[X] My Assistant

You've organized your post in such a mess that the data is completely incomprehensible, but your conclusion is nothing we did not already know. It was already known that Haste does help surviving, and may allow one to spread his attacks evenly instead of killing one mob after the other.

Haste is helpful only in that you do kill monsters a bit faster before they can unleash their SP attacks, but you're also having your Regen and draughts acting slower. Shadow Veil does a better job at that. Or... use feathers on your equipment to restore your evade stat, that works too, that doesn't consume mana and it also reduces your mana consumption.

You know, before being an aburdly tanky player, I've been through that moment when I had to run PFUDOR IW100 at Lvl320-340 (basically when I got my first Peerless Power from the lottery), so I've tried about everything to enhance surviving at the maximum. So you can trust my words on that. (IMG:[invalid] style_emoticons/default/wink.gif)

[imgur.com] Shadow Veil IW100 PFUDOR (Wed) [imgur.com] Haste + Shadow IW100 PFUDOR (Thu)
[imgur.com] Shadow Veil IW100 IWBTH #1 (Wed) [imgur.com] Targeting IW100 PFUDOR #1 (Thu)
[imgur.com] Shadow Veil IW100 IWBTH #2 (Wed) [imgur.com] Targeting IW100 PFUDOR #2 (Thu)
[imgur.com] Shadow Veil IW100 IWBTH #3 (Wed) [imgur.com] Targeting IW100 PFUDOR #3 (Thu)
[imgur.com] Shadow Veil IW100 IWBTH #4 (Wed) [imgur.com] Haste + Shadow IW100 PFUDOR (Fri)

Survivability is the reason that Haste exceeds its usual arena performance ranges in PFUDOR Item World, so this should apply to Shadow Veil as well, which was pointed out to even have an arguably better survival effect.

2,153 turns 0:16:13 (2.213 t/s) (Attack: 2023, Counter: 3357, Protection: 1, Spark: 1, Shadow: 2, Shield: 10, Heartseeker: 4, Regen: 19, Cure: 61, Full: 4, Mana Draught: 9, Spirit Draught: 5)

Here I replaced Haste with Shadow Veil on Innate Arcana. Subtle issues make fairly comparing the two a bit tricky but I think this result is indeed equal. Note that Shadow Veil provides less points of evade than Haste grants slowing by a factor of 25/33 = 0.75 so we do expect less turn penalty and less turns/second boost. I started casting Spirit Shield halfway through, earlier than with Haste.

2,029 turns 0:14:43 (2.298 t/s) (Attack: 1964, Counter: 3008, Protection: 4, Spark: 5, Shadow: 5, Heartseeker: 5, Regen: 15, Cure: 13, Mana Draught: 1, Spirit Draught: 1)
2,116 turns 0:15:44 (2.242 t/s) 6.9% slower (Attack: 2056, Counter: 3180, Protection: 4, Spark: 3, Shadow: 3, Heartseeker: 4, Regen: 16, Cure: 13, Mana Draught: 3, Spirit Draught: 2)
2,083 turns 0:15:15 (2.277 t/s) 3.6% slower (Attack: 2023, Counter: 3077, Protection: 3, Spark: 3, Shadow: 3, Heartseeker: 6, Regen: 16, Cure: 9, Full: 1)
2,070 turns 0:15:21 (2.248 t/s) 4.3% slower (Attack: 2009, Counter: 3217, Protection: 4, Spark: 3, Shadow: 3, Heartseeker: 4, Regen: 16, Cure: 13, Mana Draught: 3, Mana Potion: 1)

On easy missions it's expected Shadow Veil performs worse than Haste. My best IWBTH run was difficult to distinguish from just using Spirit Shield (or nothing) but the others were worse. I project Shadow Veil to worsen turns by 0.75*12% = 9% in arenas then make 2~4% turns/second back due to spread wasting and minimal casts. That's consistent with my results considering I hardcast and cure a bit more than the minimum in IWBTH Item World.

2,248 turns 0:16:47 (2.232 t/s) (Attack: 2150, Counter: 2569, Protection: 3, Spark: 3, Haste: 3, Shadow: 8, Shield: 5, Heartseeker: 3, Regen: 11, Cure: 45, Mana Draught: 4, Spirit Draught: 4)

Since I cure a lot in PFUDOR Item World even after applying Haste, I thought to try both Shadow Veil and Haste together. I began casting Shadow Veil in round 48 and Spirit Shield in round 67. This was identical to my worst Haste run. What happened? I would have improved, except the number of times I hardcasted Shadow Veil equaled the Cures that I saved, so the result became the same.

I could have put Shadow Veil on IA4 instead but then it would have been useless and hurtful in the early rounds. I do believe any natural evade and attack speed from my body, or Swift Strike, continues to improve my defense and clear time here. This should be true so long as a player needs to cure every few rounds.

