The Unbearable Triteness of Preening

Monday, August 29, 2011

Some black magic - test server findings

There are at least 20 things I could post about more interesting than my checking parameters for black magic spells (BG forum), but those other things would take too long.

That Breakga appears to last only 30 seconds unresisted is notable since the recast time is 50% longer (45 seconds). Compare this to single-target Break (30 second duration, 30 second recast). Even Horde Lullaby II lasts 60 seconds (with only a 24-second recast).

Comet's direct damage/MP efficiency doesn't compare favorably to Blizzaja and Thunder IV without the usual ice or lightning potency merits.

I also checked the magnitude of Kaustra's direct damage, but surprisingly the direct damage appears to be variable in a way that can't be explained by resists. With 85 INT on SCH95/MNK1 I managed to get 378 damage 3 times and 340 twice.

Sunday, January 2, 2011

"Naked" Aspir and Aspir II at 400 dark magic skill

At 400 dark magic skill and with no Aspir-enhancing equipment/traits, unresisted Aspir I appears to have a maximum of 160 MP (actually observed maximum: 160 MP) and minimum of 80 MP. (The two lowest observed values, 74 MP and 76 MP, may reflect a resisted state.)

At 400 dark magic skill and with no Aspir-enhancing equipment/traits, unresisted Aspir II appears to have a maximum of 240 MP (actually observed maximum: 238 MP) and minimum of 120 MP (actually observed minimum: 121 MP).

Consistent with previous results, the variability of MP drained with Aspir II is higher than that with Aspir I, which appears to be a result of the range of possible values increasing as the maximum increases. (The maximum of Aspir II, for a given level of dark magic skill, is higher than that of Aspir I.)

Dot plot

Raw data (first column observed MP drained, second Aspir I or Aspir II, third order of data collection)

Monday, September 13, 2010

Drain, Aspir, and Aspir II after the update.

Previous posts have shown that increasing dark magic skill beyond 300 did not increase Drain potency prior to the September 8, 2010 version update, but the version update details state that "maximum values have been increased for certain enhancing magic, dark magic, and blue magic spells whose potency is commensurate with casting skill." The following summarizes the results of checking the potency of Drain, Aspir, and Aspir II without any equipment or other factors that "enhance" any of these spells other than equipment with dark magic skill bonuses.

Raw data is available upon request or I may just post them in the comments below after a week if there is no feedback.

Drain

The distribution of Drain potency (unresisted Drain values) was checked at both 303 dark magic skill (my current dark magic skill level without any equipment) and 340 dark magic skill (with this equipment) on Zvahl Fortalices in Castle Zvahl Baileys (S). The following dot plots show an obvious increase in the maximum value of Drain. Although I was unable to check the distribution of Drain at 300 dark magic skill, it seems that the change in the maximum Drain was done merely by removing the "cap" on increasing it with dark magic skill beyond the 300 level, holding all other potency-increasing factors (maximum-increasing factors) fixed. (In this case, what was fixed was that there were no other factors "present" to affect my results.)

Previously, I gave a formula for the maximum Drain as a function of dark magic skill (without any other potency-affecting factors) as (skill) + 20. The following summary of the minimum and maximum suggests that this formula does not apply beyond 300 skill.

                        ·--------·---------·--------·------------·
                        |    n   |   Min   |   Max  |   Median   |
·-----------------------·--------·---------·--------·------------·
|   Drain (303 skill)   |   60   |   170   |   321  |    247.5   |
|   Drain (340)         |   60   |   173   |   345  |    248.5   |
·-----------------------·--------·---------·--------·------------·

For now, the maximum observed Drain without any other maximum-increasing factors present (or, for shorthand, "naked") is 345.

I do not have access to Drain II so I cannot check it at this time.

Aspir and Aspir II

I actually checked the distributions of Aspir and Aspir II (not omitting resisted values) before that for Drain, and I collected data at both 300 and 302 dark magic skill (on two separate occasions), thinking I had done so only at 300 without realizing later I had skilled up beyond 300. I was lazy and pooled the data, but it is straightforward to see that, similarly to the Drain maximum, the increase in the Aspir (and Aspir II) maximum was done merely by removing the "cap" on increasing it with dark magic skill beyond the 300 level. I checked at 339 dark magic skill with this equipment set. All data collection was done on Stone Eaters in North Gustaberg (S).

