How does Critical Strike work?

[EEWorzelle]

Wereboar, sounds good. It will be interesting to see what you discover.

Note that, in the case of Psionics and Alchemy, that Senses and Dexterity are the primary controlling attributes, respectively, for those Skills. That doesn't prove anything either, but it is a point worth noting in our investigation of this stuff.

[ 09-02-2003, 01:20 PM: Message edited by: EEWorzelle ]

[ScottG]

I haven't gotten to the Senses vs. critical strike thing yet but I'll post it when I do. What I can say though (at least from my own observations) is that your chance for a critical isn't determined soley by level and/or critical strike skill. It also depends on the skill level of the character's weapon skill and possibly the skill level of the character's ranged or close combat skill (depending on what type of attack your making). In addition the class of the character may have some bearing on this ability as well (in much the same way that a fighter will receive more attacks earlier than any other class all else equal).

ex. as soon as my ninja achieved a 75 in dagger (close combat and critical skills the same) the freq. of criticals went up dramatically (but still not good). The character had a ninja modified critical skill of 101, Senses and Speed of 55, close combat of 61, dual wield of 72. The character would typically hit and penetrate 19 out of 20 "swings". At 74 in dagger the critical ratio was about 1 out of 50 "swings", which would be aprox. 1 out of 47/48 (taking into account the no hit/penetrate). At 75 in dagger the critical ratio was about 1 out of 20 "swings" (adjusted to 1 out of 19). Typically though it would be 40+ swings w/out a critical and then... 2 (and on rare occasions 3) in a row. If my ratios are anywhere close to accurate then your still not getting 10% from a 100 in critical strike skill - (more like 5%) at least as the current stats stand.

Note: even with multiple attacks (over 30) against a Slasher (level 21) vs. level 8 character - no critical was ever achieved. In otherwords level or some level dependent bonus to AC or such may perhaps be working against the probability of a critical. With this in mind I believe the Bank Guards are level 7 so that there might be a "skew" in favor of a critical in the results above IF its purely a level vs. level thing and not a level dependent bonus thing.

[ 09-03-2003, 01:42 AM: Message edited by: ScottG ]

[Wereboar]

Well, that 50% adavantage of the psionic at level 8 in my pürevious post was misremembered. It was 50% more at some point - skill 18 vs skill 12 - but it must have been at a lower level (or it improved a lot at level 8).

As i re-checked the numbers, the party was (still) level 8, but with high experience enough to level to 9 when i enter Trynton upper branches).
Psionic has 92 int and 60 senses, the mages have (98) 95 int and 45 senses.
(mage 1 was 3 pints ahad IIRC, but stopped at 95)
artifacts (int/senses): psionic 32, mages 27/27
mythologie (senses/int): psionic 29, mages 8/12
Note that the psionic starts with 9 points in mythologie.
At level 9, at the time the party was killed by some sprites, it was:
artifacts (int/senses): psionic 34, mages 29/28
mythologie (senses/int): psionic 32, mages 9/15

One thing to notice is that the psioinc has about 6 more points in artifacts, with the primary skill 3 (6) points lower, and the secondary 15 points higher.
So primary skill might count more than secondary, or might not - difference too low to decide IMO.

Most surprising thing is the difference in mythologie of the two mages, with nearly identical stats. Its the one that had 3 more points (the fairie) that is way behind the other one (elf)

Originally posted by ScottG: If my ratios are anywhere close to accurate then your still not getting 10% from a 100 in critical strike skill - (more like 5%) at least as the current stats stand.

funny you quote that figure, scottg, as i ran across a post at rpg during the last week (when IW was down).

in a thread about how to develop samurai,

http://www.rpgdot.com/phpBB2/viewtop...ninja&start=15

there's a post by otter, about half-way-down, that reads

"Critical Strike can't add more than 4% to your chance to insta-whack, so i usually leave this one to develop itself. "

So it sounds like you get +1% to instant kill for each 25 points of the skill, with the ninja getting +5%, with their critical skill bonus.

As a follow-up to reading this, I went back and played my ninja/rogue/priest/mage party from just before they entered the cosmic circle.

The ninja is using CoC/winterwand (12 attacks, but only 6 at 15% instant, plus 117 critical strike skill). The rogue has fang/stilleto (12 attacks, all 12 at 10% instant, no critical strike skill or item bonus).

If the formula I'm interpreting from otter's post is correct, the ninja's six strikes would have a 15% + 4.68% = ~19.7%, or an expected value of 1.18 critical strikes per round (conditional upon hit/penetrate).

For the rogue, the 12 swings at 10% have an expected value of 1.2 critical strikes per round (conditional upon hit/penetrate) - MORE than the ninja with the *best* weapon in the game.