2,266 turns 0:16:15 (2.324 t/s) (Attack: 2178, Counter: 2461, Protection: 3, Spark: 3, Haste: 3, Heartseeker: 5, Regen: 15, Cure: 41, Full: 1, Mana Draught: 6, Spirit Draught: 1)
2,333 turns 0:17:13 (2.258 t/s) (Attack: 2231, Counter: 2565, Protection: 4, Spark: 4, Haste: 4, Heartseeker: 3, Regen: 12, Cure: 53, Full: 2, Mana Draught: 4, Spirit Draught: 1, Health Potion: 1)
2,274 turns 0:16:23 (2.313 t/s) (Attack: 2170, Counter: 2438, Protection: 5, Spark: 3, Haste: 4, Heartseeker: 3, Regen: 14, Cure: 51, Full: 2, Mana Draught: 5, Spirit Draught: 1, MPot: 1, HPot: 1)

I checked my earlier claim that if I target and kill enemies individually once halfway through, my clear times worsen. I was correct but it's nearly identical. I left hovering on and didn't bother to target Stunned enemies. This is marginally faster and less tiresome than my normal targeting style, though less engaging and looks sloppy. Occasionally I left hurt enemies to die if they were no threat. Spirit Shield not required.

2,124 turns 0:16:13 (2.183 t/s) (Attack: 2006, Counter: 2441, Protection: 1, Spark: 1, Haste: 1, Shadow: 7, Shield: 7, Heartseeker: 3, Regen: 12, Cure: 52, Mana Draught: 4, Sprt Draught: 3, Pot: 7)

Friday is normally the fastest day for melee players as that's the only day that gives 5% boost to void damage. However, the light trouble caused by PFUDOR Item World is enough to make me slower because enemy damage also increases. Furthermore, enemy crushing/slashing/piercing/elemental attacks with 5% reduced mitigation actually increase in damage by up to 10~15%.

This could be why a few melee players, especially those in DW or other styles, report sometimes being fastest on the day of their weapon element. My best Friday run here barely managed to be equal to other days. I changed to Infusion of Lightning and Wind Shield but the damage was equal to before with neutral Infusion of Divinity.

This post has been edited by BlueWaterSplash: Apr 25 2020, 00:38

...

When you make a giant post like this, can you start putting in a hypothesis & TL;DR conclusion at the start? I'm starting to glaze over whenever I see this giant wall of vague 1H research (IMG:[invalid] style_emoticons/default/tongue.gif)

Also, the way you're laying out the data & comparing is really unclear. Can you consider a table layout like sssss2 provides for his comparisons? (See page 3 for that)

Particularly, it's unclear because it's not certain what you're even comparing. Like the 4 lines with the red 'x% slower's - slower than what? What run are you comparing to?

This post has been edited by lestion: Apr 25 2020, 10:06

The issue is more like a big lack in rigor when it comes to how to run an expertiment.
A proper, sure-fire, easy to understand way of doing things would be:
1) What am I trying to check?
2) What is the hypothesis (H0) of what I'm trying to check?
3) Description of the experiment.
4) Data
5) Analysis
6) Conclusion

ssss's way of doing is not right either. There is no hypothesis, no analysis, no conclusion, just a big dump of data that only a rigorous person could analyze properly. So many people (starting off with BWS...) tend to assume and take erroneous conclusions out of these.

Keep things simple, people. But "simple" implies a serious thinking ahead.

It's kind of a no-win situation when dealing with large amounts of complicated data that are sometimes comparing different things all at once. By omitting explanations and titles I shorten the post, so it's a trade-off. My private secret is to check my used spells for keyword "titles" (Haste, etc) to help figure out what I did in each run. Others should take my word that I have the proper/same spells on IA, as was discussed earlier.

I doubt I presented things in the best way but it's also far from the worst way. The data is actually already in a table layout, it's just not aligned and sideways from how sssss2 presented his.

Sometimes I explained exactly what my "slower", "faster", "stronger", "weaker", "quicker" meant (and what exactly they compare to) and sometimes I didn't, or just hinted at it. I'll say this: whenever I didn't explain something, I deliberately left it out to keep things shorter. I recommend others take my word that I'm trying to provide the most fair interpretation of results, or double check some calculations if they want more insight.

My previous post, despite being long, was mainly an afterthought. There was no clear, single hypothesis. Just an aggregate of interesting observations (to me). The real post is the Haste vs No Haste test in Item World.

I do plan to use these recent data posts to bring up various additional analyses, ideas, and conclusions. They will be coming over time.

I'm sure you can rearrange things in a way so that we aren't brainfucked anymore. (IMG:[invalid] style_emoticons/default/happy.gif)
If you're merely trying things out "just like that", thing I do frequently myself, keep it short: it's only there to give yourself hints about what seems to work, what doesn't. It is only then that more serious stuff occurs, like having an hypothesis and building up youe experiment around it.

In all cases, it's always a lot of work to have a strong conclusion to a stupid and apparently small hypothesis. That's how scientific research goes.

Here's PFUDOR IW99/100 data from Pope Killdragon who is level 362 and uses a 1H Shortsword [i.imgur.com] build similar to what decondelite used to have. Since he is stronger and more forged than I, he does not cure much. In this respect his results should be similar to a high level player, with Haste and Shadow Veil no longer benefiting from improved survivability.

