Dexterity Parses - for Wiz

C

cvasihotep

Dalayan Beginner
I asked Wiz in /ooc in one of the instances when he was online and actually communicating with us in ooc why I do not see an improvement in hit rate when I get a +50 DEX buff. According to the ToK, hit rate should depend on that. Wiz said that it only seems that there is no improvement to the naked eye. He also advised me to post parses on the forums so he can take a look at them.

So I did. Here is the situation:

Duel between my 2 chars.

Defender - Monk, 65, Combat Agility 2, Lightning Reflexes 2, Defense 253 (+3 from items), Offense 250, Agility 224

Attacker - Beastlord, 65, Combat Ferocity 2, Offense 250, Weapon Skill (2HB) 245.

These are the constants of the duel. The Beastlord always hit the Monk, who just stood there, face to face, attack off.

I buffed my BST with his own Dex buff, so its dexterity was 224. I used a Shovel (2/38) so that i do not have to worry about the Monk dying in the duel.

This is how the fight started...

Hits Misses Block Dodge Riposte Total Hit/(Hit+Miss)%
66 100 7 12 5 190 39.8%

After some while...

111 167 11 14 7 310 39.9%

And then a little while more...

149 226 13 18 11 417 39.7%

I think it is pretty clear for everyone at this point that the hit rate given the above mentioned constants is about 40%.

Now ALL I did was remove the +50 DEX buff, and a few items with high DEX value on them.
Note: these items did NOT have any weapon skill modifier or offense modifier on them. They DID have some STR on them, so the BST lost some ATK. However, according to the ToK, hit rate does NOT depend on ATK.

I start the duel again and let it run just like above... I try to hit the same hit # and compare:

67 92 4 5 6 174 42.1%

Surprise, but I thought it's just statistical variation... I let it run more...

111 152 13 10 8 294 42.2%

I am totally baffled to I let it run some more...

190 267 16 17 13 501 41.6%

So... hit rate consistently stays around 42% with only 151 DEX, as compared to 224 DEX above.

This is NOT the first time I see that +50 DEX does not make a difference. I wish Wiz or the devs can look into it and see what is going on, or let me know if there is a clear reason why I got these confusing results.

If you want the log itself, I can provide it as well.

Sincerely,
Cvasihotep/Shoparla
 
Parses need to be done for a long time for it to be the most accurate. There might be a problem with dex and hit rate, but I doubt it from just seeing these parses.
 
Afaik it's been widely known by most melee's at the high end that dex is neigh on worthless. Certainly my 'strat' has always been to pump str over dex (I know Yona originally suggested this to me, so must agree). The combat precision line of AAs, however, do make a nice difference afaik.

As for the parses, they're too short. I suspect in reality DEX simply makes very little difference, if any, rather than having a negative effect :psyduck:
 
I agree that the parses are a little too short, although if I do the statistics, I can still hit the 95% confidence levels. Note that these are NOT the first parses I have ever done. In the past I got the same non-noticeable difference.

One more thing to notice too: in both parses, I stopped at different benchmarks to see what the H/(H+M) rate is. In both cases it remained constant - either 40 or 42%. The argument that the parses are too short would be strong IF at any of these benchmarks I would get different percentages, but I do not. It seems that the rate is pretty consistent, and the variations are not that "wild" once the game calculates what the hit rate should be.

I will parse more but, as I said, this is intended as an alarm signal, since I repeatedly get these kind of results. I see a big improvement in hit rate with Weapon Skill, with Offense Skill, but not with Dexterity - ever; I was hoping that +74 dexterity will give me SOME hint of a better hit rate, even marginal (2%?). Instead, I got the opposite - if anything, THAT should not have happened.

Best again,
Cvas
 
Each parse should be looked at as an event, not an individual swing. Get me 50+ parses that tell the same story and I will start to think that the affects of dexterity needs to be changed.
 
Do some parses on an actual NPC as opposed to a PC, and for an extended period of time (50 minute logs, oh god). Multiple instances of parses like this will help to enforce (or disprove) your point.
 
I do not think that damage vs PC on vs NPC should be different in structure. Besides, it is hard to know the defense/agility stats of the mobs you are fighting. To always be sure you compare apples to apples, you would have to fight the exact same mob, and ensure (HOW??) that its level and stats are the same every time. This is impossible.

As for parses, yes, they should be treated as individual events. Look at my numbers in the official post and see that I already did that, and the story is consistent.

I am currently running longer parses, but whatever the result is, it is clear that +73 dex makes a difference that, IF it exists, it is so minimal that it can only be seen in very long parses.

Just as a lesson in statistics, for those who might not know - the larger the effect, the shorter the parses needed to conclude that the effect exists.
The point of this post is that a whole 73 Dex makes little difference. Whether it is TOO little, only Wiz can tell.
 
cvasihotep said:
I do not think that damage vs PC on vs NPC should be different in structure. Besides, it is hard to know the defense/agility stats of the mobs you are fighting. To always be sure you compare apples to apples, you would have to fight the exact same mob, and ensure (HOW??) that its level and stats are the same every time. This is impossible.