I had no idea what the distribution of Aspir II was like compared to Aspir before the update. The previous image shows that, at least after the September 2010 update, the variability of Aspir II (controlling for resist level) is higher than that of Aspir (no reason to assume anything about this beforehand) even though the maximum Aspir II is higher (as expected) as shown in the above image and following summary statistics table:

                            ·--------·---------·--------·------------·
                            |    n   |   Min   |   Max  |   Median   |
·---------------------------·--------·---------·--------·------------·
|   Aspir (300/302 skill)   |   50   |    60   |   120  |     85.0   |
|   Aspir (339)             |   50   |    50   |   135  |     95.5   |
|   Aspir II (300/302)      |   50   |    83   |   180  |    130.5   |
|   Aspir II (339)          |   50   |    56   |   202  |    152.0   |
·---------------------------·--------·---------·--------·------------·

Previously, I gave a formula for the maximum Aspir as a function of dark magic skill (without any other potency-affecting factors) as (skill/3) + 20 (decimals truncated where necessary). The above data shows that this formula does not really apply above 300 dark magic skill. I make no attempt to suggest a formula for the maximum of Aspir II at this time.

The maximum "naked" Aspir observed is 135, and the maximum observed Aspir II is 202.

Summary

Announced in the September 2010 version update, the increase in the maximum Drain, Aspir, and Aspir II values seems to be the result of allowing their potency to increase by increasing dark magic skill beyond 300, holding all other maximum-increasing factors fixed. Before the update, the Drain and Aspir maximum did not increase beyond 300 dark magic skill.

The variability of Aspir II values (controlling for resist level) is higher than that for Aspir values.

I observed a maximum "naked" Drain of 345 (340 dark magic skill), a maximum "naked" Aspir of 135 (339 skill) and maximum "naked" Aspir II of 202 (339 skill). The actual maxima are likely higher and are likely to be attained only by increasing dark magic skill further (well above 340).

Friday, August 27, 2010

Blitzer's Roll

Blitzer's Roll (COR Lv.83)
Reduces melee attack delay for party members within area of effect. Lucky number: 4. Unlucky number: 9.

Seeing as this is the only interesting thing from the update as far as TP-burning for one-handed melee DD vs. Hasso (you self-identify as WAR/DRK/DRG/SAM? Are you a tool? It's either piercing Hasso or non-piercing Hasso) is concerned (although that supposes anyone still cares about FFXI), let's anticipate what the effect might actually be.

Thoughts:

If this really reduces melee attack delay a la Sword Strap, maybe it will be exactly the same form of delay reduction as that from dual wield. (Why not?) If Blitzer's Roll actually stacks with dual wield and the bonuses are good, this would benefit dancer and ninja the most.
One-handed melee jobs get more from Blitzer's Roll even if it reduces weapon delay separately from dual wield, but it doesn't mean it'd be bad for two-handed melee jobs (depends on the actual bonuses), just that Fighter's and Samurai would be better. n-hit builds are overrated.
What if it's "job ability haste"? Wouldn't it be funny if the bonuses were large enough to obviate the use of Hasso and Haste Samba? Among other things that would follow: why are we using two-handed jobs for TP-burn then?

Friday, August 13, 2010

Samurai Roll and Fighter's Roll... for ninja

Having shown in the past that Fighter's Roll can be better than Samurai Roll for increasing warrior's damage output (despite the lack of consideration of WS delay, which actually would favor Fighter's Roll), I thought it might be nice to have a worked example for ninja (this time accounting for WS delay).

This image (Imageshack host) summarizes the computations, based on 95% hit rate, 15% DA rate, 55% haste, 40% dual wield, and a "cRatio" of 1.5, with a 10% critical hit rate, with Blade: Jin as the weapon skill. The specific katana combination is Hochomasamune/Uzura, which is something I would consider getting as a "high-value" option, meaning that it's a highly effective option that I don't have to waste as much time doing boring shit to get this as I would for other options.