He played on Saturday with wind infusions and shock shield. Each of his results are an average of ten runs. I think he uses OFC with non-Imperil style mainly. I'm posting this for him because he is avoiding the forum EXP bonus.

CODE

+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+
| Test type                       | Turns | Profit/Turn | Attacks | Counters | Imperil | Regens | Cure | Full-Cure | Health Pot | Mana Pot |
+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+
| Imperil, no haste, no sveil     | 1534  | 5.64        | 977     | 2283     | 436     | 11.1   | 13.4 | 0         | 0.1        | 17.6     |
+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+
| Imperil, haste, no sveil        | 1678  | 5.08        | 1119    | 1911     | 436     | 8.9    | 6.8  | 0         | 0.1        | 21.7     |
+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+
| Imperil, no haste, sveil        | 1568  | 4.85        | 1000    | 2261     | 445     | 11.4   | 6    | 0         | 0          | 19       |
+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+
| Non-imperil, no haste, no sveil | 1709  | 5.48        | 1535    | 3166     | 0       | 14.4   | 39.2 | 0.2       | 2.2        | 0.2      |
+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+
| Non-imperil, haste, no sveil    | 1875  | 5.36        | 1740    | 2552     | 0       | 10.8   | 19.6 | 0.1       | 0.7        | 0        |
+---------------------------------+-------+-------------+---------+----------+---------+--------+------+-----------+------------+----------+

Haste worsened turns by 9.4% in Imperil style and 9.7% in non-Imperil style. Apparently Imperil is not as incompatible with Haste as I thought. My new explanation: although casting Imperil causes Haste-enhanced Overwhelming Strikes to wear off, it does the same for non-Haste players to a lesser degree. Meanwhile, Imperil greatly increases the elemental strike damage of the main attack, which favors a player who uses Haste.

The results for Imperil Rapier style would likely be similar, so it's just non-Imperil (and maybe also partial Imperil) Rapier that can do better with Haste due to improved PA spread. In my opinion this result is still not bad and some time could be made back by turns/second boost, if you believe in it. The turn loss should also be smaller in arenas, just as with my rapier tests.

- snip -

Haste worsened turns by 9.4% in Imperil style and 9.7% in non-Imperil style. Apparently Imperil is not as incompatible with Haste as I thought. My new explanation: although casting Imperil causes Haste-enhanced Overwhelming Strikes to wear off, it does the same for non-Haste players to a lesser degree. Meanwhile, Imperil greatly increases the elemental strike damage of the main attack, which favors a player who uses Haste.

What I see here that I don't understand, is that in the three with Imperil, using Haste makes him use less Regen and Cure (and this is expected). : but it also seems to consume more Mana Potions, which I wouldn't expect at all; I would have thought that the Mana consumption from Haste would be more or less fully compensated by the savings in Regen+Cure.

Edit: and there are no data on Draughts; he uses them, does he ?

This post has been edited by mundomuñeca: Apr 30 2020, 15:54

What I see here that I don't understand, is that in the three with Imperil, using Haste makes him use less Regen and Cure (and this is expected). : but it also seems to consume more Mana Potions, which I wouldn't expect at all; I would have thought that the Mana consumption from Haste would be more or less fully compensated by the savings in Regen+Cure.

Edit: and there are no data on Draughts; he uses them, does he ?

The amounts of cures and regens are fairly low (and we can see from comparing imperil/non-imperil that imperil is where all the mana consumption happens anyway). But that aside: I'm not actually sure why haste costs more mana potions on average than non-haste - perhaps it could be related to IA consumption in the increased number of turns required? Though it'd be odd for IA to be consuming mana based on actual turns and not game ticks. (IMG:[invalid] style_emoticons/default/blink.gif)

If I had to guess, I would imagine that when using haste, the less counters means you have worse OC generation. If it's enough to drop out of spirit stance more, the cost of all magic goes up, resulting in more mana draughts/potions being needed

Pope Killdragon says he does use draughts but discounted their usage, as they are cheap. So we can't take his potion count as necessarily reflecting his mana usage. Also, using Haste made his draughts recharge too slow, so the balance shifted towards more potions.

Dropping out of spirit stance was a factor as well: he reports an average of 28 reactivations with Haste, and 17 without Haste. I'm guessing this is with Imperil, since that is the data you guys questioned. He affirmed he doesn't use OFC at all with Imperil.

He focuses on a profit/turn analysis which is pretty close between each playing method. His method of calculating can probably be questioned, especially since Imperil style costs are player market driven (draughts, aether shards, etc). What I extrapolated from it is that non-Imperil Rapier style could be the most profitable.

So non-Imperil Rapier style shouldn't be looked down upon despite being allegedly slower than Imperil Rapier style. I actually planned to study this next and have already begun gathering data. I'm not directly concerned with cost but will instead measure the performance of varying amounts of Imperil usage. Because rapiers do not need full Imperil as shortswords do.

