill, compared to the effect of 1.3% less attacks that miss completely?

Yeah from the beginning most of us thought that overpower's 1.3% is quite significant. Plus it gives more overcharge for spirit stance, and penetrated armors.

Uncle, what he's thinking is that overpower is surely at least a 1.3% improvement as on paper, but Butcher may not be the 2.3% paper improvement. And Fatality even more so. The reason is, suppose we had Daemon Duality 100. And we kill every monster in 1 hit. Now Butcher is useless, right?

So the question is, how close are we to the Daemon Duality 100 ultimate being? I might argue though, that we are still very far. Perhaps 2.3% to 2.2% far....

What fraction of enemies out there fall into a gap region where a 2.3% increase in damage never makes a difference? I would say, almost none. Because the randomness in our plain attack damage is big, like 1.5 times, plus there are criticals, penetrated armor, etc.

Until we are basically one hitting a significant fraction of monsters without a critical, we shouldn't devalue the 2.3% benefit of Butcher too much. But a super powerful 1H player might indeed one hit quite a few monsters with a critical, so it could be justified to significantly devalue the 2.3% benefit of Fatality. (And if Fatality is worthless then so is Savage armor).

One source of real world data is the DwD thread. I've been posting my times there in part for this purpose. So far, increasing my damage by a certain amount has yielded equal benefit in turns. It's only recently starting to plateau slightly. However DwD is not an ordinary arena. For me the plateau is caused by one hit Vital Strike overkills. Since my DwD times continued to improve linearly well after I became capable of overkills, that suggests we should not be too quick to devalue Butcher.

Oh and people may be too quick to dismiss Swift Strike, it's the trickiest to quantify. I might only be its second biggest supporter. Fudo Masamune found that Haste improves his clear times, at least in difficult challenges like grindfest and IW. Better defense and ease of playing increased his t/s more than the loss in turns. When I scaled his Haste down to Swift Strike, it turned out to be a slightly better improvement than Butcher.

assuming an ideal even distribution of monster HP and damage rolls, and a large enough sample size, the time saved by butcher is equal or very close to its damage boost, and i still consider butcher to be a very useful potency. but with hidden variables and thresholds involved, i can't be as sure of it having its full effect as with overpower, and with the additional benefits given by overpower the full effect becomes difficult to calculate and dependent on equipment choice. so in the end it's mostly speculation and personal preference, and everybody is wrong about everything

let's look at an extreme example:

monsters have between <<1% and 200% health, distributed evenly. your attack always deals 100% damage. you 1-hit 50% and 2-hit the other 50% of monsters. against n total monsters, the total number of hits is (1*0.5+2*0.5)*n = 1.5*n

now you gain 10% attack damage. instead if 1-hitting all monsters up to 100% health, you now 1-hit those up to 110%. this means you 1-hit 55% of monsters and 2-hit 45%. against n monsters, you need (1*0.55+2*0.45)*n = 1.45*n hits

the reduction in hits needed from 10% damage boost would be (1.5-1.45)/1.5 = 3.3%

This post has been edited by sickentide: Jan 14 2019, 12:35

What if you used your script to log all damage dealt, and also log the amount of HP overkill on the final blow of each monster? Could that give us a number that could we could use to calculate Butcher waste?

I guess that's not a correct method of analysis either. We might jump to the conclusion that monster A wasted X damage as overkill, yet monster B was underkilled and could have benefited from X more damage. Hmm.

I don't need calculations to know if my playstyle is satisfying or not. I just see how often I need to heal, how smooth it feels and how much time it takes me to clear a challenge. As for the effectiveness of overpower, I wasn't thinking about how it can help avoiding to waste turns and ultimately lead to a nasty SP blast in our face, but it does make sense. Haste is not stupid either in that matter, provided one can generate enough overcharge, but I'm not sure the overall clear time can get reduced that way.

At least I can confirm that a bullshit high defense saves quite a lot of time, which isn't necessarily true for 5-10% more damage.

