Bards and BL and Crit and RoB

Hey everyone,

I was a bit bored, so I wrote a little program to calculate the effect of critical chance on RoB procs, and the amount of BL you do.

The simulation ran over the course of 360 seconds (6 minutes), with a base crit rate of 15% (~500 crit), and assumed IR usage.

100000 trials were run, and here are the results:

Base Critical Rate: 15.0
Trials: 100000 Procs: 3308577 BLs: 4626290
Average Procs: 33.08577 Average IR Procs: 11.98838 Average BLs: 46.2629 Average IR BLs: 11.98838
Total seconds saved: 24542659 Average Seconds saved: 245.42658999999998 Average Seconds saved per Proc: 7.417889624451841

EDIT: the original model was wrong, but this should be right. Pretty graphs incoming.

# of BL's and RoB procs as Crit increases:

http://i.imgur.com/2WizHQ6.jpg

Rate of Increase of # of Procs and BLs as Crit increases:

http://i.imgur.com/3Ns157u.jpg

Quote:

---

The simulation ran over the course of 360 seconds (6 minutes), with a base crit rate of 15% (~500 crit), and assumed IR usage.

100000 trials were run, and here are the results:

Trials: 100000 Procs: 6034321 BLs: 6556674

---

So if I'm interpreting correctly, your 15% crit rate simulation resulted in 360 * 100,000 / (6556674) = 0.1821 bloodletters per second.

The assumed uptime for both WB and VB is 100%? And the simulation is based on 2 chances to crit simultaneously every 3s? 100% Straight Shot uptime for 10% more crit?

Quote:

---

Originally Posted by EasymodeX

So if I'm interpreting correctly, your 15% crit rate simulation resulted in 360 * 100,000 / (6556674) = 0.1821 bloodletters per second.

The assumed uptime for both WB and VB is 100%? And the simulation is based on 2 chances to crit simultaneously every 3s? 100% Straight Shot uptime for 10% more crit?

---

Yes, the assumptions are

• 100% uptime for VB, WB, SS.
• 2 simultaneous proc chances.
• You only BL during offGCDs

Actually for the last assumption, I also, on a whim, tested what would happen if you assumed you could BL every 10th of a second, ignoring all other attacks and all animation locks.

Guess what happened? You gained between 0 and 1 BLs over the course of 360 seconds.

Safe to say, you lose no dps by only using BL on off GCD windows.

I hope that finally puts that discussion to rest.

Also: informative chart and effect of increasing Crit on RoB procs in a minute, once I get them uploaded.

Just to verify, when IR is down, each tick is resulting in a 22% chance for a BL proc right?

Quote:

---

Originally Posted by EasymodeX

Just to verify, when IR is down, each tick is resulting in a 22% chance for a BL proc right?

---

I assumed SS added a flat 10%, so I used .25.

Was that wrong? I haven't played archer in a while.

If thats wrong, then what does IR add? Maybe I got that incorrect as well.

No, I have a model that's structured significantly different than what you're doing. I was just checking to see that the # of bloodletter procs were in the same vague ballpark. I'm getting about half the # of BL procs that you are.

My model looks at a base 15% crit rate, adds 10% for straight shot and a modeled 5% (20% * 15s / 60s) for IR when IR applies (I'll ignore IR for now).

At a 25% crit rate, every DOT tick has 2 chances to crit*. The chance for a crit and therefore a "ROB-eligible" tick is: (1 - (1 - 25%) * (1 - 25%)) = 43.73%*. Half of that is 21.86% chance for a ROB proc*.

* These are all based on the server-side DOT tick assumption.

I ignore hardcasts and only assess proc bloodletters.

That is a 22% ROB'd BL every 3 seconds, or 0.073 BL per second. With a modeled IR, that is 0.085 BL per second.


Your 15% simulation has 6556674 / (360s * 100,000) = 0.182 BL per second, which is significantly different.

So, I'm trying to see whether the simulation / hardcasts / animations is causing such a difference, or if it's something in the way I'm incorporating BL in my model. Or maybe your sim ^_^.

Oh, I know exactly what the problem is.

I didn't only have a 50% chance of proccing.

Lemme go fix that.

