[X] My Assistant

Follower of Decondelite Perk - costs 50k hath to be paid to DJNoni. The effects are great! It really, really helps*.
*really helps DJNoni, that is (IMG:[invalid] style_emoticons/default/ph34r.gif)

Hmmm.... what do I get?

Both Decon and Uncle Stu have been saying this for a pretty long time

I've been saying that for even longer. That's why basically everyone knows it's basic knowledge...(IMG:[invalid] style_emoticons/default/heh.gif)

- snip -

Thank you very much for the explanations. (IMG:[invalid] style_emoticons/default/wink.gif) Despite the simplifications made, it was possible to understand why the negative effect of using Haste and Shadow Veil for the 1H style. Considering all the parameters involved, the mechanism is quite complex. (IMG:[invalid] style_emoticons/default/wacko.gif)

As Juggernaut Santa just said, this is all basic knowledge. (IMG:[invalid] style_emoticons/default/heh.gif) But I decided to do these simple tests just to see, in numbers, all these concepts that have been said over the years. A lot was just said (in text), and justified in most cases only on the basis of theory (equations and considerations). And since I'm not very good with these equations (in fact, I don't even know what to do with them (IMG:[invalid] style_emoticons/default/blush.gif) ), I preferred to see them in practice. I think that the presentation of real numbers always helps to understand the complex mechanisms of the game (especially for those who do not understand this just with theory, or because they do not know it).

Follower of Decondelite Perk - costs 50k hath to be paid to DJNoni. The effects are great! It really, really helps*.
*really helps DJNoni, that is (IMG:[invalid] style_emoticons/default/ph34r.gif)

Not even for free. (IMG:[invalid] style_emoticons/default/tongue.gif) (IMG:[invalid] style_emoticons/default/laugh.gif)

This post has been edited by Basara Nekki: Jun 14 2020, 18:57

I'd already thought of lestion's explanation for why shadow veil and haste might be extra bad together, but I'm not sure that's the correct reason, so I didn't mention it. It helps, but there might be something else more important that I haven't figured out.

There's some other interesting issues with haste and counter attacks that I noticed while testing out my action speed theories (the log file). Once any enemy recovers from stun, it no longer matters whether you have haste or not in terms of the number of counter attacks you perform (except for increasing the stun duration from 4 to 6~7 turns, which is the biggest effect on loss of counters anyway).

If I interpret what I experienced correctly, it seems that ~95% of all enemies we encounter are chaosed enough to have over 25% attack speed bonus (more than half chaosed, so 10+ levels to attack speed). Combined with the 25% attack speed bonus of PFUDOR difficulty this means that whether we have haste or not, most monsters will attack 1H players twice every time they recover from Stun.

And if we approximate that 1H style deals 3 counter attacks and 2 stuns every turn, then this creates a perpetual loop. Because in many turns, 2 monsters will recover from stun, and those 2 monsters will attack 4 times, for an average of 3 counter attacks (75% chance) and once again 2 stuns (~70% chance of stun with a counter attack).

Note the wiki says "monsters receive a free action when a stun expires on them" which is as I just explained. But I think that's only true for 1H style. I haven't used DW club in a long time but I think it can restun a monster with your main hit without the monster making any attack at all.

Both Decon and Uncle Stu have been saying this for a pretty long time - that you should be wearing a mix, at the very least. Force Shields in particular benefit more from warding than they do from protection.

I've also said and done the same thing since I started playing, but End of All Hope might have said it before me, I don't know. It's not just force shields but all power and plate armor that have a bigger boost with the warding suffix than protection. I think it's the same difference in boost if you look at it additively.

This post has been edited by BlueWaterSplash: Jun 15 2020, 01:04

This analysis is also related to the misconception that Protection equipment is better than Warding. I believe Warding is much better than Protection for 1H style. Some people before analyzed their script logs and saw that Protection reduces way more total damage. It's true, but so what? That's the wrong metric for measuring damage.

No it's not unless proven.
I also don't see why are you linking evade with spirit attacks, it's not like only spirit attacks are evaded.
It's even more complex if you factor in spirit shield (how many spirit attacks are mitigated by spirit shield? Is better to use spirit point or take the damage even if you may need to cure? Would warding suffix prevent more hp loss or sp loss in case of spirit shield?).

The correct way to measure damage is the amount of Cures saved, and I believe Warding would save way more Cures than Protection equipment. No one has tested it, but we can see in this Shadow Veil tested example that the damage taken and Cures saved gave entirely different results.