I know from experience that the single Imperil style I promoted will never drop out of Spirit Stance with or without Haste (unless dealing with very small mobs) making that method even more attractive. I have long used Imperil whenever I have excess mana in order to keep my proficiency up, just never in tests or reported runs.

He focuses on a profit/turn analysis which is pretty close between each playing method. His method of calculating can probably be questioned, especially since Imperil style costs are player market driven (draughts, aether shards, etc). What I extrapolated from it is that non-Imperil Rapier style could be the most profitable.

Note that it's worth losing small amounts of average consistent profit for more chances at loot generation. Those rare legendaries or peerlesses could be valuable!

As if 2 Aether Shards over an entire PFFEST meant jack shit...

According to him a tool is used which analyzes Monsterbation's log and includes the market value of the actual equipment drops. That being said, trainings would affect this analysis, and the tiny chances of dropping something great (which ultimately could be a large part of expected profit) would likely not have been taken account for.

I might instead only look at credit losses (including draughts) and leave equipment out of it, acknowledging that a strong player who grinds heavily may make costs back with speed of play. Meanwhile, a player who only does a set amount of play per day, such as all arenas (thus not needing energy drinks) drops the same equipment either way and is more likely to be concerned with whether or not a few minutes of their time is worth that much small credits of savings. (I was not even convinced infusions were worth it from this point of view).

According to him a tool is used which analyzes Monsterbation's log and includes the market value of the actual equipment drops. That being said, trainings would affect this analysis, and the tiny chances of dropping something great (which ultimately could be a large part of expected profit) would likely not have been taken account for.

I might instead only look at credit losses (including draughts) and leave equipment out of it, acknowledging that a strong player who grinds heavily may make costs back with speed of play. Meanwhile, a player who only does a set amount of play per day, such as all arenas (thus not needing energy drinks) drops the same equipment either way and is more likely to be concerned with whether or not a few minutes of their time is worth that much small credits of savings. (I was not even convinced infusions were worth it from this point of view).

At the end of the day player comfort is another valuable metric too; generally, playing the game faster (and getting rewards faster) feels better and more fun. (IMG:[invalid] style_emoticons/default/tongue.gif)

At least that's worth more than ~150c per infusion. But maybe 25,000 per gum/vase pack (which is massively unsustainable without another income source like H@H) is too far, even if it's way more fun to blow through DWD in sub-500 turns (IMG:[invalid] style_emoticons/default/laugh.gif)

Butcher

I've copied some of sssss2 peerless warrior stats. A peerless rapier has pxp0 = 377, so 277 / 25 * 0.0854 ≈ 0.9462 adb part can't be forged. True base is 51.33 - 0.95 = 50.38 adb, most endgame legendary rapiers have 49~50. Each level of butcher adds 0.98~1 which forged and scaled to level 500 is 1.667 * 31 ≈ 50.6~51.7 damage.

A full peerless slaughter set ends with 15000 damage with shortsword and 14378 damage with rapier, depending on your distribution of stats, or 15400 damage and 14778 damage with 600 proficiency. A strong legendary warrior could be expected to lose 1 adb per piece, so 6 * 51.7 ≈ 310 damage. Including str and dex loss, most players can strive to reach 15000 damage with shortsword and 14400 with rapier.

Not wearing a power of slaughter loses 25.73 - 18.05 = 7.68 adb or 397 damage on the helmet, let's call it 400. Approximate damage increases of Butcher 5 are as follows.

Rapier + 0 Power of Slaughter: 5 * 51 / (12400 - 5 * 51) = 2.100%
Rapier + 1 Power of Slaughter: 5 * 51 / (12800 - 5 * 51) = 2.033%
Rapier + 2 Power of Slaughter: 5 * 51 / (13200 - 5 * 51) = 1.970% (compare with 2.312% in Research for 1H)
Rapier + 3 Power of Slaughter: 5 * 51 / (13600 - 5 * 51) = 1.911%
Rapier + 4 Power of Slaughter: 5 * 51 / (14000 - 5 * 51) = 1.855%
Rapier + 5 Power of Slaughter: 5 * 51 / (14400 - 5 * 51) = 1.803%

Shortsword + 0 Power of Slaughter: 5 * 1.2 * 51 / (13000 - 6 * 51) = 2.411%
Shortsword + 5 Power of Slaughter: 5 * 1.2 * 51 / (15000 - 6 * 51) = 2.082%

If you have weak proficiency, which could be expected on players near level 499 (as well as before the proficiency patch) you might subtract 400 damage from your total and move yourself one tier. These Butcher results can be expected to hold for a wide range of levels, since the relevant stats will scale down linearly together. Research for 1H may have utilized a disproportionately forged rapier.

Counter Attacks

Butcher is the only weapon potency that improves counter attacks. The Haste tests provided counter attack data for a variety of playing styles and situations, which can be used to calculate optionally applicable derating factors = 2.4 / (2.4 + 0.75 * counters) for Fatality and Overpower.

