A6 - Dual Wielding and how much is enough?

[Hanto]

Have been working with trainer supplied 3 levels of DW for my fighters for some time now and so far have enjoyed most of the game. Now I have to consider doubling up my DW levels to 6 vice present 3. Easy enough at present 3 SP per level, but still, is it worth it? Also considering buying training levels for Priest/Mages when more credits available.

Would like opinions as my Priest/Mages are also DW ready.

[Lilith]

You don't need a lot of dual wielding skill to dual-wield effectively: the benefits drop off pretty sharply after the first few points. How are your other combat skills looking? If you haven't unlocked Blademaster or all the battle disciplines yet, you're probably better off working towards those.

[Randomizer]

Increasing Dual Wielding decreases the damage penalty for using two swords. The to hit penalty goes away pretty quickly.

[Lilith]

The damage penalty also goes away pretty quickly. Slarty did an analysis comparing it to other skills in value-for-skill-points; unfortunately, the only link I can find to his thread no longer works, but the take-home message was that it's a pretty marginal improvement after the first few points.

[Hanto]

I appreciate what you are saying Lilith and Randomizer, but an actual # that most would believe enough is more my question then is the fact that the value goes down as level increases. So is 6 to much or should I go so far as to leave it at trainer supplied 3. I don't know if I have ever seen Jeff give any of his trainers the ability to supply all levels needed for particular skill just by paying the trainer. That would be a new one on me. Thanks for all replies.

[Lilith]

6 was enough for me. Even 3 might be enough, honestly. There isn't an exact number that's optimal because again, it depends on what your other skills are.

[Hanto]

Just to answer part of your question, my battle disciplines are minimums of 16 at this point for ' all ' characters and I've been using a lot of Blade Shield for a long time in this game. Blademaster has be unlocked, else I would never have got 5 levels Reposte. Only reason I stopped at 5 was it was costing to much and I still need one more Dex to unlock Gym. I kinda believe that Gym depending on how much it costs me will be of some importance in my game.

[Lilith]

How about Quick Action and Quick Strike?

[Randomizer]

Gymnastics is useless above normal difficulty because everything can hit you. It's only good for going ahead in battle.

 

Edit-

 

I found Slarty's calculation and dual wielding of 10 is the most really needed.

 

Originally Posted By: Slarty

Tests are in! And as usual, our theories are not quite right. And as usual, more testing is probably warranted.

 

I ran two tests. Both involved attacking the first NPC in the game with the first weapons in the game -- those "3-6" daggers. [i recorded damage BEFORE armor, that is, I added together the damage done with the damage blocked.] Each run, 7 in all, involved 20 successful attacks. I know that's not enough to be statistically bulletproof (hi, Thuryl) but it should give us a vague idea of what direction to look in.

 

I am suspicious that the daggers were a limiting factor; see below.

 

DUAL WIELDING AND TO-HIT

 

To-hit scores with 2 points of Strength and no Melee Weapons or Blademaster were expectedly low. For single wielding they were 28%, but for dual wielding they fluctuated between 13%, 18%, 23%, and 28%. With 0 DW they were often 13% and never 28%; with 20 DW they seemed to always be 28%. Given that this range of 15% is far off from the estimates of others (Randomizer's 50%) and that the weapon I was using had a bonus level of 3, I am suspicious that what dual wielding really does is remove the weapon to-hit bonus, then give a chance -- greatly increaed by DW skill -- to regain some or all of that bonus.

 

 

DUAL WIELDING AND DAMAGE

 

Let's begin with my data:

 

Test 1: Level 1, 6 Strength, 9 Melee Weapons.

00 SW: Avg 22.50, Range 16-28

00 DW: Avg 18.14, Range 10-26

02 DW: Avg 19.00, Range 11-26

 

00 DW / 00 SW = 0.80

 

Test 2: Level 61, 2 Strength, 20 Melee Weapons.

00 SW: Avg 27.35, Range 21-36

00 DW: Avg 21.55, Range 15-30

10 DW: Avg 27.00, Range 19-33

20 DW: Avg 29.30, Range 21-37

 

00 DW / 00 SW = 0.79

 

The good news is that, however paltry the data set, it is at least well-behaved, particularly the single-wielding to unskilled dual-wielding damage ratio. A penalty of 20% to damage seems pretty likely. This changes our formula above:

 

Poles Avg / Lvl = 2.5 * 1.0 * 1.0 = 2.5

Melee Avg / Lvl = 2.0 * 2.0 * 0.8 = 3.2

 

This makes swords more attractive. The other thing of note is that Dual Wielding skill did increase damage significantly. At least in this trial, 10 points was enough to counteract most of the penalty.