I instead believe you should test it instead.
Also it's obvious that damage taken and cures are not linear, you are steadily regaining health, when the damage taken is less or equal the health you're naturally regenerating the cures are nearly 0.
If you're curing a moderate amount of time a decent defensive spell will bring the number of cures down very fast.
I don't think that this proves anything.

Btw I personally believe that for 1H you should not care about mitigations, which was pretty much what most people said when there were the debate of protection vs warding (and that the only data we had aka damage taken shown that protection would prevent more damage).
I'm honestly surprised that it's still discussed, still unproven and still wildly speculated.

Returning to the 1H analysis of swift strike and low amounts of attack speed, the prior calculations apply to non-imperil rapier style. If any spells are cast the cluster of turns with enhanced ailment durations will move up. Since cast time varies a little with proficiency I will save the full discussion of what spells to use with specific amounts of attack speed for later.

With 1% attack speed, casting spells usually results in the entire loss of enhanced ailments because the first turn is removed while the ailment cluster fails to move up enough to happen. Originally I thought this was disastrous, but since realizing that undesirable stuns are also removed, this is actually a decent outcome. The result is simply 0%~0.04% drop in offense, which is basically nothing.

The next possibility is that the full ailment cluster occurs in the early turns. I will spend this post redoing the prior calculations for this scenario. We begin with the case of arenas without haste. Average PA duration increases to 8. Scaling to the 10.66 duration of haste gives 1 * 1.66% / 3.66 = 0.454545% boost to offense.

There will be 5 turns of enhanced Overwhelming Strikes, which I boost to 7.5 since none will kill an enemy if spread attacking. The average duration is 4.5 turns if rounds end in 15 attacks. Scaling to the 6.33 duration of haste gives 0.5 * 3.33% / 2.33 = 0.714286% estimated boost to offense.

There will be 4 turns of enhanced Stuns which isn't that much worse than before because we are modeling 5 enemy mobs. This means that only 5 long Stuns will be inflicted. Assuming 13 stuns overall the average duration is 3.384615, scaling to the 4.66 duration of haste gives 0.384615 * 6% / 1.66 = 1.384615% which I'll reduce by 1/5 to 1.107692% penalty after considering the side benefits of stun.

In arenas, haste displayed a turns/second boost of 3% which was taken to be 1.5% spread wasting and 1.5% casts. In the current case we have (1.384615% + 0.04%) * 1.5% / 8% = 0.267115% time bonus from spread wasting, and the same 0.2% as before from casts.

The overall impact on offense is thus 0.454545% + 0.714286% - 1.107692% - 0.04% = 0.021139% so effectively no change in turns. This leaves 0.2% + 0.267115% + 0.021139% = 0.488254% benefit to time, which is a bit more than butcher. This is the result of glitch abuse that I originally alluded to, but I guess it's still not broken in the end.

In grindfest, PA duration is 8 for each of eight turns, then 7 for the remaining eleven for an average of 7.42105, which corresponds to 0.42105 * 1.66% / 3.66 = 0.191388% boost to offense. Added turns of Overwhelming Strikes are taken to be 7.5 over the course of the 27 turn round, so 4.28 average, which becomes 0.28 * 3.33% / 2.33 = 0.396825% estimated boost to offense.

There will be 4 turns of enhanced Stuns which is worse in this case because we have a full mob of 9. I will assume 7 enhanced Stuns will be inflicted. Compared to 31.5 stuns overall, the average stun duration is 3.222222 which gives 0.222222 * 8.25% / 1.66 = 1.1% which I'll reduce by 1/9 to 0.977778% penalty due to side benefits of stun.

For grindfest, haste gets a turns/second boost of 6% which was taken to be 4% spread wasting and 2% casts. This yields (1.1% + 0.055%) * 4% / 11% = 0.42% time bonus from spread wasting, and the same 0.148148% as before from casts.

The overall impact on offense is 0.191388% + 0.396825% - 0.977778% - 0.055% = -0.444565% penalty in turns and 0.148148% + 0.42% - 0.444565% = 0.123583% benefit to time. Though contributing components are wild the final result is mundane. I surmise that grindfest's large mobs created a balance with the good and bad aspects of enhanced ailment durations.