I don't need calculations to know if my playstyle is satisfying or not.

ok, but i'm having fun with this. some more silly numbers:

monsters have 100 < health <= 300. your attacks deal 100, 2-hitting one half and 3-hitting the other half for 2.5*n total hits

raising damage by 50% lets you 1-hit 25% and 2-hit the rest for 1.75*n hits, or a time save of 30%

raising damage by 10% gives you (1*0.05 + 2*0.55 + 3*0.4)*n = 2.35*n, a 6% save

a 1% damage increase gives you (1*0.005 + 2*0.505 + 3*0.49)*n = 2.485*n, saving 0.6%

now someone tell me if this is close enough to the real situation and compare it to an ideal model where the probability of taking a set number of hits approaches 0 as said number approaches infinity and you can elegantly calculate the limit of total hits saved per damage increase to be exactly 1:1 as damage increase approaches 0

I don't need calculations to know if my playstyle is satisfying or not.

Therefore, and on the subject you were discussing earlier, you could do a real test by comparing OP5B4 vs (B5F4 or B4F5).

As you have another Peerless Shortsword Slaughter, you could just do the IW and upgrade at the same level. And for the tests, use an infusion of the same element of the other sword, so we have the same elements in both tests.

Therefore, and on the subject you were discussing earlier, you could do a real test by comparing OP5B4 vs (B5F4 or B4F5).

As you have another Peerless Shortsword Slaughter, you could just do the IW and upgrade at the same level. And for the tests, use an infusion of the same element of the other sword, so we have the same elements in both tests.

Err...
Someone got the funds for this research? Because not me. (IMG:[invalid] style_emoticons/default/heh.gif)

i was about to say, that's a pretty big "just"

another thing i've been considering is the IW for a DW club. common advice has it that you should go without overpower because stuns compensate for it, but i like to look at it this way:

instead of 3-hitting an unstunned target, you effectively 1-hit it if you land an initial stun. whether you need 1 or 3 successful hits, overpower is effective all the same. log data on this matter would be deceptive unless it distinguishes between stunning attacks and free wipe-up hits

Err...
Someone got the funds for this research? Because not me. (IMG:[invalid] style_emoticons/default/heh.gif)

Start pooling now. (IMG:[invalid] style_emoticons/default/tongue.gif)

This can be a long-term test.

I'm not in a hurry to see the results. I can wait. (IMG:[invalid] style_emoticons/default/heh.gif)

Overpower is probably more effective in DW club than in 1H, I'll return to this question sometime soon.

let's look at an extreme example:

monsters have between <<1% and 200% health, distributed evenly. your attack always deals 100% damage. you 1-hit 50% and 2-hit the other 50% of monsters. against n total monsters, the total number of hits is (1*0.5+2*0.5)*n = 1.5*n

now you gain 10% attack damage. instead if 1-hitting all monsters up to 100% health, you now 1-hit those up to 110%. this means you 1-hit 55% of monsters and 2-hit 45%. against n monsters, you need (1*0.55+2*0.45)*n = 1.45*n hits

the reduction in hits needed from 10% damage boost would be (1.5-1.45)/1.5 = 3.3%

Your example feature a strong loss in Butcher efficiency but let me try a similar example:

Monsters' health is between 0% and 220% distributed evenly, your attack always deals 100% damage. You 1-hit 45.45% of monsters, 2-hit 45.45% of monsters, and 3-hit 9.1% of monsters.

Then you gain 10% attack damage. Now you 1-hit 50% of monsters and 2-hit 50% of monsters. You shrink from 1*0.4545+2*0.4545+3*0.091 = 1.6365 hits/monster to 1.5 hits/monster. The improvement is (1.6365-1.5)/1.5 = 9.1%

Suppose you gained 8% damage. Now you 1-hit 49.1% of monsters, 2-hit 49.1% of monsters, and 3-hit 1.8% of monsters. You shrink from 1.6365 hits/monster to 3*(0.491+0.018)=1.527 hits/monster. Turns improvement is (1.6365-1.527)/1.527 = 7.2% nearly the same efficiency relative to 8% damage.

Suppose you gained 12% damage. Now you 1-hit 50.9% of monsters, 2-hit 49.1% of monsters. You shrink from 1.6365 hits/monster to 1*0.509+2*0.491 =1.491 hits/monster. Turns improvement is (1.6365-1.491)/1.491 = 9.76% a noticeable drop in efficiency relative to 12% damage.