CODE

stage         spell usage   difficulty   counters/attack   derating factor
--------------------------------------------------------------------------
4 mob arena    haste          pfudor         0.85              0.79
5 mob arena    haste          pfudor         0.95              0.77
4 mob arena    none           pfudor         1.17              0.73
5 mob arena    none           pfudor         1.30              0.71
item world     haste           iwbth         1.18              0.73
item world     none            iwbth         1.65              0.66
item world     haste          pfudor         1.25              0.72
item world     none           pfudor         1.75              0.65
item world     imperil        pfudor         2.33              0.58

Fatality

Crit damage does not scale with level. Without savage armors or fatality, peerless crit damage is 50% + 1.36% + 1.61% + 1.24% + 1.49% + 1.11% = 56.81% at best, and 56.1% at worst. I'll copy Research for 1H and assume +60% crit damage without fatality, presumably from a savage piece or two.

Fatality gets better at high level because crit chance increases. The peerless full slaughter warrior has 53.2% crit chance. With Heartseeker it becomes 57.88% crit chance to do +75% crit damage. The base multiplier of the main hit is now up to 2.434 which becomes 2.49198 with Fatality 5, an improvement of 2.49198 / 2.434 - 1 = 2.382% (compare with 2.45373 / 2.4 - 1 = 2.239% in Research for 1H).

The old result is still useful since most players are not peerless. Note that fatality is even better without infusion (2.655% peerless, 2.499% old) or with skills (4.043% peerless, 3.838% old). The attractive damage boost with skills may be partially wasted because the damage is often overkill. Plus it's around twice this much damage boost when it crits and nothing when it doesn't, so it's not consistent.

In order to come to the opinion that Butcher is better than Fatality it is necessary to be slightly biased and insist on including counter attacks. Taking 1 counter/attack, the derating factor is 0.76 so Fatality 5 increases damage by 2.239% * 0.76 = 1.706%, or 1.821% if using the new peerless numbers.

Balance Armor

In conclusion, Butcher vs Fatality may still be a very close contest for 1H. Plus if butcher is overrated then slaughter may be as well, and a mix of balance and slaughter could be best for non-imperil rapier. Using less slaughter will help butcher as noted previously, but if those are balance pieces fatality will improve slightly more.

Balance Armor

In conclusion, Butcher vs Fatality may still be a very close contest for 1H. Plus if butcher is overrated then slaughter may be as well, and a mix of balance and slaughter could be best for non-imperil rapier. Using less slaughter will help butcher as noted previously, but if those are balance pieces fatality will improve slightly more.

It is not; I have an extensive spreadsheet which calculates a 'heart score' (same concept as arcane score for mages) which puts full savage balance below full savage slaughter quite definitively.

Fatality works out slightly better than butcher for straight damage, regardless of your setup. But it's not MUCH better. Butcher is close enough to outweigh it through counterattacks alone. I also believe overpower to be stronger than either, but the math for that is a little more plain.

Average monsters have somewhere around 20% parry, I believe (should be roughly comparable to resist - in which case it's slightly higher, even more in favour of overpower). If you imagine your unparried damage to be 100%, you lose 20% of it to parries, on average. Therefore, with 20% counter-parry, you gain 4% of it back - a 5% increase (0.84/0.8).

This is significantly higher than butcher or fatality. Consider this heart score table (numbers given for a max-forged level 500 player with 550 proficiency, rapier and force shield with strength & dex present):

CODE

|Gear              |   B0F0 |     B5 |     F5 |  B5/F4 |  F5/B4 |
|------------------|--------|--------|--------|--------|--------|
|S.Slaughter 100%  |  26231 |  26735 |  27221 |  27542 |  27635 |
|------------------|--------|--------|--------|--------|--------|
|S.Slaughter 0%    |  22291 |  22698 |  23103 |  23359 |  23436 |
|------------------|--------|--------|--------|--------|--------|
|S.Balance 100%    |  24379 |  24923 |  25441 |  25792 |  25891 |
|------------------|--------|--------|--------|--------|--------|
|S.Balance 0%      |  19877 |  20315 |  20719 |  21004 |  21081 |

The greatest gain is F5/B4 on the peerless savage slaughter row, gaining 5.3% combined.

This all applies to non-savage gear as well, and I cannot see a reason why it would not apply to a mix - crit chance does not layer well.

edit: the tl;dr takeaway from this for people who don't care: B5 is still preferable for 1H 99% of the time, Op4 would be ideal in fests where you're facing monsters with higher average parry, B5/F4 is probably still ideal for schoolgirl arenas

This post has been edited by lestion: May 4 2020, 21:34

Here's some counter attack data when enemies are killed one-by-one. I divide counters by 2 when calculating these derating factors because they don't benefit from Penetrated Armor in this case.

CODE

stage         spell usage   targeting   counters/attack   derating factor
-------------------------------------------------------------------------
4 mob arena    haste       individual       0.70              0.90
5 mob arena    haste       individual       0.75              0.90
4 mob arena    none        individual       0.94              0.87
5 mob arena    none        individual       1.06              0.86

Overpower

The monster parry formula is min(10, (dex / 100), (dex - level) / 75)). All monsters have 10% parry (19% at pfudor) except arthropod/celestial/dragon below PL 1056 (dex below 1000) or below PL 2250 at level 500 (dex below 1250). As well as elemental below PL 1953 at level 500, and mechanoid/undead below PL 1287 at level 500, and avion/giant below PL 945 at level 500.