Even with the ninja's critical strike maxed, this yields the same expected value of 1.2 as the rogue with fang/stilleto.

So in running the group again, I would have expected the rogue and ninja to be about equal at achieving criticals, in this party. And sure enough, that was the case. I ran them for several hours and counted exactly the same number of criticals (and interestingly, the same hit/penetrate ratio) for both.

I realise there are a lot of IFs and assumptions in this, but it seems to confirm what was stated by otter, and indicated by scottg.

If this is true, it's no wonder the critical strike skill seems so whimpy compared to weapons that do 10% and 15% instant kill! (even for getting how difficult they are to level up - the rogue had 10% instant in both hands, and more strikes with each hand, long before the ninja got six with her 15% one-hand). In fact, it would completely substantiate not using hybrids just for access to criticals.

[ 09-03-2003, 07:14 PM: Message edited by: sultan ]

[EEWorzelle]

Wow, excellent catch, Sultan.

Somebody actually measured it!

Okay, so it's 4% (5% for Ninja with bonus).

I haven't been posting (or even reading) on that board. Any word on whether it's fractional or just at 25 Skill Point jumps?

I read the post you linked and agree with Otter's Samurai development advice, too. Strength and Dexterity, then Senses, forget Int and Powercast for the Samurai. That's the way I develop mine, too.

But what I am asking for is if there are other threads you know of where this is gone into more, where players discussing observations like yours (which are excellent by the way).

I would call your, may I say brilliant experiment, convincing verification of what Otter posted. It is as good as I have ever seen on these boards. Very cool!

Scott's estimates, based upon his observations, were getting closer to those numbers also, which is additional "reality check" type verification for them.

[ 09-03-2003, 08:33 PM: Message edited by: EEWorzelle ]

i cant say for certain whether the calculation is fractional or straight 1% leaps at the appropriate times. the test i played really didnt have enough opponents (lots of 2-4 creature encounters) that i could say for certain whether the ninja was getting 19% or 20% IK with the cane.

i did do a fair bit of searching on RPG for followups to that, and found a number of people praising otter's knowledge of stats for the game, but never found a post with any reasoning behind how he knows nor collating such findings.

however, i also ran across a website of a regular (lanastasia, maybe?) who has chest and monster drop item chances out to two decimal places, including variations by experience level (of your party, not the monster). putting it together, you find out some fascinating things like there are 6 ways to get the enchanted wakizashi (can anyone say 6 samurai party?)

i'll see if i can track down the link...

here it is, from the VN board...

http://www.flamestryke.com/wizardry8...trykes_w8.html

eew, i see you posted in this thread, so you must know it.

the reason i mention it is that someone must have cracked into the code to get this kind of detail - there's no way you could reverse engineer that!

so perhaps otter did something similar?

[ - EDIT - ]

this got me thinking so i started another search. i havent found anything on criticals or general stats yet. but i did find an interesting thread on testing spellcasting that you all might find interesting (particularly spellcasting fiends, like eew )

http://www.rpgdot.com/phpBB2/viewtopic.php?t=9571

whoa, a fantatic thread on criticals, tested by someone using a game editor. pretty conclusive stuff, confirming some of our thoughts above.

http://www.rpgdot.com/phpBB2/viewtopic.php?t=9258

[ 09-03-2003, 10:56 PM: Message edited by: sultan ]

[Ziggurat]

All of this makes good sense, from my experience in many games. Unlike some, I don't measure, just have fun and observe. So I make mistakes in presumptions sometimes. I thought the Eagle Eye Bow was something worth having (it cost enough!), and it does some extra damage and has slightly better To Hit. But with a kill bonus of 1%, even the Hunter Bow beats it with 5%. I have a little more to say on the Bows thread.

i spent a fascinating evening testing this further.... my results from the other day were just counting criticals and *guesstimating* hit rates. i decided to be a bit more scientific.

hypothesis: the ninja's CoC, with 15% instant, plus the ~5% critical strike bonus, should get about twice the ratio of criticals to the rogue's fang/stilleto combo, with 10% instant.

method:

i used the same party described above, ninja/rogue/priest/mage, with the priest mage on defend unless needed for buffing/healing, wandering ascension peak.

the ninja is a faerie, currently level 26 using CoC and Rod of Sprites (didnt have this before running around - a nice bonus from the test), with a critical strike of 121.

the rogue is a hobbit, level 28 using fang/stilleto, with no critical strike skill or assassin's cap.

i counted the results of each swing of the CoC for the ninja and each swing for both fang/stilleto for the rogue. i ignored any results for the ninja's off-hand weapon and the rogue's bow.

the results:

........ninja...rogue
swings....234.....374
hits......225.....357
misses......9......17
# IK.......25......20
% hit...96.2%...95.5%
% IK....11.1%....5.6%

both characters have 85+ in close, dual, and the individual weapon skills. so it's no surprise they each hit/penetrated 95%+ (although i didnt record it, i believe only 1 of the 26 misses for both characters was actually a no penetrate).

the percent instant kills for the ninja is almost exactly twice that of the rogue.

so, this would seem to confirm the 20%/10% hypothesis discussed above, at least in as much as the ninja gets twice the ratio of instant kills to the rogue.

of course, the burning question for me is why are these figures only slighly better than half what we expected (20%/10%)?

one thing i noticed (and something i didnt record) is that, after two level ups during the test (and maybe i shouldnt have executed them), criticals became a lot harder to come by, because instead of fighting level 24ish monsters we ended up facing a lot of level 30 and 34ish monsters. before the first level up, the small sample figures i had were closer to 14% for the ninja and 11% for the rogue.

conclusion:

i think we can safely say that 5% is the equivalent figure for translating a maximum (125) critical strike skill into the figures shown for instant kill for weapons. however, there are clearly other factors at play in ultimately bringing those criticals to fruition, with opponent level seeming to be the key factor.

[ 09-04-2003, 09:19 AM: Message edited by: sultan ]

[EEWorzelle]

Good stuff, Sultan.

Yeah I post on VN also, but haven't been regularly following RPGDot. Dorkus' work is good. I wonder what editor he is using and whether he is able to use it with v 1.2.4.

The work He did on Powercast is inconclusive and does not match higher resistance (non-scientific) observations made in my games on the Peak. I have had the opportunity to compare no Powercast Spells with Powercast 100 against monsters at numerous resistances and often the damage there, on the Peak is 0 without Powercast.

Also I worked a particular example there a bit, with a comparison of a Alchemist and Gadgeteer casting Tsunami, and a Mage casting Blizzard. They were attacking 12 creatures with a resistance of only 90 or 100 in the Water Realm.

The results were all three casters hitting all twelve monsters for an average of

Alchemist - 107
Gadgeteer - 72
Mage - 96

All spells were successfully cast at Power Level 7. The Mage actually had slightly higher Powercast (107) versus Alchemist (99), but Blizard is only a Level 6 spell. I studied this one because I found the 1300 Damage in a single round of the Alchemist interesting and the fact that the three of them "averaged" over 1000. At the time players were posting that magic is less effective on the peak. I was also using this to discuss the wisdom of taking on multiple parties of foes at once, for a Magic-Heavy party (contradicts traditional wisdom but is often smart to do).

However I found the Bards attack spells, with or without Powercast - makes no difference for the Bard - were often ineffective (Damage = exactly 0) on the peak.

Good job on your confirming measurements. Now that is quality information. It is indeed interesting, that your measurements show less than half of what the weapons give. We have known there are other factors, like level of party and foe, but the impact of those is not quantified yet. What I am wondering is if it is really 4% for Critical Strike Skill = 100, like Otter proposes, or a higher % with these same "other" factors having an impact. i.e. maybe Scott's original presumption of 10% was closer to correct, in an apples to apples comparison to Kill % on weapons, but that other factors reduce both, in practice, to something much less than that. Just an idea.

No, wait a minute. That would not match the data in your previous experiment, would it, where you found critical strikes should be and were the same with the Ninja and Rogue using 4% of Critical Strike Skill in your Ninja calculation? Did you take numbers for your earlier test?

This is definitely getting much closer to a complete picture, isn't it?

[ 09-04-2003, 10:50 AM: Message edited by: EEWorzelle ]

[EEWorzelle]

Sultan, regarding cracking the code, 7thHeaven did a lot of work with items, to decode the more subtle characteristics (like which number or status bit means the Shiurken will return, which is for Extended, which 2-handed, etc., plus some data about chests. Some information used by the Wizardry 8 executable is kept in files, and these are what the program uses. He or she seemed to get frustrated when players didn't immediately drop everything else they were doing to help with the process.

That work is incomplete, but others are carrying on with it. My involvement in that has been minimal and off line, as a reviewer of others calculations.

That does not mean the binary code within the executable, nor the formulas embedded deep in that code can be known by this means. It is experiments such as yours and those by Dorkus that uncover those facts.

I think I have a save somewhere just before the battle on the peak mentioned in my previous post. Maybe I can do some experiments using the Bard, Gadgeteer, Mage, Alchemist, Priest and Psionic. This save is actually from my first game that reached that point. I foolishly game the Bard and Gadgeteer Powercast, too, and put points into it, but apparently it does nothing for them. At least there were no increases through "use."

I have to leave town for a few days, but will pick this up when I return... humm, after I have completed another thing I have committed to. I will be away from computers, and having a good time, but look forward to catching up here when I return.