Offhand for almace

[Transmit]

This will be fun.

(Note - In my "Epeen" sig I wasn't using a STR magian)

[Prothscar]

Based on the total amount of damage it took to kill the same number of mobs, the Khanda set were obviously higher level by a substantial amount. This average higher level is going to push down weapon damage, WS damage, and spell damage.

That being said, the per strike damage is nearly identical (about 1 point of damage different). This means the more frequent ODD strikes are heavily skewing the averages up even if you are getting a huge number of small Khanda hits, as I expected.

After looking at WS fequency and numbers of hits, I think Khanda comes out as the clear winner because of the level correction working against it.

I expect the Khanda fights felt harder because of the slightly higher level that pushed damage down and monster HPs and damage up. The number of Dream Flower casts also indicate twice as many bad pulls during that period, so that probably led to longer resting times because of MP spent and more damage that had to be healed.

Interestingly enough, the number of Blank Gaze casts is almost the same, so either the mob used Cocoon a lot less on the Khanda set or the TP feed was not as bad as expected.

I think the take-home lesson is that the Str and Att aren't giving people substantial damage increases at 99 even if the numbers look very big. You can barely see the effect in this parse before accounting for level-correction, and it might not even be empirically provable after taking into account level-correction, and that's pretty damaging.

After looking at this raw data, I'm more inclined to look at other stats as a way to increase combat efficacy for this one slot. The stats that NA players focus on seem designed for high-damage epeen screenshots to put in their sigs and don't reflect overall effectiveness or usefulness to the party.

I think the take home lesson is you are in a severe case of denial. Not only did you get your inaccurate parse that you demanded, but you've now had no less than six people testifying as to why your sword choice is garbage compared to other options. You need a reality check.

You're looking for a way to rationalize your choice even though it's complete shit. What other evidence do you need? You've now had the calculations and the parse, on top of several people stating why you're wrong.

If you can seriously look at that parse and try to say that it proves your point, then I don't know what to tell you.

[Tashan]

Based on the total amount of damage it took to kill the same number of mobs, the Khanda set were obviously higher level by a substantial amount. This average higher level is going to push down weapon damage, WS damage, and spell damage.

No

That being said, the per strike damage is nearly identical (about 1 point of damage different). This means the more frequent ODD strikes are heavily skewing the averages up even if you are getting a huge number of small Khanda hits, as I expected.

The Mode, Median and Average melee hit for the Khanda test was:

+ 34: 352 ^ 37: 88 M+0.Avg: 47.11

The Mode, Median and Average for the STR Shamshir Test was:

+ 66: 143 ^ 68: 111 M+0.Avg: 69.94

By swapping to the Khanda the Mode was reduced by -48.48%; Median -45.59% and Avg: -32.64%. I'm not sure how you got to 1 point of damage difference.

After looking at WS fequency and numbers of hits, I think Khanda comes out as the clear winner because of the level correction working against it.

I don't see this. The Khanda test did 31.44% more CDC's than the Shamshir test, but only increased WS damage by 4.03%.

And to add further insult to injury, even though WS damage increased by 4.03% the following took place. The Khanda test did 72.01% more melee hits, it lost 3.39% of it's melee damage.

I'm not sure it can be explained anymore simply than this. The increased frequency is incredibly unbalanced.

I expect the Khanda fights felt harder because of the slightly higher level that pushed damage down and monster HPs and damage up. The number of Dream Flower casts also indicate twice as many bad pulls during that period, so that probably led to longer resting times because of MP spent and more damage that had to be healed.

In regards to personal feelings, I don't warrant them out of data but if I must humor it I'd disagree completely. The fights were not harder, they were longer. While there's no way to justify this idea that the majority of crawlers were of higher level there is something in front of us that can. It's elementary. Less damage was being dealt and thus it took longer.

There's nothing on there to accurately dictate the resting times on the parse.