 

N.B.: there could well be other factors this test did not account for, such as weapon type -- I am suspicious of that one. So, we can hazard guesses based on this data but not ironclad conclusions.

 

That said -- given that the Dual Wielding skill is relatively cheap and easy to access, and that there doesn't *seem* to be much advantage to maxing it out, EW and DT still probably beat Ambidextrous. However, dual wielded blades now compete with and possibly outshine magic in terms of damage output, when you take Quick Action into account:

 

2.0 * 2.0 * 1.0 * 1.5 = 6.0 (rough estimate with 10 DW and 10 QA)

 

Compare to about 3.25 for Poles with QA, and 3 or 3.5 for spells. Admittedly Pole users can get the Slith bonus and save skill points on DW, but that simply won't make up the difference between 3.25 and nearly 6.0.

 

BUT ALSO

The SW numbers were not what we would expect. Given the assumption we've made since A4 about the damage formulae being simple, we would expect a maximum range of 1-2 * (2 + 6 + 9) or 17-34, with an average of 25.5, for the first test, and a max range of 1-2 * (2 + 2 + 20) or 24-48, with an average of 36, for the second test. But the averages were lower: 22.5 and 27.4... it appears that Strength and Melee Weapons do not add to damage as simply as we thought they did. It may be more like Nethergate: Resurrection where they add at differing proportions. More testing is required -- this would be very advantageous to figure out!

 

[Hanto]

QA and QS are still low, been concentrating on enabling skills, but this issue will be rectified soon because those two skills come in real handy and are very cheap at the moment.

 

Randomizer, I didn't understand ' It's only good for going ahead in battle '. What are you trying to say?

 

I had just got done reading Slarty's DW comments when I saw your post. Thanks. I may go above 6 on DW, but that will be later on I'm sure.

[Randomizer]

Gymnastics does two things: it decreases your chance of being hit and acts like higher dexterity for determining when your character attacks in battle. Most monsters in the later part of the game will always hit so trying to decrease their chance of hitting is useless.

[Lilith]

Originally Posted By: Randomizer

Gymnastics is useless above normal difficulty because everything can hit you. It's only good for going ahead in battle.

Edit-

I found Slarty's calculation and dual wielding of 10 is the most really needed.

I found that already, it's not what I was looking for. There was another thread comparing all the different combat skills to each other in one big preformatted table, in terms of how much extra damage per skill point you got out of each.

The upshot of it is that after the first 3 points or so, Dual Wielding should be a slightly higher priority than Anatomy and Lethal Blow, but a lower priority than everything else.

[Hanto]

So you're saying that even if my GYM was 40 or even for the sake of discussion 100 ( I know that that 100 doesn't exist ), some of the monsters would still get a hit on me at each and every opportunity. That doesn't sound like a physical attack to me, but instead a area effect attack delivered physically to only one player kinda like the slith bridge guards east of Fort Monestary. But at least their attacks affected multiple players just like area attack should. In other words I can understand why GYM would not help with the slith bridge guards however when an attack is being delivered physically at only one player, then GYM should have a modifying effect at least occasionally. Just an opinion, but your experience is definitely much greater then mine.

[Lilith]

Originally Posted By: Hanto

So you're saying that even if my GYM was 40 or even for the sake of discussion 100 ( I know that that 100 doesn't exist ), some of the monsters would still get a hit on me at each and every opportunity.

That's irrelevant, because you won't ever get your Gymnastics skill anywhere near 40 due to skill costs increasing as you raise them. The fundamental issue is that high-level enemy hit rates scale up faster than anything you can do to increase your evasion rate, and so by the endgame they'll be hitting you 95% of the time no matter what.

[ex post slarto]

You want this table:

 

http://www.spiderwebforums.com/forum/ubbthreads.php?ubb=showflat&Number=191367#Post191367

 

Lilith expanded on this in a follow-up by adding something else to it (I forget what), but I can't find that one.