Suddenly the Butcher efficiency is near full, and all it took was a rigged maximum monster health. Let's try another example:

Monsters' health is between 0% and 240% distributed evenly, your attack always deals 100% damage. You 1-hit 41.67% of monsters, 2-hit 41.67% of monsters, and 3-hit 16.67% of monsters.

Then you gain 20% attack damage. Now you 1-hit 50% of monsters and 2-hit 50% of monsters. You shrink from 1*0.4167+2*0.4167+3*0.1667 = 1.75 hits/monster to 1.5 hits/monster. The improvement is (1.75-1.5)/1.5 = 16.67%

Suppose you gained 10% damage. Now you 1-hit 45.83% of monsters, 2-hit 45.83% of monsters, and 3-hit 8.33% of monsters. You shrink from 1.75 hits/monster to 3*(0.4583+0.0833)=1.625 hits/monster. Turns improvement is (1.75-1.625)/1.625 = 7.7% a small drop compared to the previous result.

I don't know what the pattern is and I don't know if this line of examples are close enough to reality. Probably they aren't, as monsters don't have health anywhere near approaching zero, but it would be nice if this line of examples could be used as a starting point to figure out the real situation.

I don't know what the pattern is and I don't know if this line of examples are close enough to reality. Probably they aren't, as monsters don't have health anywhere near approaching zero, but it would be nice if this line of examples could be used as a starting point to figure out the real situation.

yes, this illustrates the difficulty in building an accurate model, and the further complication of different relative increases yielding different degrees of effectiveness

BlueWaterSplash, sickentide, I hear you like numbers >_>

Here is some data for your OP5 vs Fat5 discussion ( 1H Heavy ) [using PDUFOR arena The Trio and the Tree]. Not really clean, as stats and level (382-390) a little different.
Weapon is But4, Fat5. Heartseeker is enabled at the battle's start. Attack speed is 0%, Haste is off. Weapon has 2nd Infusion (different by day). Friday is marked. No OFC, Vital Strike only in the last round.
I think it should be possible to calculate potential decrease in monsters' parry count to rival Fat5 (let's say that every non-parried attack from OP5 is 100% critical). Compare it to theoretical decrease from OP5, answer "Is it worth it?" and later at some point compare it to OP5 in-game data (like month vs month at same level and equipment for the definitive answer.

 q.zip ( 7.44k ) Number of downloads: 10


This post has been edited by tox01: Jan 15 2019, 22:01

tried a new PA scheme for a mage:

STR = 0
DEX = Level - 100
AGI = Level
END = Level
INT = Level + 100
WIS = Level + 100

works good.

Hm, not bad. I am really proud of my improvements. (IMG:[invalid] style_emoticons/default/happy.gif)

This post has been edited by Uncle Stu: Jan 27 2019, 19:10

finally got to Lv. 350...
any feedback is welcome... (IMG:[invalid] style_emoticons/default/smile.gif)

finally got to Lv. 350...
any feedback is welcome... (IMG:[invalid] style_emoticons/default/smile.gif)

looks like you are doing this right. You can further improve with more slaughters but you know that already.

I need a better pants.

looks like you are doing this right. You can further improve with more slaughters but you know that already.

More Slaughter is not an improvement at Lvl350, unless all he does is arenas.

looks like you are doing this right. You can further improve with more slaughters but you know that already.

More Slaughter is not an improvement at Lvl350, unless all he does is arenas.

yeah, I plan to upgrade them to slaughter or at least the savage ones...
my priority is the gauntlets and the boots as their adb are too low IMO...

and currently I only doing arena and sometimes IW as I can't find the time to doing the GW...
I need ~6 hours to doing daily arenas afterall... (IMG:[invalid] style_emoticons/default/laugh.gif)

finally got to Lv. 350...
any feedback is welcome... (IMG:[invalid] style_emoticons/default/smile.gif)

Wow...May I ask how much credits it cost you to upgrade all those equipment( and getting them)? It's still a while for me to get to 350, but I can't imagine getting gears that good by then.