With haste in arenas, we get 3.33 turns of extra PA duration over 5 turns, with the remaining turns not mattering, for 3.33 / 5 = 0.666 average added PA duration. This corresponds to 0.666 * 1.66% / 3.66 = 0.30303% boost to offense. There are effectively 2.33 turns of enhanced Overwhelming Strikes, boosted to 3.5, and the average duration is 4.23 turns. Scaling to haste gives 0.23 * 3.33% / 2.33 = 0.3333% estimated boost to offense.

There are effectively 3.66 turns of extra Stun which I'll suppose affects 3.66 * 5 / 4 = 4.583 enemies. The average stun duration increases by 4.583 / 9.5 = 0.4825 turns, for a 0.4825 * 6% / 1.66 = 1.736842% loss which I'll reduce by 1/5 to a 1.389474% penalty due to side benefits of stun. I reduce these by a 1.5 factor to 1.157895% and 0.926316% respectively.

Turns/second boost is (1.157895% + 0.04%) * 1.5% / 8% = 0.224605% from spread wasting and 0.066% from casts.

The overall impact on offense is 0.30303% + 0.333333% - 0.926316% - 0.04% = -0.329953% penalty in turns and 0.066% + 0.224605% - 0.329953% = -0.039348 penalty to time. Small amounts of attack speed generally perform worse with haste.

With haste in grindfest, average added PA duration is 7.66 / 17 = 0.4506 which corresponds to 0.4506 * 1.66% / 3.66 = 0.204813% boost to offense. Overwhelming Strikes gives 0.12963 * 3.33% / 2.33 = 0.185185% boost to offense.

There are 3.66 turns of extra Stun which I'll suppose affects 3.66 * 7 / 4 = 6.417 enemies. The average stun duration increases by 6.417 / 22.5 = 0.2852 turns, for a 0.2852 * 8.25% / 1.66 = 1.411666% loss which I'll reduce by 1/9 to a 1.254815% penalty due to side benefits of stun. I reduce these by a 1.5 factor to 0.941111% and 0.836543% respectively.

Turns/second boost is (0.941111% + 0.055%) * 4% / 11% = 0.362222% from spread wasting, and 0.04888% from casts.

The overall impact on offense is 0.204813% + 0.185185% - 0.836543% - 0.055% = -0.501545% penalty in turns and 0.04888% + 0.362222% - 0.501545% = -0.090443 penalty to time. These calculations all make simplifications and assumptions so even though I perform them precisely, the results should be taken as approximate order of magnitude guesses.

This post has been edited by BlueWaterSplash: Jul 2 2020, 01:40

Wait I forgot, Basara actually did something different in his test. He didn't test haste with infusion, so haste might actually improve his turns if he gets around to testing it.

Now I did what was missing. (IMG:[invalid] style_emoticons/default/wink.gif) In fact, I should have posted this earlier. (IMG:[invalid] style_emoticons/default/heh.gif)

Just for comparison, I put again some results of my previous post.

(IMG:[imgur.com] https://imgur.com/1l50AMy.jpg)

Briefly speaking, seeing the results I think the use of Haste did not help the effect of the Infusion. It was the opposite.

Even if I had used the Infusion corresponding to the day of the week (Cold), I do not believe in a big change (maybe a minimum number of turns).

Looks like about 5.5% turn loss with haste this time. I did a rough calculation and estimate the overall influence of utilizing the Day of the Week element is expected to be 2% or less. It could be significant but can't explain that performance.

Maybe it's just due to randomness. A couple more trials with and without haste might be useful. When I compared haste in arena 60/65/80 I also got 1% turn loss once, but usually it was 2~3%.

One of the things I have been doing is that I always infuse holy+dark for tests. This helps for having a neutral infusion, but it also helps because I can analyze my battle log and check the ratio of holy to dark damage. There is a "normal" ratio of more dark damage, but on a rare run I would get more holy damage.

So I could sometimes identify the "bad" runs and throw away the data (I used this method for my upcoming Imperil test data). This occurs when bad luck makes you fight lots of tough HP undeads, and not enough celestial/sprite/human/etc. This method was easier for me than doing mass trials every day in the same arenas for weeks, to take an average.

P.S. - I now think your way is right to keep Haste off IA. Because I think you don't have IA on anything else, either. So it's fair, as you said before.

This post has been edited by BlueWaterSplash: Jun 22 2020, 04:24

So I could sometimes identify the "bad" runs and throw away the data (I used this method for my upcoming Imperil test data). This occurs when bad luck makes you fight lots of tough HP undeads, and not enough celestial/sprite/human/etc. This method was easier for me than doing mass trials every day in the same arenas for weeks, to take an average.