First, let's start with Samurai Roll (augmented by the presence of a samurai). The following is a table summarizing the relative increase in the (theoretical maximum) rate of damage from each of several desirable Samurai Roll outcomes.

Roll total      Bonus     Damage/second     Relative efficiency
---------------------------------------------------------------
 -              -               113.607                       -
 2             42 STP           123.979                   9.13%
 7             26               119.672                   5.34%
 8             30               121.410                   6.87%
 9             32               121.414                   6.87%
10             34               121.458                   6.91%
11             50               126.684                  11.51%

The relative increase for each store TP bonus is decent, but not as high as you would expect for two-handed weapons since the proportion of total damage from Blade: Jin for katana is realistically never going to be as high as the proportion of total damage from the (generally) best weapon skills for two-handed weapons (e.g. Drakesbane, Raging Rush, Tachi: Gekko).

In contrast, Fighter's Roll (here augmented from the presence of a warrior) increases auto-attack damage, weapon skill frequency, and weapon skill damage, so it's not a surprise that Fighter's Roll is generally easily superior to Samurai Roll for increasing rates of damage:

Roll total      Bonus     Damage/second     Relative efficiency
---------------------------------------------------------------
 -               -              113.607                       -
 5             15% DA           129.567                  14.05%
 7             11%              125.291                  10.28%
 8             12%              126.358                  11.22%
10             13%              127.427                  12.16%
11             19%              133.857                  17.82%

Note that these are supposed to be the actual DA percent bonuses after the infamous August 2007 version update (source), you know, the one that resulted in ninja being disregarded as DD. If they were what they were before that update, assuming there was a change (higher than what they are now), Fighter's Roll would be even better.

So, if Fighter's Roll can be shown to be better for the job that gets the least relative benefit from Fighter's Roll (warrior), as well as for ninja (though this could be considered self-evident), are there even any legitimate reasons to use Samurai Roll? No, "TP overflow" is an incorrect and stupid answer that betrays a lack of conceptual understanding.

Now, that's not to say Samurai Roll has no application. Maybe it would be desirable to maximize the ratio of damage efficiency to TP "fed" to a mob rather than consider damage efficiency in isolation. I doubt Monk's Roll would do the job, but a case could be made for Samurai Roll.

Also, if it is desirable to maximize weapon skill frequency, Samurai Roll would generally be better for that purpose than Fighter's Roll. (Also related is maximizing TP per hit, which would be desirable for dancer.)

Also, Fighter's Roll can be said to be "contraindicated" for use with multi-hit weapons (that don't use virtue stones) such as Magian multi-hit weapons. (But this assumes Magian multi-hit weapons are actually good, which is not necessarily true.)

Saturday, July 31, 2010

Aspir and Drain modeling: an incomplete picture

Earlier, I made some bold statements (at least by my standards) about the distribution of unresisted values of Drain and Aspir. I proposed a model from which I can make explicit, testable predictions.

Specifically, I wanted to see if the model holds at 114 dark magic skill, which can be attained by subbing /DRK on any job without any native dark magic skill (in my case NIN/DRK). Based on the model I described previously, the maximum Aspir without any potency-increasing equipment used is 114/3 + 20 = 58 MP, and the minimum is half that, or 29 MP.

I went out to cast Aspir on Stone Eaters (North Gustaberg (S)) and after the seventh cast I obtained an Aspir of 63 MP, which exceeds the stipulated maximum, so the model doesn't hold for 114 dark magic skill. The following are the observed data values (in order observed):

51 44 56 49 25 51 63 53 35 62 62 46 41 32 44

I then cast Drain on the same Stone Eaters with 246 dark magic skill (NIN75/SCH35 with Dark Arts) and obtained the following results (stem-and-leaf plot):

(The reason I used Drain and not Aspir was that my scholar is only level 35, and Aspir is accessible at level 36. I would prefer to gather data for Aspir because Aspir values are obviously less variable than Drain values.)

Given 246 dark magic skill, the predicted maximum for Drain is 266 and the predicted unresisted minimum is 133 (both without any potency gear or day/weather effects), so the distribution of Drain (as represented by the sample) seems consistent with the model.

