Crit PDIF Distribution and other interesting data

[Kirschy]

I realize this is a weird topic for a thread, but I thought some of the information was useful/interesting. When I did my Liberator Trials I purposely used 3 crit atmas, and compiled the data. Below are some interesting facts I found over the large data sets.

The following is a distribution graph for the critical hits during one trial.
Notes:
-3057 total crits
-Lv75 Liberator was used: (DMG 93 + 10 = 103)
-AF3+2 procs were removed

http://img827.imageshack.us/img827/4804/crits.png

Most crits are in the 3.0-3.15 pdif range as expected. (103 DMG + 19 fSTR) * 3.0 pdif * 1.35 Crit DMG Boost * 1.05 Randomizer = 518 which was the max damage I saw. 59 times to be exact! (~2% chance.) The lowest recorded hit was 413 damage, which only occurred once, this estimates around 2.5 pdif with the crit.

For the above set, I used 3 pieces of AF3+2. In the 3121 critical hits, 64 of them were AF3+2 set bonus procs. 64/3121 = ~2% for 3 pieces. Around ~.68%/piece.

In another trial, I used 5 pieces of AF3+2. In 3224 crits, 155 of them were AF3+2 set bonus procs. 155/3224 = ~4.8% for 5 pieces.

For my last trial, I mostly fought one mob at a time which gave me some great data on Gnarled Horn's counter rate:

185 Counters and 1784 hits taken. This comes out to be: ~9.4% Counter Rate. Factoring in the 95% acc, comes out to be ~9.9% counter rate. Seems like solid evidence to put Gnarled Horn @ 10% counter rate.

Since my final trial was on Arcana, I had the unique situation to parse my Arcana Killer. KParser has it listed as 10.46% intimidation rate.

EDIT: Forgot to add one more statistic. For the last trial, I killed 421 mobs, and saw 235 chests, making the chest drop rate /w 3 Abyssites around 55%. (Zone was Misar, I did end up with like 4 pearl lights according to parser, but I also re-entered 3 times throughout the parsed trial.)

EDIT 2: Forgot to add another chest statistic. For another trial I killed 330 mobs, and got 133 chests. Making the drop rate ~40%. Zone was La-Theine, and as with Misar I got a few pearl lights, but also re-entered a few times. Seems possible the martello bonus might affect chest drop rate, similar to dominion status in hero zones.

[Nightfyre]

Not really surprising. fMax(cRatio) values can and do exceed 2.0, meaning that all values >=2.0 will become critpDIFa=3.0 after the cap is applied. I'd imagine the distribution would be more in line with expectations once level correction becomes a factor, pushing the maximum value below 2.0.

[Kirschy]

Not really surprising. fMax(cRatio) values can and do exceed 2.0, meaning that all values >=2.0 will become critpDIFa=3.0 after the cap is applied. I'd imagine the distribution would be more in line with expectations once level correction becomes a factor, pushing the maximum value below 2.0.

Ah right, I should have added, those were done on Even Match birds, so there was no level correction involved!

[Lyramion]

So if Gnarled Horn is a 10% counter rate with MINOR, what is Roaring Laugher then? 15? 20? Can a Sphrai MNK reach 95% Counter Rate or is there a Lower Cap on the actual Counter Proc Rate (not meaning the ACC bound cap)?

[Appie]

Roaring Laughter so far looked like a typo from what I heard and that it's actually "minor". As for that counter cap, I am going to test that after I pinpoint Spharai lv90 counter some more. Still need to get Roaring Laughter too.

[Kirschy]

Appie had a decent parse with just Lv90 H2H + GH. Showed ~78% counter rate not counting the acc thing. Would prob have to bug him to run a test with Counter+ atma and counter+ eq to see if there's a cap!

EDIT: Beaten by Appie! Interested in seeing the Counter Cap test!

[Lyramion]

Le Awesome, can´t wait.

[Kitrane]

Friend on Sylph with Spharai told me a while back that she didn't think the additional tiers granted through magian increased the counter rate. She had the 85 version at the time I believe. Has that been officially disproven now? If so I'm really curious to see the testing as well.


EDIT: Above referenced friend corrected me in tells earlier today. Apparently there is no difference in the counter boost between the 80 and 85 versions. (programming wrong? idk)

[mdkuser]

Not really surprising. fMax(cRatio) values can and do exceed 2.0, meaning that all values >=2.0 will become critpDIFa=3.0 after the cap is applied. I'd imagine the distribution would be more in line with expectations once level correction becomes a factor, pushing the maximum value below 2.0.

This doesn't explain why most of the crits have above 3.0 pdif... The min is 2.5 (computed by masamune with 3 digits) and the max would be 3.5 so only 50% of the hits should hit the jackpot( in case of uniform distro). All it shows is that pdif is not uniformly distributed between min and max. Most likely, the distribution is affected by attack ( through ratio value). Anyone that ever fought low level mobs knows that crits are almost always near cap.