Isn't throwing away runs like this just doctoring the data to support your hypothesis? You should use all real data and just test more to get proper averages...

It's a regrettable alternative if you don't have the time or desire to take averages with many trials. As long as its disclosed, more thorough tests could be done later if the results prove sufficiently close and interesting. At least the method is consistent, utilizing the dark/holy damage ratio as a predictor of abnormality.

I threw away runs in my second batch of haste tests in arenas (without infusion) but I displayed all and explained why some were (partially) ignored for analysis. Usually it was because infusions made me slower which is actually the same problem Basara just had: his latest run with infusion was about the same speed as without.

Usually it was because infusions made me slower which is actually the same problem Basara just had: his latest run with infusion was about the same speed as without.

But in my case, I said that using Infusion saves me a little time (and turns), but using Haste makes me lose a little time (and turns). This means that the use of Haste cancels the beneficial effects of using Infusion. That's why the results were similar to the one where I don't use anything.

Still on the use of infusions, I did the tests below last month. The tests were done on Saturdays (Elec day), the first one I used Dark Infusion, the second Elec Infusion and the third I did not use infusions.

(IMG:[imgur.com] https://imgur.com/OuYa0aN.jpg)

In the specific case of the 150 round arena (DwD), at least for me, the use of Dark Infusion has a better effect than the use of the infusion of the day. This difference is smaller for the arenas of 125 and 110 rounds.

The importance of using infusions is proportional to the size of the challenge.

I was referring to comparing your runs [8] vs [15]....haste without infusion vs haste with infusion. The difference is below 2% which is not how it is supposed to be (expected ~6% or more). It suggests to me that your run [15] had abnormally bad luck, so that is the first one I would re-test. A complete re-test of runs [5], [7], [8], and [15] would be better for fairness.

When I tested these same things earlier (at my lower level) on numerous occasions I had a run that was significantly faster without infusion (everything else same). So I experienced the same problem as you, and I either needed to re-test or partially ignore those results. In my case, since I tested multiple arenas, I just chose to partially ignore those results and rely on the other data a bit more.

DwD isn't like regular arenas, especially regarding dark infusion vs elec infusion (even on Saturday) because of School Girls being very weak to dark. So I think it's not safe to assume that performance extends to arena 80. Though I don't know for sure how dark and electric compares on arena 80. I only estimated that the difference could be up to 2% turns (not elec vs dark damage) but it might also be less or zero.

Isn't throwing away runs like this just doctoring the data to support your hypothesis? You should use all real data and just test more to get proper averages...

Either that or do a different type of average either by trimming or winsorizing the highs and lows.
It there are exceptionally bad or good runs that may be the best approach even if that was not my experience.
When I used 1H, most runs were pretty comparable, and even with DWD maging where RNG plays an active role (imperil resists, number of riddlemasters, placements of SGs) I'm not seeing more than 2-3% deviation from my average result.

4+% from your expected result seems a pretty big deviation and it's quite unlikely there is no reason behind it (it can also be a bad day, when I'm sleepy or exhausted I refrain from taking time measurements, I know I would play like crap and get inconsistent results anyway).

Either that or do a different type of average either by trimming or winsorizing the highs and lows. If there are exceptionally bad or good runs that may be the best approach

4+% from your expected result seems a pretty big deviation and it's quite unlikely there is no reason behind it

Exceptionally bad or good runs are indeed the problem here and make the average-taking approach require an extreme number of trials. So focusing on median runs as you recommend seems best. It's actually similar to what I suggested, by just throwing out a rare data that you can identify as abnormal.

Regarding the dark/holy damage ratio as an indicator of abnormality, I forgot that since Basara has a holy rapier and I have a dark rapier, our ratios won't be the same. There is a fake ~1.5% elemental damage bias due to the order of strikes. The normal difference in elemental damage is expected to be about 1% favoring dark, so for Basara's holy rapier an average run might appear to do just barely more holy damage.

Finishing up my 1H analysis of swift strike and attack speed, the final possibility is that casting a spell moves the ailment cluster into the late turns. Rather than redoing the painful calculations, I reuse the first batch of results. In this case the extra harmful stuns don't matter since they occur when nearly all enemies are dead. Added PA doesn't help either, but improved Overwhelming Strikes does.