Any bad pulls meant having to logout of the game because I'd be fighting Puks not Crawlers. While it's possible to link the crawlers, there are only 3 in the camp area. They were rarely all spawned at the same time.

Interestingly enough, the number of Blank Gaze casts is almost the same, so either the mob used Cocoon a lot less on the Khanda set or the TP feed was not as bad as expected.

Actually there is another possibility. It used other TP moves.

This data is under Defense but I must apologise, it looks like I saved recovery data as this for STR Shamshir one instead. You can view the Khandas one from the Defense page, and the STR if you can open an SQL file. Here's the TP moves used in the STR parse:

Ability Usage

Ability Used/Prepared
Cocoon 32/34
Poison Breath 35/34
Sticky Thread 29/30
Total 96/98

And for the Khanda:

Ability Usage

Ability Used/Prepared
Cocoon 39/40
Poison Breath 55/57
Sticky Thread 32/35
Total 126/132

I think the take-home lesson is that the Str and Att aren't giving people substantial damage increases at 99 even if the numbers look very big. You can barely see the effect in this parse before accounting for level-correction, and it might not even be empirically provable after taking into account level-correction, and that's pretty damaging.

To be honest, this would have been irrelevant had I chosen my own criteria for the parse. But to meet your arguements basis, I did as you requested on monsters higher than Abyssea Merit Party targets. As such, the parse was always going to be susceptible to the uncontrolled variable of Level-Correction.

However, this arguement still doesn't hold much ground because the idea of having a large sample for test is to work to the effect of minimizing random effects. Rather, the smaller the sample the greater the effect of randmness. This is precisely why I opted to do 100 monsters per test rather than say 10 or 20.

Has this worked? Certainly. The Offense Detail pages support this.

After looking at this raw data, I'm more inclined to look at other stats as a way to increase combat efficacy for this one slot. The stats that NA players focus on seem designed for high-damage epeen screenshots to put in their sigs and don't reflect overall effectiveness or usefulness to the party.

I'm British, but sure.

If I had to be inclined to take anything from this, what I would take from this is that you are not aware of how to evaluate correctly. This is kind of what I was saying a few weeks back in my first post on this thread.

And had "epeen" been the basis of any arguement in this thread I would have mentioned that the STR test pulled off a CDC that did 2130 damage as opposed to 1894.

However that would be stupid as it only happened once. It's unreliable data.

[Prothscar]

To go along with Tashan's synopsis I'd also like to point out that the difference between two monsters of the same type, within 3 levels of each other (the maximum that those crawlers could be apart), will be minimal outside of max HP. Max HP scales at a much higher percentage than base stats such as DEF, VIT, AGI, and STR. Usually, the AGI and VIT barely change if at all. Level correction would be in STR swords favor too, unless you don't know how to read and somehow think that the Khandga is a bastion of overcoming level corrected adversity.

Hint: It's quite the opposite.

So, Mr. Demonofhunger, just admit you're wrong and move on. Your sword is trash.

[Demonofhunger]

No


The Mode, Median and Average melee hit for the Khanda test was:

+ 34: 352 ^ 37: 88 M+0.Avg: 47.11

The Mode, Median and Average for the STR Shamshir Test was:

+ 66: 143 ^ 68: 111 M+0.Avg: 69.94

By swapping to the Khanda the Mode was reduced by -48.48%; Median -45.59% and Avg: -32.64%. I'm not sure how you got to 1 point of damage difference.

The Weaponskill and TP rates page lists the mean damage per hit at 21.3 and 22.15 respectively.

Also, why did you not remove the Cocoon from the Khanda set 8/40 times? Failing to do that is going to bring down the Khanda numbers.

As for the two sets, an explanation for the 58,702 extra points of damage that it took to kill the Khanda set is that they were overall higher level and thus had more HPs. Being higher level, the level-correction that takes place in the damage equations would have forced down all damage numbers and lead to longer fights as compared to the Shamshir set.

