What's Better?

[Michelob]

regain earring is now beaten by ghillie +1 in uncapped attack situations. it also has 3 acc. I keep mine (acc+4 regain+1) for an accuracy build.

And what do you use for Smite... lol?

[NynJa]

The thread on FFXIAH mentions that Jambiyah is a slightly weaker version of STR Dagger. but does not say why/how. Can anyone explain to me why that is? Not great with mechanics/math here. xP

Just wondering if it's worth making this dagger or just going with the Jambiya. I have neither, and would prefer to make the better one even if it's only slightly better.

Its pretty close as is, but it also depends what you're pairing it with.

For the following, I'm not altering gear, though it shouldnt really matter. /dnc solo, not in abyssea, exneterator WS
Mandau95 paired against both those daggers, Thok wins.
Twash90 paired against, Aluh wins.
Paired with Corus (stronger in main always), same WS dmg, slightly lower DPS

You forgot to mention what your main weapon would be, but since you said DNC, that rules out Mandau. Either way, Thok seems to be slightly better for the most part.


On the topic of moonshade, I still havent turned mine in, any point to it on thief?

[Darkmagi]

The difference is so minimal that it is absolutely not worth making the Thokcha anymore though.

[NynJa]

The difference is so minimal that it is absolutely not worth making the Thokcha anymore though.

55 flame geodes at 50-80k a piece + 15 ifritites (2.75m to 4.4m) as well as all that trial time vs 1.5m Aluh

unless youre REALLY serious about the job, just go with the Aluh.

[pancakesandsex]

Its pretty close as is, but it also depends what you're pairing it with.

For the following, I'm not altering gear, though it shouldnt really matter. /dnc solo, not in abyssea, exneterator WS
Mandau95 paired against both those daggers, Thok wins.
Twash90 paired against, Aluh wins.
Paired with Corus (stronger in main always), same WS dmg, slightly lower DPS

You forgot to mention what your main weapon would be, but since you said DNC, that rules out Mandau. Either way, Thok seems to be slightly better for the most part.


On the topic of moonshade, I still havent turned mine in, any point to it on thief?

I use the attack / tp bonus for rudras.

[Michelob]

Been wondering about switching my current TP feet for MNK from AF3+2 to Thaumas Nails... good idea? Here is my current TP set:

Vere/x/x/potestas
ganesha's/rancor/ghillie+1/brutal
thaumas/af2+2/epona's/rajas
atheling/bb/thaumas/af3+2

Working on switching to rancorous/nefarious as well, but anything else you would change? Also was thinking of switching to a set that involves thaumas hat for TP, good idea?

[Taint]

regain earring is now beaten by ghillie +1 in uncapped attack situations. it also has 3 acc. I keep mine (acc+4 regain+1) for an accuracy build.

w

[pchan]

And what do you use for Smite... lol?

uncapped attack or acc : ghillie/+1 beats anything and if not vulcan does .
capped attack : vulcan's pearl.

TP bonus +25 is still shit even though diasetsu tested it to work on crit ws. Noobs generally see x% crit as an awesome stat. STR4 on a buffed victory smite beats 2% crit because at high ratio, 1 crit is only worth +46% more damage than a normal hit. At average ratio, 1% crit is about the same as 4 attack and I'm being generous. Nevermind the fact that attack is more consistent damage than crit.

You can probably find a setup where TP bonus moonshade beats both gillie+1 and vulcan's pearl but it's probably going to be very high str, medium/high attack at which point it's probably better to boost ?

[Churchill]

You can probably find a slew of setups where it wins. These same setups would also outparse anyone from france.

[pancakesandsex]

You can probably find a slew of setups where it wins. These same setups would also outparse anyone from france.

Flag down motenten and ask him add a "french? 1/0?" toggle to his spreadsheets so we'll know for sure.

[Kalmado]

55 flame geodes at 50-80k a piece + 15 ifritites (2.75m to 4.4m) as well as all that trial time vs 1.5m Aluh

unless youre REALLY serious about the job, just go with the Aluh.

This is why I stopped with my str dagger. Mainly because I'm a cheap ass. I also haven't unlocked exneterator so I haven't bothered with the AGI dagger either as well as being too cheap to even make that. I've considered making dex dagger just so I have something better than twilight/triplus (yes I know gimp) while not spending much. I'll probably bite the bullet and buy Aluh again.

[Motenten]

Re: Smite and Moonshade -- I'm not going to run a lot of samples, but for a gear set I have set up, Moonshade Att/TP earring beats: Str+4 earring when attack capped; Ghillie+1 with attack uncapped, acc capped; Ghillie+1 with attack uncapped, acc uncapped (checked down to a 78% hit rate). Moonshade still wins even when adding 30% crit rate for Champion's Drink, though the margin is pretty slim at that ponit. Moonshade still wins whether or not fStr is capped (when comaring with Str+4 earring). Checked vs DC Dyna mobs and Pil. Checked both lvl 90 and 99 Vere.

There's enough of a margin that I can't imagine any situation Moonshade would -not- win vs those other two options, unless you're weaponskilling at 300 TP. I even checked worst case crit rate increase scenario (TP bonus only giving 3% crit rate, rather than its average 3.75% crit rate), and Moonshade still won.

The only way it might not win is if there's an error in my modelling of weaponskill damage in the spreadsheet.

Note that the above is only for weaponskill damage. If you model full DPS (with uncapped acc), they go in order of:

MoonshadeRegainAcc(TP) + Ghillie+1 (WS)
Ghillie+1(TP) + MoonshadeAttTP (WS)
MoonshadeRegainAtt(TP) + Ghillie+1 (WS)
Aesir(TP) + MoonshadeAttTP (WS)

The margin between each of the above is about 0.6%, or about 2.2% from first to last. With capped accuracy they're all basically identical, except the MoonshadeRegainAtt TP config is about 0.5% ahead of the other three.

