What's Better?

[Byrthnoth]

So, the math I've done indicates that any time you're /WAR, Hagun is better both in terms of DPS and WS damage.

If you're /DRG (with both sets), then Soboro will win.

Here is the excel sheet I made:
http://www.megaupload.com/?d=X7M6546N

You'll notice that you can't vary Haste or Delay. In my opinion, they aren't relevant for this comparison because you would make the same decisions regarding them regardless of which weapon you're using. This makes it impossible to compare SAM/DRG Soboro to SAM/WAR Hagun. You could include Haste with a minimum of effort, I just didn't feel like it because I'm not particularly invested in the Samurai job let alone this specific weapon comparison.

Things that aren't accounted for:
1) Zanshin - Doesn't seem like a big deal, but likely gives Hagun an effective 97-98% Accuracy on first-swings that weren't already Double Attacks and doesn't give that boost to Soboro.
2) Job Ability/Weaponskill Delay - Soboro's WS frequency is ~50% higher than Hagun's. This is a huge factor, especially considering Soboro is "better" under very high haste conditions.
3) Berserk's effect on DoT isn't considered.

Funny things: Hagun with 17% DA actually has higher DoT than Soboro.


Still, I'm just going to feel like I'm waving my hands this way and then waving them that way until someone quantifies the JA/WS delays.


Edit: By the way, I think I might edit that document to compare Fort to the Mag DA GA some time this week. =p

[ringthree]

So, the math I've done indicates that any time you're /WAR, Hagun is better both in terms of DPS and WS damage.

If you're /DRG (with both sets), then Soboro will win.

Here is the excel sheet I made:
http://www.megaupload.com/?d=X7M6546N

You'll notice that you can't vary Haste or Delay. In my opinion, they aren't relevant for this comparison because you would make the same decisions regarding them regardless of which weapon you're using. This makes it impossible to compare SAM/DRG Soboro to SAM/WAR Hagun. You could include Haste with a minimum of effort, I just didn't feel like it because I'm not particularly invested in the Samurai job let alone this specific weapon comparison.

Things that aren't accounted for:
1) Zanshin - Doesn't seem like a big deal, but likely gives Hagun an effective 97-98% Accuracy on first-swings that weren't already Double Attacks and doesn't give that boost to Soboro.
2) Job Ability/Weaponskill Delay - Soboro's WS frequency is ~50% higher than Hagun's. This is a huge factor, especially considering Soboro is "better" under very high haste conditions.
3) Berserk's effect on DoT isn't considered.

Funny things: Hagun with 17% DA actually has higher DoT than Soboro.


Still, I'm just going to feel like I'm waving my hands this way and then waving them that way until someone quantifies the JA/WS delays.


Edit: By the way, I think I might edit that document to compare Fort to the Mag DA GA some time this week. =p

Someone posted some testing information from Alla about WS and JA delays recently, but I don't remember where on BG it was posted.

[Byrthnoth]

Someone posted some testing information from Alla about WS and JA delays recently, but I don't remember where on BG it was posted.

Bah, although I said I wanted it I was actually lying! Results become much more specific when you start including WS delays, as your total Haste (Buffs, gear, etc) and delay reductions (Sword strap) suddenly matter. I might look for it later, but for now I am at work. >.<;

[Thorny]

Someone posted some testing information from Alla about WS and JA delays recently, but I don't remember where on BG it was posted.

If it's the Shamaya stuff, it's not accurate. Using JA0Wait and spamming JAs nonstop won't slow down your attack more than the last one would, implying if there is a wait it's somehow tied to animation delay. This in itself is a fallacy though, as server side cannot see your animation delay and I would hope melee delay is always server side.