[So Much Killing]

Slarty, regarding your analysis reposted by Randomizer, and in particular your comments at the end on expected vs. actual damage, there is definitely something off about Jeff's random number generator or the way he uses it. It seems streaky, or he is caching some of the results and re-using them when we wouldn't expect. The best evidence of this is the weird streaks of parry and riposte: you will sometimes get a guy with a 6% chance of parry succeeding three time (or more) in a row. On average, that should happen maybe 1 in 5000 tests, but it happens much more often than that. The same happens with low riposte numbers. It usually lasts a turn, then things go back to "normal". It's so obvious that when I see it happening in a turn, I switch my later PCs attacks to another enemy.

 

If I'm right, statistical tests may never match up with the formulas, even if the formulas are correct.

[So Much Killing]

Hanto, if you've unlocked Riposte, you're far enough along to be missing out by not using QA and QS. With QA above 10, your dual-wielder will hit three and often four times per attack. With increased battle speed and/or QS to get you to 10AP per round, you double that number, and can triple it if using haste in combination with adrenaline rush, speed potions, speed burst scrolls, or rods of alacrity. Even tough opponents die pretty quickly when you take 600 hp out of them each turn.

[ex post slarto]

Originally Posted By: So Much Killing

Slarty, regarding your analysis reposted by Randomizer, and in particular your comments at the end on expected vs. actual damage,

The quoted analysis is slightly old. Since then, we discovered that the damage formulas are slightly more complicated than we thought: for all damage, the "bonus dice" contributed by skills (strength, melee weapons, spellcraft, sharpshooter, etc) are equal to 75% of the skill total, not all of it. That is what is responsible for the damage difference and once that is taken into account, I believe the averages line up quite neatly with the formulas.

Quote:

there is definitely something off about Jeff's random number generator or the way he uses it. It seems streaky, or he is caching some of the results and re-using them when we wouldn't expect. The best evidence of this is the weird streaks of parry and riposte: you will sometimes get a guy with a 6% chance of parry succeeding three time (or more) in a row. On average, that should happen maybe 1 in 5000 tests, but it happens much more often than that. The same happens with low riposte numbers. It usually lasts a turn, then things go back to "normal". It's so obvious that when I see it happening in a turn, I switch my later PCs attacks to another enemy.

We can't rule this out, but I think it's more likely that Jeff modifies the chances in some way after they are displayed onscreen but before they go into the RNG. He's definitely done this with ceilings and floors on to-hit chances in many games.

Another possibility is that there is some unknown-to-us circumstantial modifier, which could possibly explain your observation that it "usually lasts a turn." For example, suppose a successful parry gives you a hidden bonus to parry for the rest of the round?

And it's also possible that your results are statistically normal and just look weird. For example, your 1 in 5000 number is about accurate if you do trials of three attacks at a time and look for only 3 parries WITHIN one trial. However, your anecdotal evidence presumably comes from an essentially undivided string of attacks, so you are looking for ANY chain of 3 parries. Were the attacks split up into groups of 3, this would include, for example, a group of 2 hits and a parry, followed by a group of 2 parries and a miss. This increases the odds substantially. According to my calculations, the odds of getting a run of 3 parries (with 6% parry chance) out of 9 attacks are about 0.14%, or about 7 in 5000. The chance will rise disproportionately as the number of observed attacks goes up, at first.

Either way, I really doubt the RNG is the culprit. If you are really interested, I think the next thing to do would be to test parry with a skill of 1 under repeatable, as-neutral-as-possible conditions, and see if we observe (a) an unusual number of parries, and ( any pattern to their occurance. But I am a bit skeptical.

[Earth]

You act before others and enemies.

[ex post slarto]

Originally Posted By: Earth Empires

You act before others and enemies.

This is inaccurate. Often you will act first -- it doesn't take a whole lot of Dex/QA/QS/Gymnastics to do so -- but if enemies have more than you, they WILL act before you in battle.

[So Much Killing]

You may be correct that there are other factors that affect parry and riposte. It's possible that the number displayed is some base chance that is modified by other factors (or is just plain wrong).

 

But at this point it probably isn't worth the effort of chasing it down. If I notice the same thing when Avadon comes out, I'll bring it up again. Or you can look for it in the beta if you're one of the testers.

[ex post slarto]

Yes, those are possibilities. But again, there is no convincing evidence to suggest that the displayed numbers are actually wrong.