[Kirschy]

This doesn't explain why most of the crits have above 3.0 pdif... The min is 2.5 (computed by masamune with 3 digits) and the max would be 3.5 so only 50% of the hits should hit the jackpot( in case of uniform distro). All it shows is that pdif is not uniformly distributed between min and max. Most likely, the distribution is affected by attack ( through ratio value). Anyone that ever fought low level mobs knows that crits are almost always near cap.

For what it's worth, the distribution in the graph showed 1447 values below 3.0 PDIF, and 1610 values at 3.0 or higher PDIF. So more than half the values (~52.6%). If I get around to it, I'll try to get a distribution of 2H non-crit in the same situation to match it up.

[mdkuser]

Err yes the data does seem to be symetric with respect to 3.0 pdif but not uniform. It seems that without the 3.15 cap you would have a large maximum at aroud 3.0 which could be verified on non crits. Anyway what happens is that the maximum of the distribution is greater that 3.0 before the cap is applied. My guess is that with a tiny lower pdif not that many hits would exceed 3.0.

[Byrthnoth]

So, with Apoc's major benefit negated by bountiful Haste options in gear and its other major benefit (being self-healing) being negated by the massive damage taken and mages having infinite MP, how do you think Liberator fares against it these days?

Also, random question, but does Liberator's AM3 overwrite itself? I keep getting into situations with Twashtar where I don't think it does.

[Kirschy]

So, with Apoc's major benefit negated by bountiful Haste options in gear and its other major benefit (being self-healing) being negated by the massive damage taken and mages having infinite MP, how do you think Liberator fares against it these days?

Also, random question, but does Liberator's AM3 overwrite itself? I keep getting into situations with Twashtar where I don't think it does.

It didn't used to. I haven't re-checked though tbh. The AM lasts 3min now which is a huge increase. I used to save around ~30 seconds to re-gain another 200 tp, so I had 90 seconds to spam WS. Now that time has increased to 150 seconds. Also with the extra sTP/haste options available, I can 5-hit with capped haste while keeping in a ton of attack. (I'm sure with a 5-hit build, getting to 300 TP will be a lot easier too.) Also, Liberator seems to be getting the largest increase from Magian Trials. It started off as 93 DMG at lv75, and ends up at 133 by lv 90. And last, Liberator has high WS damage output due to the AM affecting WS DMG and weapon skill frequency, the Lv90 version gives us a 15% boost in that area which I think is major improvement. Haven't actualli used it for anything, but I'll be sure to post my experiences =p.

[Raelia]

* 1.05 Randomizer

You're always wrong until everyone forgets that you were right.


'Synthetic' atma seem to all use different scalings than Dropped atma.

[Kirschy]

You're always wrong until everyone forgets that you were right.


'Synthetic' atma seem to all use different scalings than Dropped atma.

What do you mean? The 1.00 - 1.05 random term on crits has been in the game since before abyssea existed. And what scaling difference are you talking about between synthetic and dropped atma?

[Cream Soda]

I've nothing useful to add, just wanted to thank you for a good job, like always.

[Byrthnoth]

So, useful for those of us who like to do napkin math. What did your average crit pDIF parse as?

My guess is 2.94, or 484~5 damage!

[Kirschy]

So, useful for those of us who like to do napkin math. What did your average crit pDIF parse as?

My guess is 2.94, or 484~5 damage!

I'll get that info for ya when I get home.

[Masamune]

Nice data Kirschy !

Can you post your data in the "standard format" like used by Kparser please ? (ie 1st column with each individual observed damage value, and 2nd column next to 1st one with corresponding parsed frequencies). Up to you to remove "false crits"

Also i'm wondering if your tests werenot screwed by level of mobs fought: were they ALL really even match ?

Lastly, just to try to explain why seeing way more frequency values near cap:
imagine your data had no cap and has "approximately" a uniform distro shape. Now, "cut" the highest values to 0% frequency, but add all those cut frequencies to the last uncut frequency, thus making visually a frequency spike.
And finally, add a secondary randomizer like raellia explained: the aforementioned spike will become "sparsed" between 3.0 pDIF (max uncut value) and 3,15 pDIF(max *5% 2ndary randomizer). The result is what you showed in the graph of your 1st post.
Hoping to be clear^^

[Thorny]

Minor thing I didn't see pointed out(not bashing, great info, just thought it might be worth mentioning):

64/3121 = ~2% for 3 pieces. Around ~.68%/piece.

You don't get set bonus with one piece do you? I'd think it's 1%-2%-3%-5% for 2-3-4-5 pieces to have a similar pattern to other af3 sets.