The overall impact on offense for 1% attack speed is 0.064935% + 0.285714% - 0.443077% - 0.04% = 0.245714% or about two-thirds of a butcher in turns, and 0.2% + 0.111346% + 0.245714 = 0.445714% benefit to time due to more convenient casts. Spread wasting no longer applies since stuns are ignored in this case.

The overall impact for grindfest is 0.023923% + 0.15873% - 0.279365% - 0.055% = 0.10373% benefit in turns and 0.148148% + 0.134286% + 0.10373% = 0.251878% benefit to time.

The overall impact with haste on is 0.151515% + 0.142857% - 0.64% - 0.04% = 0.102857% benefit in turns and 0.066% + 0.1575% + 0.102857% = 0.168857% to time.

The impact with haste in grindfest is 0.044563% + 0.079365% - 0.41284% - 0.055% = 0.024365% benefit in turns and 0.04888% + 0.18888% + 0.024365% = 0.073245% benefit to time.

This latest batch of results are modestly positive but more importantly safe even if I calculated imperfectly; it makes sense that if extended stuns occur at the end of rounds then there is no negative effect of swift strike, only positive. This is because counter attacks essentially don't drop at all in this case.

Note that if the timing works out such that the "final" turn of extended ailments occurs even a turn beyond the end of the round, you won't get the partial amount of extended ailments that precede it. You'll simply get no ailment changes for the entire round, and no effect on offense. I didn't explain this previously but it's due to how the turns count down: the key turn in which the turns remaining counter fails to decrement is that final one.

Even if you don't know how the exact timing of your spell casts work in conjunction with your particular amount of attack speed, the calculations suggest any timing is likely to have a decent result out of 3 main possibilities. The worst possible scenario is non-imperil style (no casts) but even that wasn't so bad.

There's some other interesting issues with haste and counter attack...once any enemy recovers from stun, it no longer matters whether you have haste or not in terms of the number of counter attacks you perform (except for increasing the stun duration).

It seems that ~95% of all enemies we encounter are chaosed enough to have over 25% attack speed bonus...combined with the 25% attack speed bonus of PFUDOR difficulty this means that whether we have haste or not, most monsters will attack 1H players twice every time they recover from stun.

And if we approximate that 1H style deals 3 counter attacks and 2 stuns every turn, then this creates a perpetual loop. Because in many turns, 2 monsters will recover from stun, and those 2 monsters will attack 4 times, for an average of 3 counter attacks and once again 2 stuns.

Unfortunately I think there is another phenomena that hurts the performance of swift strike in non-imperil style. Only enemies stunned in the first turn have increased duration, yet my prior calculation didn't take into account that when applied stun durations are not all uniform, it creates a "log jam" of enemies stuck at the counter cap when they all recover in the same turn.

Thus I revise my calculation by doubling the harmful effect of the added stun duration, as an approximation. After an extra added turn of being unable to counter attack, the "log jammed" enemies will probably clear out on average. There's a lot of complexity here, as even the default behavior without any attack speed can "self jam" anyway.

The overall impact for 1% attack speed is 0.064935% + 0.285714% - 0.886154% - 0.04% = -0.575505% penalty to turns and 0.2% + 0.222692% - 0.575505% = -0.152813% penalty to time. A doubled spread wasting term slightly mitigates the doubled harm from stuns.

The overall impact for grindfest is 0.023923% + 0.15873% - 0.55873% - 0.055% = -0.431077% penalty to turns and 0.148148% + 0.268572% - 0.431077% = -0.014357% penalty to time.

I think these are the only calculations that need revision. Haste alone by itself creates the "log jam" effect, so with haste on any additional "log jam" effect from swift strike is probably small. If spells are cast in a round the "log jam" effect does not apply to our calculation for various reasons.

I'm not sure my revised non-imperil calculation is better than my original, so it's good to consider both results as this is ultimately just a way of guessing the true performance. I roughly halved the benefit of Overwhelming Strikes in some of my calculations due to killing blows, when perhaps I should not have.

Finally, it should be kept in mind that these calculations were based on extrapolating my personal experiments with haste, tweaked to be pessimistic. Basara at high level has better haste results in turns. He doesn't notice turns/second boost which makes sense if using hotkeys for casting, and might generally be expected at high level due to less casts.