[mdkuser]

No, this is entirely incorrect, this is why I asked for a parse of your Pachipachio, and this is why a majority of the math done by you and Pchan on these forums is incorrect. All of your math assumes that one will weaponskill the second they hit 100TP every time. The only TP overflow you ever account for is the double attack probability distribution, which is incorrect. The fact of the matter is, in things like Pachi vs Hagun, any human error begins to favor Hagun, to the point that Pachi no longer wins when you put a human being behind the controls. Things like not WSing at 5%, or taking an extra swing in a high-haste situation as you move to get overwhelm, or just being distracted for a fraction of a second and swinging an extra time - these things do happen, even to the best players. Using your personal TP/WS(and ideally meditate WSes separate from TP'd WSes) split will give you what is actually best for you, as a human being in a real situation. Using what you use will actually not result in you optimizing your damage. This also accounts for missing overwhelms (if you miss them regularly, this will skew your TP/WS split, and also does influence it...if you have a large sample, you can assume that you will probably keep a similar ratio of missing overwhelm in the future).

It is much better to gather data from your personal play, then apply theoretical to it, than to use purely theoretical numbers from start to finish.

you're just mad and are desperate for arguments towards hagun. either you do the maths or parse, do not combined both. The best weapon is teh one that is going to do teh best damage in controlled conditions. I laughed inside when I saw your analysis on Rindomaru ( and pachipachio it seems), your conclusion can be summed up in the following statement

"Hagun is all the better, the more you go afk"

kk thnxby.

ringthree joined your fanclub too.

[Byrthnoth]

If it's the Shamaya stuff, it's not accurate. Using JA0Wait and spamming JAs nonstop won't slow down your attack more than the last one would, implying if there is a wait it's somehow tied to animation delay. This in itself is a fallacy though, as server side cannot see your animation delay and I would hope melee delay is always server side.

Well, I know I can go Waltz II -> Samba -> Waltz II between swings on WAR/DNC with a 504 GA and 19% Haste in gear. Waltz is an 8 second recast and it takes me 6.8 seconds to swing, so there's some kind of delay there.

[FailureMidgard]

So, the math I've done indicates that any time you're /WAR, Hagun is better both in terms of DPS and WS damage.

If you're /DRG (with both sets), then Soboro will win.

Here is the excel sheet I made:
http://www.megaupload.com/?d=X7M6546N

You'll notice that you can't vary Haste or Delay. In my opinion, they aren't relevant for this comparison because you would make the same decisions regarding them regardless of which weapon you're using. This makes it impossible to compare SAM/DRG Soboro to SAM/WAR Hagun. You could include Haste with a minimum of effort, I just didn't feel like it because I'm not particularly invested in the Samurai job let alone this specific weapon comparison.

Things that aren't accounted for:
1) Zanshin - Doesn't seem like a big deal, but likely gives Hagun an effective 97-98% Accuracy on first-swings that weren't already Double Attacks and doesn't give that boost to Soboro.
2) Job Ability/Weaponskill Delay - Soboro's WS frequency is ~50% higher than Hagun's. This is a huge factor, especially considering Soboro is "better" under very high haste conditions.
3) Berserk's effect on DoT isn't considered.

Funny things: Hagun with 17% DA actually has higher DoT than Soboro.


Still, I'm just going to feel like I'm waving my hands this way and then waving them that way until someone quantifies the JA/WS delays.


Edit: By the way, I think I might edit that document to compare Fort to the Mag DA GA some time this week. =p

I'm unsure what your exact point is with this sheet, and I guess it's marginally useful, but you're drawing incorrect conclusions from it. I went ahead and added a few things to your sheet. Since we agree Hagun is already winning /war, I'm going to ignore that for now and just look at /drg, so I changed DA to 5% for Soboro and 7% for Hagun.

For Soboro, I added the following cells:
WS DPS : =60*B47/(B38*B33) = 9.9586
Meditate DPS : =(150+16.9)*B47/(100*150) = 2.6366
Total DPS = 24.4189 (WS+Meditate contribution : 12.5952 )

For Hagun:
WS DPS : =60*B33/(B24*B20) = 9.4139
Meditate DPS : =(150+16.9)*B33/(100*150) = 3.9259
Total DPS = 24.5892 (WS+Meditate contribution : 13.3398 )

The difference in average WSes more than makes up for the difference Soboro gets from autoswing damage/WS rate. This of course is all assuming a constant pdif, meaning attack capped on normal swings. Regardless of that though Hagun according to this has both a higher white swing DoT, and a higher total WS contribution, meaning regardless of what you do to pdif, Hagun will win. Admittedly, this so far is assuming all the TP is used perfectly, but when I subtract off ~20 TP as wasted (and not increasing ftp), Hagun is still coming out ahead, and realistically I don't see that much even wasting under the perfect conditions that your spreadsheet is simulating.