Of course, you might have just been wasteful of TP and/or MP on the Khanda set and that extra damage comes from wasting the damage on mobs that were close to death. Wasting damage like that would be a different reason for why the fights might have taken longer.

From the spell usage it doesn't look like the fights were too much longer if we assume you used the spells as often as possible, so the extra 50Kish WS damage had to go somewhere.

That being said, a sample size of 100 is not large enough for the very small numbers we are pulling out of the data. This was always going to be a very rough estimate and it's ambiguous enough to resolve the matter for me.

[Prothscar]

You realize that you just said ambiguity solved the matter for you? Whatever little credit you had is now gone. You're absolutely hopeless.

[Tashan]

Good morning!

The Weaponskill and TP rates page lists the mean damage per hit at 21.3 and 22.15 respectively.

That's TP gain rate, not damage.

Also, why did you not remove the Cocoon from the Khanda set 8/40 times? Failing to do that is going to bring down the Khanda numbers.

If it wasn't removed it's because it couldn't be removed (It was dying during an animation such as a a spellcast or skillchain.)

---

Looks like we're really scraping the barrel now:

--> Level Correction as an explanaition for the difference in damage.

To go along with Tashan's synopsis I'd also like to point out that the difference between two monsters of the same type, within 3 levels of each other (the maximum that those crawlers could be apart), will be minimal outside of max HP. Max HP scales at a much higher percentage than base stats such as DEF, VIT, AGI, and STR. Usually, the AGI and VIT barely change if at all. Level correction would be in STR swords favor too, unless you don't know how to read and somehow think that the Khandga is a bastion of overcoming level corrected adversity.

--> Wasted TP on overkills makes fights take longer, yet spell damage made it not.

Self contradiction, ok.

Have you looked at how long the fights took? I have given you the data:

Khanda: Avg.Time/Fight: 338.99 s
Shamshir: Avg.Time/Fight: 136.70 s

That being said, a sample size of 100 is not large enough for the very small numbers we are pulling out of the data. This was always going to be a very rough estimate and it's ambiguous enough to resolve the matter for me.

1,000,000+ damage is a small number.

Ohkay.

[Prothscar]

Sir, the data is clearly ambiguous enough to draw a conclusion according to Mr. Hunger. We need not argue further.

[Tashan]

True that.

[Demonofhunger]

1,000,000+ damage is a small number.

Ohkay.

No, but 100 is small number.

Since the 100 mobs for each sample has a level that varies by as much as three, if one set is higher on average then one set of mobs is going to be harder to kill because they take less damage from the same attacks and spells and have more HPs.

Do you know how level-correction works? You do less damage to mobs as their level increases because they have higher stats as their level increases. The extra damage that it took to kill the Khanda set is clear evidence that they were higher level because the idea that you somehow wasted over fifty thousand points of damage on inept spell and WS use (over 5%) is laughable.

As for your times, I honestly don't have an explanation. If the times were true and the fights took twice as long, you should have used a lot more BLU and NIN spells. Since you clearly didn't, it looks like the calculation must also take into account resting times or you were fighting with less than full MP so you would run out in the middle of the fight or doing something else to make the fights harder for yourself. It might have just been the need to cure yourself wasted so much MP that you couldn't keep up spell casting the whole time and so artificially prolonged the fights. I don't really know what your fighting style is or if you were giving it your full effort or simply did the Khanda set in a lazy and half-hearted fashion.

Heck, if the battle had actually taken twice as long then you should have taken close to twice as much damage. This and many other issues make me not believe that your times are accurate.

That being said, even with the massive issues skewing with the parse, the overall issue is still very close. That ambiguity leaves only one conclusion: the parse issues negatively affecting the Khanda set are probably the only reason it looks slightly worse, and if they aren't then the two swords are too close to call.

Still, both of you should feel free to keep being rude and dismissive. It reveals a lack of confidence in your analysis that will keep damaging your credibility long after you've forgotten about this thread.

Thanks again for providing the parse data, and please don't feel the need to respond since I won't be checking this thread any more.