In conclusion, if one were to be technical in describing the scope of the model I proposed earlier, I would say the model is (likely) valid for Drain between 246 and 300 dark magic skill. It is valid for Aspir between 269 and 300 dark magic skill. And, finally, it is valid for Drain II between 285 and 300 dark magic skill. What happens in the 100s is just not that relevant. (In case you are wondering what data I am referring to, you'd have to check my old posts on Drain and Aspir).

Saturday, July 24, 2010

The "new" melee pDIF

This post is definitely not for those who neither understand nor care about what melee "pDIF" is all about and why it can be of interest, so I find no point in making some sort of "for dummies" kind of introduction and will just jump into the results.

First, a reference to the "new" melee pDIF should be seen as a sarcastic gesture, as there likely have no been wholesale changes to pDIF after the August 2007 version update that brought the gameplay-altering "two-handed weapon adjustment." Therefore, the following results are assumed to reflect the actual changes to pDIF made in August 2007.

Data and results

The guy who plays Masamunai (currently of Cerberus) provided this spreadsheet of data, having tabulated the observed damage values for various ratios of attack to defense (without level correction), using both one-handed and two-handed weapons, on level 63-65 Lesser Colibri and then "standardizing" them to approximate observed pDIF values (acknowledging estimation error associated with in-game truncation of values). There are more details concerning the raw data and he provided his own analysis, but I prefer to do my own analysis so you don't necessarily have to review the spreadsheet yourself.

The following is an image attempting to plot 67,123 of the observed pDIF data values (almost of all the data) to show primarily how the minimum, maximum, and (most important to me) mean pDIF for both critical and non-critical ("normal") hits varies with the ratio of attack to defense:

It is somewhat difficult to plot 67,123 data values cleanly and elegantly with limited resolution, so I exploited transparency of data points, resulting in narrow "bands" that vary in opacity from top to bottom, an attempt to illustrate roughly the relative "density" of observed values. Each band represents the entirety of the data collected for a given attack/defense ratio. Another interpretation is that each band represents the observed conditional distribution of pDIF for a given attack/defense ratio.

The bands for critical pDIF are generally less "dense" or less opaque than those for normal pDIF, reflecting that fact that there are many more data points for normal pDIF (55,956 versus 11,127). Also, the bands are generally most translucent at the endpoints, reflecting the fact that the observed data at the extremes of each conditional pDIF distribution (for a given attack/defense ratio) occur relatively less frequently, which is consistent with the idea that pDIF is now a function of two uniform random variables (either the sum or the product), which follows a trapezoidal(-like) distribution. (But I will not be discussing probability distributions today.)

Aside from the plotting of the data values, regression lines for the mean pDIF (controlling for attack/defense ratio) were also plotted (lines based on ordinary least squares, which is justifiable as there are a lot of data points involved for each level of attack/defense considered). Regression was done in an informal piecewise fashion, as there are specific ranges of attack/defense ratio where the variance of pDIF is obviously not constant, specifically for three cases:

where there is a critical pDIF upper limit imposed (3.15 when attack/defense is approximately greater than 1.65)
where there is a normal pDIF lower limit imposed (1.00 when attack/defense is between 1.25 and 1.5), and
where the mode of normal pDIF is 1.00 and the mode does not occur at the left endpoint of the pDIF distribution (when attack/defense is less than 1.25). It should be noticed that it is impossible to discern the mode of pDIF (conditional on a given attack/defense ratio) based on the above graph. One would have to consult the original source as cited above.

I hope that will suffice as an explanation for the elements of the graph.

Interpretations and conclusions

These are a few of the things one could take away from the graph above.

Aside from the maximum attack/defense ratio attainable, there appear to be no differences in pDIF between one-handed weapons and two-handed weapons. I have incorrectly thought otherwise in the past, but I assumed people who cared about this knew what they were talking about. Obviously not.

While there is no data for two-handed weapons below 1.398 attack/defense ratio, I would invoke model parsimony and assert there is no good reason to expect differences at lower values of attack/defense. Although it is not shown above (and cannot be shown above cleanly), 2.00 is the maximum attack/defense ratio for one-handed weapons, and 2.25 is the maximum attack/defense ratio for two-handed weapons. Support for the these maxima can be found in the spreadsheet.