[Thorny]

Well, I know I can go Waltz II -> Samba -> Waltz II between swings on WAR/DNC with a 504 GA and 19% Haste in gear. Waltz is an 8 second recast and it takes me 6.8 seconds to swing, so there's some kind of delay there.

I don't think anyone is denying that the wait exists, it's just a matter of getting it right.

[Clerix]

What's better for BRD debuff:

Omega/Nereid or Omega/Angel's?

Obviously Nereidx2 for buff.

^

[FailureMidgard]

you're just mad and are desperate for arguments towards hagun. either you do the maths or parse, do not combined both. The best weapon is teh one that is going to do teh best damage in controlled conditions. I laughed inside when I saw your analysis on Rindomaru ( and pachipachio it seems), your conclusion can be summed up in the following statement

"Hagun is all the better, the more you go afk"

kk thnxby.

ringthree joined your fanclub too.

You really don't get it do you? A parse will never tell if X or Y is better (unless there is a huge discrepancy), there are too many external factors. That being said, a parse is a good baseline for how you generally perform. If your Meleeswings:Weaponskills ratio is always 5.3-5.4 before meditate, why would you do math to figure out what it is under perfect conditions, when you obviously do not play under perfect conditions? If you meditate as you are running between "camps" in salvage, why would you not add this into your calculations? You are not swinging your Soboro as you run, but your meditate timer is indeed ticking down. This -does- apply to clearing the zone faster, doing more damage, etc.

The fact of the matter is, if I look at one of your parses, I will not get the split that you are claiming in your math. This is pure fact. So, that leaves you with 3 options:
a. You claim that the relative value of the 2 stats being compared is the same regardless of TP/WS split. Aka, Pachi vs Hagun maintains the exact same ratio/etc if its 5hit pachi vs 6hit hagun or 6hit pachi vs 7hit hagun.
b. You admit that relative value of the stats being compared is in fact a variable based on human error, but realize that human error is not that hard to quantify with a parse, and furthermore realize that from this you can perform a more accurate calculation for what further stat increases will do for YOU.
c. You feel that math is a huge waste of time!

I hate to break it to you, but b is the best and most realistic way to go about things. You don't just calculate what something -should- be. You test and verify it is what you think it is, and when it doesn't match the theoretical, you discard anything theoretical until you can get it to match.

Don't worry though, if we took your method and applied it to real life, human reaction time would be 0, this would be assumed in the production of all car breaks, and we'd have a lot more car wrecks. Why would we ever bother to test human reaction time and include that in calculations dealing with how fast we want a car to be able to stop? Preposterous! I'd go on, but there really is no point. Inb4 lolrlanalogy.

[Thorny]

Don't worry though, if we took your method and applied it to real life, human reaction time would be 0, this would be assumed in the production of all car breaks, and we'd have a lot more car wrecks.

Just make sure you've got a WHM out of aoe for stona.

[Byrthnoth]

I'm unsure what your exact point is with this sheet, and I guess it's marginally useful, but you're drawing incorrect conclusions from it. I went ahead and added a few things to your sheet. Since we agree Hagun is already winning /war, I'm going to ignore that for now and just look at /drg, so I changed DA to 5% for Soboro and 7% for Hagun.

For Soboro, I added the following cells:
WS DPS : =60*B47/(B38*B33) = 9.9586
Meditate DPS : =(150+16.9)*B47/(100*150) = 2.6366
Total DPS = 24.4189 (WS+Meditate contribution : 12.5952 )

For Hagun:
WS DPS : =60*B33/(B24*B20) = 9.4139
Meditate DPS : =(150+16.9)*B33/(100*150) = 3.9259
Total DPS = 24.5892 (WS+Meditate contribution : 13.3398 )

The difference in average WSes more than makes up for the difference Soboro gets from autoswing damage/WS rate. This of course is all assuming a constant pdif, meaning attack capped on normal swings. Regardless of that though Hagun according to this has both a higher white swing DoT, and a higher total WS contribution, meaning regardless of what you do to pdif, Hagun will win. Admittedly, this so far is assuming all the TP is used perfectly, but when I subtract off ~20 TP as wasted (and not increasing ftp), Hagun is still coming out ahead, and realistically I don't see that much even wasting under the perfect conditions that your spreadsheet is simulating.