EDIT: its been fixed, the new numbers look like

Base Critical Rate: 15.0
Trials: 100000 Procs: 3308577 BLs: 4626290
Average Procs: 33.08577 Average IR Procs: 11.98838 Average BLs: 46.2629 Average IR BLs: 11.98838
Total seconds saved: 24542659 Average Seconds saved: 245.42658999999998 Average Seconds saved per Proc: 7.417889624451841

So its about a 0.09 BL per second, which is on the same level as your model.

What is interesting about the sim is that each proc saves you on average over 7 seconds on your BL...thats pretty impressive.

Lol, I did the same thing when I originally edited my formulas.

I fixed it...also I uploaded some graphs.

Whats really interesting though is the rate of increase.

As you increase crit, each crit is valued less and less for RoB proc increases, however...they are valued more and more for total BL increases.

I have NO idea why this is the case, but suffice to say it seems the more crit you have, the more you want crit.

Go go crazy Bards.

Quote:

---

Originally Posted by pandabearcat

I have NO idea why this is the case, but suffice to say it seems the more crit you have, the more you want crit.

---

Perhaps in terms of #BLs, but that's a bit of a sweeping statement ^_^.

Crit still has decay per point, although it decays much more slowly than for other stats / classes.


Sidenote: that result for +BL/crit is still weird.

Ok...So now that we have this information.. I would like us to know how much is this actually worth it.. Because I'm sure we all know that, in order to get Critical Rate, other statistics must be sacrificed. If determination and dexterity are sacrificed then we may be using more bloodletters but every single ability will deal a little less damage.

What we need to find is that harmonious balance of sacrificing other stats for critical, but making sure that the end result is higher damage, because afterall that is what we want; to make the job for the tank harder ;) :p

Edit: I suppose the higher crit on other abilities other than dots would offset this sacrifice and most likely will produce higher damage; just wanted to throw it out there for consideration.

what i want to know ( I am sorry if it is a stupid question), is does IR stay with DOT's or come off when IR goes down?

Quote:

---

Originally Posted by Vea

what i want to know ( I am sorry if it is a stupid question), is does IR stay with DOT's or come off when IR goes down?

---

When you use a DOT ability, it takes a snapshot of your current buffs and adds them to the entire duration of the DOT. So if you use VB or WB with 4 seconds left on Internal Release and 7 on Raging Strikes, the DOT will stay buffed with IR and RS for the entire 18 (?) seconds even though the buffs drop off soon after you've used the attack.

Well, that seems to be the theory anyway.

I am sorry I don't quite understand why BLs are superiors in numbers compared to procs. Is that the number of times there would have been no procs and that BLs would have reset naturally ?

I don't want to start a debate on BL usage but since you seem quite a knowledgeable on the matter : Did you know BL animation can be cancelled and has an effective 0 second cast (I noticed it just recently) which means you can fire any other skill afterwards without delay and on the other hand, casting anything but BL will cause an animation delay on BL. What's your view on this ?

Quote:

---

Originally Posted by Thotor

I am sorry I don't quite understand why BLs are superiors in numbers compared to procs. Is that the number of times there would have been no procs and that BLs would have reset naturally ?

I don't want to start a debate on BL usage but since you seem quite a knowledgeable on the matter : Did you know BL animation can be cancelled and has an effective 0 second cast (I noticed it just recently) which means you can fire any other skill afterwards without delay and on the other hand, casting anything but BL will cause an animation delay on BL. What's your view on this ?

---

Yes, the difference in numbers is the amount of full duration cooldowns (12 seconds w/ no procs) you incur.

Since you will NEVER miss a BL by waiting until offGCD to use BL, it is a moot point as you should never delay a GCD skill to use BL.

Quote:

---

Originally Posted by Borvos

Ok...So now that we have this information.. I would like us to know how much is this actually worth it.. Because I'm sure we all know that, in order to get Critical Rate, other statistics must be sacrificed. If determination and dexterity are sacrificed then we may be using more bloodletters but every single ability will deal a little less damage.

---

Crit is overweighted by random Bards who try to stack it to the exclusion of all else. It's actually quite comical.