Some arthropod, celestial, dragon, and elemental are clumsy enough to have below max parry. Research for 1H was performed below level 500 so some mechanoid and undead also had reduced parry. Average monster parry is slightly higher at level 500 and chaosing up to 27.1% has become more common, so new experimental data is needed.

In the meantime we can do theoretical calculations. If average monster parry is 20% then Overpower 5 grants (1 - 0.2 * 0.8) / (1 - 0.2) = 5% more hits. In Research for 1H monsters were stunned 62% of the time, this changes the above calculation to (1 - 0.2 * 0.38 * 0.8) / (1 - 0.2 * 0.38) = 1.645% more hits.

The player also gets Overwhelming Strikes in 1H style, which I'll estimate to provide 20% counter-parry on average without haste, and 60% counter-parry with haste. Stacked additively with Overpower 5 the above calculation becomes (1 - 0.2 * 0.38 * 0.6) / (1 - 0.2 * 0.38 * 0.8) = 1.618% without haste and (1 - 0.2 * 0.38 * 0.2) / (1 - 0.2 * 0.38 * 0.4) = 1.568% with haste.

There are times when Overwhelming Strikes grants 100% counter-parry which is not uncommon with haste. In these instances overpower does nothing. Haste was used when performing Research for 1H which some might consider a flaw. The experimental result was that Overpower 5 increased the hit rate by 1.3075%.

At that time, sssss2 had 66.7% block and 65.4% parry. The peerless warrior has 69.3% block and 73.6% parry. At 600 proficiency this will exceed 70% block. If we assume stuns are proportional to how often you block or parry, then (1 - 0.3 * 0.264) / (1 - 0.333 * 0.346) = 4% more stuns than what sssss2 measured are possible, which would worsen overpower.

The experimental stun rate of 62% used grindfest. Stun rate would be independent of mob size in the long run, but at the beginning of each round all monsters are unstunned and the player is temporarily swarmed with attacks, hitting the counters limit of 3 for the first few turns. In arenas the 1H stun rate may thus be slightly higher, worsening overpower.

Overcharge

There is a valid argument that the overcharge granted by overpower is worth extra, however the amount is difficult to determine and likely small. Some Haste tests earlier suggested that usage of OFC is discretized thus not usually affecting haste comparisons, and likewise may lend zero benefit to overpower. Vital Strike benefits but whether or not it helps a maxed out player is debated.

Penetrated Armor

Increased hits from overpower create more chances for Penetrated Armor. It's difficult to quantify how much an application of PA is worth but I like to think each is worth an extra hit with non-imperil style. Sometimes it's worth nothing, other times it's worth multiple future hits. There's up to a 1 - (1 - 57.88%) * (1 - 25%) ≈ 68% chance to apply PA.

Yet there is a logical error in thinking that more hits = more damage + PA. Missing only delays the chance to apply PA a turn. Overpower grants extra hits; butcher and fatality grant extra damage. The better question to ask is whether or not butcher and fatality decrease turns proportionately to the damage granted. For small increases it seems they do; thus PA can be ignored.

For non-imperil rapier spread style only, added hits from overpower have the valuable effect of preventing previously stacked PA from wearing off. I do think it's correct to add this effect. It wouldn't exist if PA didn't have a duration limit. What happens here is that more hits saves more than the same number of turns.

I estimate this effect is worth 0.68 * 0.68 * 0.66 = 0.3 hits, meaning overpower becomes 1.3x better with this viewpoint. The two 0.68 are the chances that PA previously existed, and that another PA is inflicted to refresh it. The 0.66 is because the first time an enemy is hit he can't have PA yet, and I estimate enemies to die in 3 hits.

Overpower = Fatality (1H non-imperil rapier spread style only)

Let's suppose the average monster parry rate today is 22%, then Overpower 5 without haste improves hits by (1 - 0.22 * 0.38 * 0.6) / (1 - 0.22 * 0.38 * 0.8) = 1.792% and may improve turns by 1.792% * 1.3 = 2.330% which is equal to Fatality 5. Keep in mind the math is fuzzy and PA wearing off applies less in arenas with small mobs.

For other styles the numbers suggest overpower is somewhere between 0.5x to 1x as beneficial as fatality. Even assuming 25% average monster parry, overpower can't give more than (1 - 0.25 * 0.36 * 0.6) / (1 - 0.25 * 0.36 * 0.8) = 1.940% more hits.