Okay, add Meditate to the list of things I didn't account for. I basically didn't want to account for anything with a timer because it would mean that I'd have to start comparing Haste totals. However, if you include meditate then you should also include Sekka, where Hagun doesn't get its TP Bonus.
Another thing you can add to the list of things I didn't account for is WS Double Attacks. They would have uncapped accuracy and their hit rate would depend on your WS set but likely wouldn't be 20%.

Finally, Hagun will likely be benefited by wasting TP over-TPing, not Soboro. If you waste attack rounds with Soboro you're pushing up your fTP in a lower slope range of Gekko than if you waste attack rounds with Hagun. The relatively minor difference in DoT will be compensated for and likely exceeded by the difference in fTP gained and dramatic decrease in Soboro WS frequency.

[FailureMidgard]

Okay, add Meditate to the list of things I didn't account for. I basically didn't want to account for anything with a timer because it would mean that I'd have to start comparing Haste totals. However, if you include meditate then you should also include Sekka, where Hagun doesn't get its TP Bonus.
Another thing you can add to the list of things I didn't account for is WS Double Attacks. They would have uncapped accuracy and their hit rate would depend on your WS set but likely wouldn't be 20%.

Finally, Hagun will likely be benefited by wasting TP over-TPing, not Soboro. If you waste attack rounds with Soboro you're pushing up your fTP in a lower slope range of Gekko than if you waste attack rounds with Hagun. The relatively minor difference in DoT will be compensated for and likely exceeded by the difference in fTP gained and dramatic decrease in Soboro WS frequency.

Oh wow, you're right, I'm an idiot and ignored haste (I thought it looked high). In any case though, meditate and human error both add a bit more to Hagun as you said though.

[CDF]

No one is arguing that a comparison of the maximum potential of two options in a hypothetical controlled situation should be held as what will actually happen in typical practice. Certainly, it would be an interesting exercise to compare theoretical TP:WS ratios to what actually happens, as a way to validate or invalidate theoretical comparisons. It is also a reminder that the consequences of "human error" may be disproportionate between two weapons. Letting an extra hit slip through in a max haste situation... what a "problem" to have.

A better argument against assuming a TP:WS ratio for any comparison (in terms of percent difference) is that it's gratuitous at best. Holding it fixed for two weapons being compared implies that the percent difference in "TP-phase" damage would be the same as the percent difference in WS damage... false in general.

[ringthree]

you're just mad and are desperate for arguments towards hagun. either you do the maths or parse, do not combined both. The best weapon is teh one that is going to do teh best damage in controlled conditions. I laughed inside when I saw your analysis on Rindomaru ( and pachipachio it seems), your conclusion can be summed up in the following statement

"Hagun is all the better, the more you go afk"

kk thnxby.

ringthree joined your fanclub too.

I think that most people would drop Hagun for the best weapon whatever it might be. The only ones that I have seen continually claim that Hagun is not the best under most circumstances is you and pchan.

Remember, I was the one that said, on Greater Colibri, that Rindomaru looked like it was a better option than Hagun. And I did that based on a true number of rounds, not estimations.

It was very close, and the reason that most people still use Hagun (which is also the reason you and pchan chose to ignore) is that it assumed a perfect world with no delay. The difference was so small that adding in that delay favored Hagun again. And that is not even the human delay that Failure is talking about, that is the absolute in-game delay between last swing and next WS. Failure is completely correct that there are human factors that must be included to get a correct value, but you also ignore the in-game delays that exist that always favor Hagun.

[ringthree]

Okay, add Meditate to the list of things I didn't account for. I basically didn't want to account for anything with a timer because it would mean that I'd have to start comparing Haste totals. However, if you include meditate then you should also include Sekka, where Hagun doesn't get its TP Bonus.