Not doing variants on this one.

Each of the Moonshade variants has a very situational 'best' result, if you really want something so specialized. For overall utility when accounting for other jobs, though, Att/TP is the best. You'd want a very specific reason for choosing something else.

[pchan]

The only way it might not win is if there's an error in my modelling of weaponskill damage in the spreadsheet.

Well I found different problems in the spreadsheet
1) you are using 25% anchor @ 200 TP when the devs said it was 20%
2) you are calculating the crit% from TP bonus as non integer value. It's a choice. Due to the testing we did a few years ago we know it's floored so it's probably more realistic to count it as 10%+floor((TP-100%)/10). I don't think it will matter much though.
3) In the spreadsheet you don't consider pdif but ratio, and it will have an impact when comparing setups with different crit rates. For instance with your choice, a crit makes your ratio go from 2 to 3. If you admit that my pdif model (which is nearly the same as yours as explained in the maths section) is correct for both non crit and crit, a non crit is worth on average 2,021527778 and a crit is worth 2.953518518. It's not very different but it acts as if you should add 0.93 ratio instead of 1 when a crit occurs, which necessarely unfavors crits setups.

A quick modification to your spreadsheet and I found the difference between both earrings much closer though TPbonus/moonshade still won by a hair. Might be due to the point 2) above. Nonetheless, regardless of the validity of point 2), if 2 ws sets are extremely close it's better to choose the most consistant one ( moonshade has like 8%-12% chance to proc per WS use, so technically vulcan's pearl beats it 90% of the time and once every new moon it's the other way around).

[Cyprias]

Well I found different problems in the spreadsheet
1) you are using 25% anchor @ 200 TP when the devs said it was 20%
2) you are calculating the crit% from TP bonus as non integer value. It's a choice. Due to the testing we did a few years ago we know it's floored so it's probably more realistic to count it as 10%+floor((TP-100%)/10). I don't think it will matter much though.
3) In the spreadsheet you don't consider pdif but ratio, and it will have an impact when comparing setups with different crit rates. For instance with your choice, a crit makes your ratio go from 2 to 3. If you admit that my pdif model (which is nearly the same as yours as explained in the maths section) is correct for both non crit and crit, a non crit is worth on average 2,021527778 and a crit is worth 2.953518518. It's not very different but it acts as if you should add 0.93 ratio instead of 1 when a crit occurs, which necessarely unfavors crits setups.

]A quick modification to your spreadsheet and I found the difference between both earrings much closer though TPbonus/moonshade still won by a hair. Might be due to the point 2) above. Nonetheless, regardless of the validity of point 2), if 2 ws sets are extremely close it's better to choose the most consistant one ( moonshade has like 8%-12% chance to proc per WS use, so technically vulcan's pearl beats it 90% of the time and once every new moon it's the other way around).

So you were wrong, got it.

[pchan]

The spreadsheet may still be wrong, the problem is that you, like all other BG trolls don't read the code behind it. Apprently I'm the only one who actually checks it w.

[NynJa]

You insisted youre right, corrected spreadsheet to coincide with your theory, and you were still wrong. Yet you still attempt to attack others and insist youre not wrong?

[Motenten]

1) you are using 25% anchor @ 200 TP when the devs said it was 20%

Official numbers. 20% was from the first iteration of the change, but was later relaxed. 25% is from the final changes.

2) you are calculating the crit% from TP bonus as non integer value. It's a choice. Due to the testing we did a few years ago we know it's floored so it's probably more realistic to count it as 10%+floor((TP-100%)/10). I don't think it will matter much though.

That's because the TP used per weaponskill is itself an average. Since actual TP used per weaponskill will vary a fair bit, it makes more sense to use the float-calculated crit value; using the floored value would be under-representing the TP effect.

3) In the spreadsheet you don't consider pdif but ratio, and it will have an impact when comparing setups with different crit rates. For instance with your choice, a crit makes your ratio go from 2 to 3. If you admit that my pdif model (which is nearly the same as yours as explained in the maths section) is correct for both non crit and crit, a non crit is worth on average 2,021527778 and a crit is worth 2.953518518. It's not very different but it acts as if you should add 0.93 ratio instead of 1 when a crit occurs, which necessarely unfavors crits setups.

This is a potential real issue. Obviously the spreadsheet's model was used because, when written, there was no meaningful means of converting from cRatio to average pDif. I haven't quite been motivated enough to plunge into pDif calculations again; the reference you gave, the wiki, is a mix of your formula and older formulas which are definitely wrong, so it's tedious to sort out.

[pchan]

Official numbers. 20% was from the first iteration of the change, but was later relaxed. 25% is from the final changes.

ok didn't catch this, which explains the difference.

That's because the TP used per weaponskill is itself an average. Since actual TP used per weaponskill will vary a fair bit, it makes more sense to use the float-calculated crit value; using the floored value would be under-representing the TP effect.

It would make more sense to calculate the rounded critical value per TP used on WS and then average based on the probably of it happening. It will lower the effect of TP bonus a little because you are more likely to get lower TP values than higher.

This is a potential real issue. Obviously the spreadsheet's model was used because, when written, there was no meaningful means of converting from cRatio to average pDif. I haven't quite been motivated enough to plunge into pDif calculations again; the reference you gave, the wiki, is a mix of your formula and older formulas which are definitely wrong, so it's tedious to sort out.

The average pdif on normal wiki work because they come from my model.

[Churchill]

and altogether definitely wrong

[Maguile]

Ogiers breeches vs warrior cuisses +2 vs ravagers cuisses +2 for resolution?