Future Tests

I don't have the proper weapons or resources to test overpower, but things I would like to see are:

1) Test overpower under today's monster population both with and without haste
2) Gather data not just for hits saved, but also for turns saved
3) Very brief test to check if 1H stun rate is the same in arena and grindfest

This post has been edited by BlueWaterSplash: May 6 2020, 09:10

Swift Strike

This potency effectively sacrifices offense (e.g. butcher ≈ 1 adb ≈ 0.4% of total damage) for 1.925% defense. In comparison a power helmet provides a choice of 7.68 adb with slaughter, 3.32% physical defense with protection, or 5.57% magical defense with warding. This is a good trade, the main problem is that high level 1H players eventually no longer have any use for defense.

The Best Offense is a Good Defense

Investigation has shown that even after it's no longer difficult to survive, defense continues to speed the player up on offense. Turns remain similar; rather this manifests as a turns/second boost each time a cure is prevented. If you cure once every 1~2 turns, expect a defense spell such as Haste, Protection, or Shadow Veil to improve clear time by ~10%. Scaled to swift strike, this is equivalent offense to butcher.

This benefit can be temporarily kept, at least in difficult challenges such as item world, by trading defense for offense in other areas (e.g. wear more slaughter, turn haste off, increase burden, or reduce agility). High turns/second or slow reactions (my cure is on a mouse button) can also prolong this status. Defense doesn't end when survival is achieved, but continues to help with gradually diminishing importance until invincibility is reached.

The Best Defense is a Good Offense

When I was level 250 it was suggested to me by a level 500 player that using power instead of plate would help me survive by killing off enemies faster. I tried and felt it was wrong. Back then power could cause death, especially with lower quality equipment. Even if you survive it can be a struggle of constant potion drinking which is ultimately not worth it.

That advice gains some truth at higher level with solid defense and very strong offense. In this realm offense begins to reflect on defense, although it no longer needs help by this point. At level 350 with some forging and DD5, I cleared enemies fast enough to live off gems. There is a relationship between offense and defense, and it can be possible to benefit from swift strike past level 400 depending on this balance.

Side-Effects on Offense

Assuming impregnable defense, the 1.925% attack speed from swift strike has numerous, complex side-effects on a 1H player's offense. They are easiest to determine by studying Haste and scaling its effects down. The primary contributor is loss in counters, which represents an overall loss of (2.4 + 1.5 * 0.75) / (2.4 + 0.75) = 12% damage assuming 1.5 counters/turn without haste.

More specifically, the loss in counters is (2.4 + 1.2 * 0.75) / (2.4 + 0.8 * 0.75) = 10% damage in arenas, and (2.4 + 1.75 * 0.75) / (2.4 + 1.17 * 0.75) ≈ 13% in item world or grindfest. This is approximately equal to a negative butcher per swift strike, and also applies to evade on a point-for-point basis.

Fortunately both action speed and 1H are weird, and there are secondary side-effects to counteract this. Here are the previous experiments gathered into a table. Interpretive liberties are taken for greater simplicity. Haste (+50% action speed) clear time results are extrapolated to Swift Strike 5 (+10% attack speed) by simple division.

CODE

stage          difficulty     1H substyle   infusion   haste (turns)   haste (time)   swift strike
--------------------------------------------------------------------------------------------------
4 mob arena      pfudor      spread rapier     no         -5.4%           -4.0%          -0.80%  
5 mob arena      pfudor      spread rapier     no         -5.0%           +0.2%          +0.04%
4 mob arena      pfudor      spread rapier     yes        -2.8%           -1.8%          -0.36%
5 mob arena      pfudor      spread rapier     yes        -2.0%           +5.4%          +1.08%
4 mob arena      pfudor    targeting rapier    yes        -5.0%           +0.0%          +0.00%
5 mob arena      pfudor    targeting rapier    yes        -5.0%           +0.0%          +0.00%
item world        iwbth      spread rapier     yes        -8.0%           +0.0%          +0.00%
item world       pfudor      spread rapier     yes        -4.0%           +4.0%          +0.80%
item world       pfudor    spread shortsword   yes        -9.7%            ???           -1.94%
item world       pfudor   imperil shortsword   yes        -9.4%            ???           -1.88%

This data is for players below level 390 and applies when player defense is impregnable. Clear time is what counts, but the amount of turns/second boost is controversial. Part comes from casting spells, which depends on the player and can reduce to near nothing at level 500. Another part comes from targeting issues, which depends on the player's vision and connection speed.

The turns performance of haste is expected to improve on high level players for various reasons. Data from a couple high level players suggest grindfest turn loss could be as little as 0.5% to 2%. If so then clear time would likely improve with haste even with perfect defense, but it's unknown by how much.

Recent discoveries suggest that blood wound is surprisingly almost as good as penetrated armor when spreading. In that style, action speed would slow down blood wound just as it does counters, so haste and swift strike are expected to be terrible with shortswords. The provided data is not nearly as bad as I would expect though.

Butcher ≈ Fatality ≥ Overpower > Swift Strike ≈ 0 (offensively)

From an offensive standpoint this is where 1H conservatively stands for now. Overpower might be 0.5x of butcher/fatality at worst. Swift strike is all over the place, providing competitive offense while player defense matters, and perhaps even in some cases when defense is no longer relevant.