I would like to question the role of Sekkanoki. Meditate should definitely be included, but if you are always using a WS as soon as you cross the 100TP threshold, then you will never get a chance to use Sekkanoki.

Am I wrong in this assessment? It would seem counter-productive to hold 200TP just to use Sekkanoki and lose the TP bonus on the first WS, when you could just ignore Sekkanoki and just WS when you hit 100TP each time.

This is assuming you only use Tachi: Gekko, of course. I am not factoring using Sekkanoki to set up a skillchain with lower damage WS.

[FailureMidgard]

No one is arguing that a comparison of the maximum potential of two options in a hypothetical controlled situation should be held as what will actually happen in typical practice. Certainly, it would be an interesting exercise to compare theoretical TP:WS ratios to what actually happens, as a way to validate or invalidate theoretical comparisons. It is also a reminder that the consequences of "human error" may be disproportionate between two weapons. Letting an extra hit slip through in a max haste situation... what a "problem" to have.

A better argument against assuming a TP:WS ratio for any comparison (in terms of percent difference) is that it's gratuitous at best. Holding it fixed for two weapons being compared implies that the percent difference in "TP-phase" damage would be the same as the percent difference in WS damage... false in general.

Eh, you may not be arguing that, but mdk and pchan always have. They feel that if a certain combination has the highest theoretical, that any human error or things not being accounted for must affect all things being compared equally. While I make no claims that my method is 100% correct or foolproof, it is more correct than assuming the best-case theoretical scenario - that is to get basic values from your parse such as #of hits:#of WSes:#of meditates. This is something that will never be the exact numbers that pchan and mdk claim, but at the same time if you look at 1 person across multiple parses, it will generally maintain a similar ratio for that individual, though it may differ from someone else.

My point is that theoretical comparisons are nice and all, I'm not trying to invalidate their existence, I'm merely invalidating them as a way to choose what is best unless one can mimic the model being created, which I'm relatively sure none of us can.

[FailureMidgard]

I would like to question the role of Sekkanoki. Meditate should definitely be included, but if you are always using a WS as soon as you cross the 100TP threshold, then you will never get a chance to use Sekkanoki.

Am I wrong in this assessment? It would seem counter-productive to hold 200TP just to use Sekkanoki and lose the TP bonus on the first WS, when you could just ignore Sekkanoki and just WS when you hit 100TP each time.

This is assuming you only use Tachi: Gekko, of course. I am not factoring using Sekkanoki to set up a skillchain with lower damage WS.

If you really want to be exact, Soboro also has the option to over-TP and Shiki TP to a warrior or someone with a higher WS average, not that anyone would boost someone else's numbers! Also, if Soboro is at ~85TP and lands a triple proc, you can Sekka that WS just to save 2 hits next round, whereas Hagun doesn't really benefit from that.

There's all sorts of stuff that's just really hard to properly analyze.

[ringthree]

If you really want to be exact, Soboro also has the option to over-TP and Shiki TP to a warrior or someone with a higher WS average, not that anyone would boost someone else's numbers! Also, if Soboro is at ~85TP and lands a triple proc, you can Sekka that WS just to save 2 hits next round, whereas Hagun doesn't really benefit from that.

There's all sorts of stuff that's just really hard to properly analyze.

God, the mind boggles at the calculations involved in that. LOL

That said, and I am asking seriously here, can you think of a time when you would seriously use Sekkanoki with a Hagun (besides at the beginning of a fight when you have 300TP stored up)? Meditate would just mean that your next WS came faster, not that you would hold the TP to use it with Sekkanoki.

[CDF]

My point is that theoretical comparisons are nice and all, I'm not trying to invalidate their existence, I'm merely invalidating them as a way to choose what is best unless one can mimic the model being created, which I'm relatively sure none of us can.

That is a strange position, as people use theoretical comparisons all the time to guide decision-making (other factors like value also enter into it obviously). It is not impossible to approximate human error with "modeling" either. For example, you could say that "X% of the time I waste Y number of rounds holding TP" and then adjust the number of rounds accordingly," though accounting for effects on TP bonus might be tedious.