This post has been edited by BlueWaterSplash: Jul 2 2020, 23:42

A few more simple test results. (IMG:[invalid] style_emoticons/default/tongue.gif)

[A] Arena 150 rounds, using imperil and infusion (that is, a routine run (IMG:[invalid] style_emoticons/default/heh.gif) ) (29/06/2020 - Monday)

(IMG:[imgur.com] https://imgur.com/5IiRyuX.jpg)

[B] Arena 150 rounds, using infusion but without imperil (06/07/2020 - Monday)

(IMG:[imgur.com] https://imgur.com/t3ZOJSp.jpg)

[C] Arena 150 rounds, without imperil and infusion (01/06/2020 - Monday)

(IMG:[imgur.com] https://imgur.com/0MCVMyJ.jpg)

Note 1: In test [C], as seen in the image, I ended up using Imperil only once in the first round, but by accident. I think that does not invalidate the result. (IMG:[invalid] style_emoticons/default/duck.gif)

Note 2: The use of Paradise Lost in tests [A] and [C] was also accidental. (IMG:[invalid] style_emoticons/default/heh.gif) Since the use of this spell is on the same mouse button as the use of OFC, the error occurs when I try to use OFC with an Overcharge below 210.

Obvious conclusion: Using Imperil helps a lot. (IMG:[invalid] style_emoticons/default/happy.gif) Even having to perform almost 400 more actions, it still worth the effort. (IMG:[invalid] style_emoticons/default/wink.gif)

I forgot to ask this before: what is that "dot" that appears in the damage table? (IMG:[invalid] style_emoticons/default/huh.gif)

A few more simple test results. (IMG:[invalid] style_emoticons/default/tongue.gif)

[A] Arena 150 rounds, using imperil and infusion (that is, a routine run (IMG:[invalid] style_emoticons/default/heh.gif) ) (29/06/2020 - Monday)

(IMG:[imgur.com] https://imgur.com/5IiRyuX.jpg)

[B] Arena 150 rounds, using infusion but without imperil (06/07/2020 - Monday)

(IMG:[imgur.com] https://imgur.com/t3ZOJSp.jpg)

[C] Arena 150 rounds, without imperil and infusion (01/06/2020 - Monday)

(IMG:[imgur.com] https://imgur.com/0MCVMyJ.jpg)

Note 1: In test [C], as seen in the image, I ended up using Imperil only once in the first round, but by accident. I think that does not invalidate the result. (IMG:[invalid] style_emoticons/default/duck.gif)

Note 2: The use of Paradise Lost in tests [A] and [C] was also accidental. (IMG:[invalid] style_emoticons/default/heh.gif) Since the use of this spell is on the same mouse button as the use of OFC, the error occurs when I try to use OFC with an Overcharge below 210.

Obvious conclusion: Using Imperil helps a lot. (IMG:[invalid] style_emoticons/default/happy.gif) Even having to perform almost 400 more actions, it still worth the effort. (IMG:[invalid] style_emoticons/default/wink.gif)

I forgot to ask this before: what is that "dot" that appears in the damage table? (IMG:[invalid] style_emoticons/default/huh.gif)

damage-over-time. Like bleeding wound etc.

A counter-attack's damage is 75% of normal attack (http://ehwiki.org/wiki/Fighting_Styles#One-Handed)

I've been fundamentally using this fact for 1H calculations for a long time, but now that I check the wiki, it does not state that counter-attack damage is 75% of the main hit. Did anyone ever verify it? (If not, I will try to verify on school girls sometime).

A related question that I've been asking for a while: what is the probability of a stun to be inflicted upon a successful counter attack? (This is best to investigate with customized script).

This post has been edited by BlueWaterSplash: Jul 16 2020, 10:34

Counter-attacks don't have the elemental strikes. This is where the real difference is.
And this is also why Domino Strikes sucks as 2H: no elemental strikes either.

CODE

8827 base damage, level 391, Thursday
Spirit Stance, Overwhelming Strikes always active

Main Attack 11552 ~ 17738
Void Strike 5921 ~ 8592

Main Attack 22018 ~ 31454 crit
Counter 11552 ~ 17328


Penetrated Armor 3

Main Attack 24971 ~ 36994
Void Strike 12332 ~ 18497

Main Attack 44770 ~ 67155 crit
Counter 24971 ~ 36994

Counter attacks do the same damage as the main hit. As explained the real difference is the lack of elemental strikes and crits. This still throws off many of the calculations I have done in the past; eventually I may redo some. I also manually counted that 82 out of 100 counter attacks on school girls created a stun. Better tests should be done, and this could also depend on player or monster stats in some way.