It seems likely that swift strike is offensively neutral at worst for rapiers, or minimally negative enough to not matter. Therefore it can be considered as equivalent to a combined protection+warding suffix, or the protection spell scaled down. Players who appreciate defense, or those who pay a premium for a good parry weapon or shield, might be happy with swift strike even if it's useless at high level.

Agility / Burden Manipulation

CODE

burden     attack speed     evade
---------------------------------
  75           41.4%         0.0%
  70           53.5%         9.8%
  55           83.6%        37.2%
  40          100.0%        61.1%

Non-mithril peerless equipment with ethereal weapon has 55.3 burden. Action speed is diluted according to ((burden - 40) * 0.02) ^ 1.5 while evade is diluted according to (burden * 4/3%) ^ 1.5

The standard 1H approach is to aim for 70 burden. If you have high burden you can also increase agility for a tiny bit more PMit if desired. Most importantly your evade will be low, which will improve offense. A 1H player may have over 20% evade if unburdened. Any attack speed will be half wasted but its tiny reflection to offense will also be halved.

Another option is to aim for low burden (or use feathers) then also reduce agility to 0 to get rid of unwanted evade. If you trade evade with swift strike point-for-point, you effectively transform your swift strike into a positive butcher! This comes at the cost of having forfeited swift strike's defense boost. You are also limited in how much you can do this once you hit zero agility.

Eliminating agility causes PMit to be lost. Reducing the peerless warrior's agility to 0 drops his body PMit from 62.1% to 59%, and his total PMit from 82.9% to 81.5% (equipment PMit is 54.8%) which is still greater than his 80.8% MMit thus arguably superfluous. Meanwhile, doing so eliminates 14.7% unburdened evade.

Do Feathers/Haste make Swift Strike useless?

Despite its potential uses, swift strike may be a wasted slot because anyone can mimic its effect by using feathers or haste only in desired challenges. (The evade temporarily gained by using feathers is assumed needed for defense here, thus the issue of evade not being as good as action speed is dwarfed by their comparable applicability for defense).

I think that argument is partly true. Certainly feathers and haste are more flexible. On the other hand, swift strike does tack on after using feathers and/or haste. We've seen that boosting defense helps offense not in an absolute yes/no sense, but on a gradual scale that ends when you become a tank. If you want feathers or haste to start with, chances are you'll still benefit from swift strike afterwards.

I plan to use haste in arenas until past level 400. I prefer it to spirit shield and am not ready to use neither and live dangerously. There may come a time when haste is turned off; swift strike may still be helpful for another 30 levels. After that it will still be helpful in item world, maybe only with haste turned off.

Swift strike is stuck on permanently but at least it's close to offensively neutral. Feathers might stop being useful in item world and grindfest slightly before haste because of evade penalizing first half performance. You can reforge your weapon after swift strike stops being useful, change to a new one, or go mage.

Some of the things I wrote in my original post testing action speed are wrong. For now, it seems to me like all ailments inflicted on the first turn are likely to give you an extra turn, but only if you have nonzero action speed. Having haste and 0% attack speed meets this requirement, as does 1.7% attack speed without haste. Getting parried or doing some other action on your first turn will lose this opportunity.

So the start of rounds may not be randomly jittered like I originally thought. This might be abuseable if someone could figure out the exact rules. Maybe 1% or 0.1% attack speed would also provide this marginal head start. I wasn't able to confirm when the second time you can get an extra turn of ailments would be, which is a problem.

Another discovery: attack speed doesn't work like casting speed and haste. Attack speed bonus was already known to stack multiplicatively, but I had thought it applied like haste afterwards. Haste gives +50% action speed, lowering enemy attacks to 1/1.5 = 66% of normal, so a 33% reduction. Penetrated Armor increases from 7 turns to 10~11 turns, roughly half the time each.

On my Dual Wield persona with 29.5% attack speed bonus and haste off, my Penetrated Armor lasts for 10~11 turns as well, more often 10. It will be 11 if inflicted on the first attack of the round and can also be 11 in undetermined later rounds. The "turns remaining" counter goes from 7 to 0 only when the duration is 11 turns.

That's too much increase so attack speed must apply in the denominator, which would be consistent with the fact that attack speeds stack with each other multiplicatively, just like mitigations. The game tries to prevent you from reaching 100% attack speed, because that results in an invincible character.

Even so, 29.5% or even 30% attack speed should not be enough to reach 11 turns. It's equivalent to lowering enemy attacks to 70% of normal = 1/1.42857, and 1.42857 * 7 turns = 10 turns exactly. The wiki does vaguely mention that the game menu may display the attack speed bonus incorrectly. My body has 16.6%, my rapier has 1.925% from swift strike, and my waki has 12.10% attack speed. Agility from weapons also adds speed.

Finally, I checked how 29.5% (?) attack speed stacks with 50% (or 33% in the denominator) haste. They seem to stack well, it's multiplicative once again. Penetrated Armor increased to 15~17 turns, which is strange because it's the first time we see a semi-random variance of more than a turn. The counter goes from 6 to 0 if its 15 turns, and 7 to 0 otherwise.