The ceiling on critical hit pDIF first occurs near 1.65 attack/defense. Moreover, the value of the ceiling, 3.15, is the modal (most frequently occurring) pDIF for attack/defense ratios above 1.65.

Mean pDIF, as a function of attack/defense, does NOT increase at the same rate for critical hits as for normal hits for a given value of attack/defense. A consequence of this is there is no pat way to relate normal pDIF to critical pDIF, like critical pDIF = normal pDIF + 1. To see what I mean, refer to this blog entry (JP), particularly the first image, to get a sense of how pDIF was incorrectly perceived more than a year after the August 2007 version update (a mish-mash of the critical hit pDIF ceiling of 3.15, increased attack/defense ratio maximum, and the old pDIF model).

Irrelevant considerations

This is a matter of personal preference, but I consider the so-called "secondary randomizer" an irrelevant red herring. pDIF as a product of two uniform random variables or sum of two uniform random variables, so what? (But I will note that the slope of mean pDIF without the second random factor does not change if the factor is added and does change if the factor is multiplied.) I just know it's there and I can explain what can cause it, but it is not very important for estimating mean pDIF, which is why I even made this post in the first place.

I also do not care about exactness of any pDIF model. Approximately true is fine with me as far as modeling rates of damage is concerned. (There are other factors when completely ignored or incorrectly computed that cause much more error than mere sampling error based on 60,000+ samples).

Formulas for mean pDIF as a function of attack/defense ratio and whether the weapon is one-handed or two-handed

These formulas are based on the regression estimates. (You may have noticed discontinuities in the piecewise mean pDIF functions suggested in the graph, but I do not care that much about fudging the estimates to eliminate that.) For normal-hit pDIF, which I have denoted as M_Normal, the estimated mean of M_Normal, as a function of H, the number of hands required to wield a weapon (H = 1, 2), and R, the attack/defense ratio (level-corrected or otherwise), is

For critical-hit pDIF, the functional relationship between that and H and R is

The following is the output of the regression procedure. I only include this to show that there is no reason to expect that the coefficient of determination be high, mainly because there is inherent variability of pDIF.

***Regression for normal pDIF, ATK/DEF < 1.25****

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept) 0.225403   0.010107    22.3   <2e-16 ***
ratio       0.782699   0.009748    80.3   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.1621 on 12257 degrees of freedom
Multiple R-squared: 0.3447,     Adjusted R-squared: 0.3446 
F-statistic:  6447 on 1 and 12257 DF,  p-value: < 2.2e-16 



***Regression for normal pDIF, 1.25 < ATK/DEF < 1.5****

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  0.07274    0.04129   1.762   0.0781 .  
ratio        0.90232    0.02969  30.390   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.2369 on 14254 degrees of freedom
Multiple R-squared: 0.06085,    Adjusted R-squared: 0.06078 
F-statistic: 923.6 on 1 and 14254 DF,  p-value: < 2.2e-16 



***Regression for normal pDIF, ATK/DEF > 1.5****

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) -0.294339   0.016866  -17.45   <2e-16 ***
ratio        1.162306   0.009566  121.50   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.2518 on 29479 degrees of freedom
Multiple R-squared: 0.3337,     Adjusted R-squared: 0.3336 
F-statistic: 1.476e+04 on 1 and 29479 DF,  p-value: < 2.2e-16



***Regression for critical pDIF, ATK/DEF < 1.65****

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  0.94139    0.01739   54.14   <2e-16 ***
ratio2       1.07335    0.01273   84.29   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.232 on 6745 degrees of freedom
Multiple R-squared: 0.513,      Adjusted R-squared: 0.5129 
F-statistic:  7106 on 1 and 6745 DF,  p-value: < 2.2e-16 



***Regression for critical pDIF, ATK/DEF > 1.65****

Coefficients:
            Estimate Std. Error t value Pr(>|t|)    
(Intercept)  1.59673    0.04287   37.25   <2e-16 ***
highratio    0.68607    0.02344   29.27   <2e-16 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 

Residual standard error: 0.1905 on 4377 degrees of freedom
Multiple R-squared: 0.1637,     Adjusted R-squared: 0.1635 
F-statistic: 856.8 on 1 and 4377 DF,  p-value: < 2.2e-16