The current stat weights I've estimated for a Bard rotation (autocrit SS + WB/VB + 5x HS [note: 5x HS -> 1x 100% autocrit SS is another point where a simulation would have differing results], including fire AOE once per minute, BL procs, autoattacks, IR and Barrage, etc etc; does not include Hawk's Eye which is a minor skew against Dex) are roughly:

WD 7.849
DEX 1.000
DET 0.235
CRT 0.233
SS 0.136

However, my model estimates about 8% fewer BL procs than the simulation results panda posted. After hard-coding an 8% increase in BL procs:

WD 7.849
DEX 1.000
DET 0.234
CRT 0.232
SS 0.136

Almost identical. Interesting that the CRT (and DET) values went down at all. Probably because the base character crit (341) + SS + IR are so high. The 8% adjustment magnifies those as well, slightly marginalizing crit from gear.

The only other factor is the large quantity of "hard wait" BLs in the simulation. Adding that (0.0366 hardwait BLs per second) results in:

WD 7.849
DEX 1.000
DET 0.232
CRT 0.224
SS 0.131

Unsurprisingly, adding in BLs that don't appear from crit procs reduces the relative value of crit.

Note that at higher crit levels, the number of hard wait BLs will be reduced, so the actual value of crit is very slightly lower than the number above. I'd WAG 0.223. Note: I consider the 15% crit baseline (483 crit rating) sliiightly high, so I would suggest a 0.225 weight or so. Regardless, the differences are small.

It certainly wouldn't be above dex, but I'm surprised its not above det.

Determination is a flat increase in damage independent of weapon damage and dex, but I suppose since bard's base weapon damage is so low that it is weighted higher.

For casters at least weapon damage and int are so high and lack of autoattack that crit's multiplier of the whole thing outshines det even without a crit proc.

Out of curiosity, does your sim accurately reflect the fact that flaming arrow is auto-crit damage?

Quote:

---

Originally Posted by pandabearcat

It certainly wouldn't be above dex, but I'm surprised its not above det.

---

Det appears to cost more item budget than crit/ss/acc, so it "needs" have a higher weight per point or it'll be pretty bad.

Quote:

---

Originally Posted by Viridiana

Out of curiosity, does your sim accurately reflect the fact that flaming arrow is auto-crit damage?

---

Actually no, my model has it as a normal ability that crits when you crit, etc. I read something about shadow flare autocritting; the fire AOE for Bards is the same? May be 'common' among persistent AOEs potentially sourced from dummy objects or something else strange. Doesn't seem like there's a really "good" reason why those would autocrit.

In any case, that would alter the weights to:

WD 7.849
DEX 1.000
DET 0.231
CRT 0.221
SS 0.130

Relatively small-ish dent to crit, although it's obviously dependent on the target standing in the AOE.

So, it looks like .22 is a good estimate, at least going off of the stats you're using to run your sim. I don't suppose this is available online anywhere, is it? I'm curious what it would say the weights are with my stats, bearing in mind that I'm just under the crit rating you're using.

Recommendation for a good file upload site? It's in Excel.

Edit:

Nevermind; I guess google can help answer that. http://www.filedropper.com/ff14

Edit:

Oh, any references to manual corrections for DOT clipping are obsolete; clipped DOTs are calculated automatically (but the game system invalidating a DOT overwrite is not handled, so don't clip or if you do, prepare to adjust your rotation manually to compensate).

Okay, playing around with numbers, I'm getting pretty similar results. Only thing I can't figure out is what's going on with Skill Speed, as changing the values for that seems to do nothing. >_>

Are you referring to the skill speeds in cells Z30:Z35 on the Bard sheet? Those don't do anything except keep track.

Skill speed is evaluated as part of the actual potency-per-second estimate of the rotation. Adjust skill speed in cell U1 at the top. That will change U2 (dt = delta time = the GCD) that the table on the left uses to model the rotation. You want to then copy the resulting potency per second from U9 into the appropriate cell within Z30:Z35.

The stat weights area is more or less a completely separate thing from the rotation model. The inputs on the stat weights area are the rotation PPS column (Z30:Z35), and the other stats (not SS) in columns U:Y. Column Z is pretty important to update.

Okay, once I opened it up in Excel (Open Office wasn't reading the rotation section properly) and did what you suggested, I wound up with these numbers:

WD: 8.015 (not a +1 yet, so yeah)
DEX: 1.000 (Baseline)
Det: .235
Crit: .224
Speed: .105

Now to go plan some gear upgrades, as if I have much choice in the matter. . .

Skill speed is butts, is the conclusion I've come to for basically everything.

I guess spell speed is okay for BLM though.