As for parses, yes, they should be treated as individual events. Look at my numbers in the official post and see that I already did that, and the story is consistent.

I am currently running longer parses, but whatever the result is, it is clear that +73 dex makes a difference that, IF it exists, it is so minimal that it can only be seen in very long parses.

Just as a lesson in statistics, for those who might not know - the larger the effect, the shorter the parses needed to conclude that the effect exists.
The point of this post is that a whole 73 Dex makes little difference. Whether it is TOO little, only Wiz can tell.
Click to expand...

The defense of the mob doesn't matter as long as you have enough of the same mob type. I would suggest kobolds in Sorcs. Now the other thing you need to do is gimp your dps so that you can get longer fights. For this i would reccommend getting some dalium weapons. They are extremely fast and low dmg, which will give you more attempted hits for smaller dmg thus prolonging the fight and maximizing the # of swings. If you want to be super accurate about it make sure you dispell the mobs alot before engaging just in case some roamer or w/e buffed them.
 
i still dont see how this would be better than dealing with a PC whose stats are always the same, always known. Dex would need to affect hit rate in pvp as well, it doesn't make sense not to. The OVERALL message is what counts, that dex isn't doing much at all.

I am using a shovel, it's 2/38. This way I do not deal with the off-hand, and its low damage makes it easy to not worry about one char dying.
 
More parses:

DEX 151

Hits - 460 Misses - 739; H/(H+M) = 38.4%

DEX 151 (log 2)

Hits - 476 Misses - 778; H/(H+M) = 37.95%

------------------------------------------------------------------------------------------------

DEX 224 (log 1)

Hits - 497 Misses - 775; H/(H+M) = 39.07%

DEX 224 (log 2)

Hits - 490 Misses - 677; H/(H+M) = 41.9%

DEX 224 (log 3)

Hits 308 Misses - 486; H/(H+M) = 38.7%

If we add all the logs I parsed today we get the weighted average of today's parses... :

DEX 151

H: 1127 M: 1784; H/(H+M) = 38.7%

DEX 224

H: 1444 M: 2164; H/(H+M) = 40.0%

So I guess on average +73 DEX gives you +1.3% better hit rate, if that's even statistically significant. I think that at this point no one can still say that my parses aren't conclusive. All I am saying is that a whole lotta dexterity doesn't have an effect that can be noticeable after 4000+ swings...

Wiz - any comments? :)
 
Honestly that sounds about right. SoD works in very small percentages... (it's the nature of the beast).
 
True... although 73 DEX is a whole lot I thought... and all for 1%... makes you wonder if it's worth it, especially incrementally.

From earlier parses that I ran, i observed that a 32 drop in weapon skill drops the hit rate by about 3.2%, so it seems that the formula for hit rate has in it something like:

(weapon skill - defense skill)/10

While the divider for DEX/AGI is much higher, something like

(dex - mob agility)/60

The 60 was estimated to fit the above 1.3% increase, but is by no means accurate. The weapon skill numbers are much more accurate. The way I determined that is because my BST had piercing at 218 and H2H at 250. His hit-rate on the same enemy dropped by 3.2% when he was wielding a piercing weapon as opposed to just using fists.

I would adventure to propose the following formula for hit rate (HR):

HR = 50% + (Offense - Mob Defense)/10 + (Weapon Skill - Mob Defense)/10 + (Dexterity - Mob Agility)/60

of course AA's and styles will change that, but by Wiz's account they will do that in an additive manner, i.e.

- each combat agility level decreases HR by 1, 2 and 3% respectively
- each lightning reflexes AA decreases HR by 3% each (? this number is added here based on third party untested reports)
- each combat ferocity AA increases HR by 1,2 and 3%
- each unerring precision increases HR by 3% (same here, invented, but I would think UP and LR are perfectly equivalent, same as CF and CA).

and styles

- defensive decreases HR by x%
- aggressive increases HR (on you) by the same x%

etc.

All these numbers would be simply ADDED to the HR formula presented above.

PS: if the 60 divider for DEX/AGI is correct, it means that a +200 increase in DEX (from native 100 to maximum possible 305) only increases hit rate by 3.3%... sure, any increase is good, but you work so hard to get to 305 dex and all you get to show for it is a mere 3.3%, when a simple haste buff brings you 50%+ more damage...

It also means that the role of dex and agi are marginal - if you maxed everything else, getting additional dex/agi won't hurt... but one should not base gearing decisions on these stats only, and should not sacrifice other stats (HP, AC, mana, int, wis - depending on class) for dex/agi.
 
I make no claims about the validity of your parses. However, is a 1-2% increase really unreasonable for +75 dex?

Honestly you're looking at a player with dex. 151 vs. 224.
The max dex is 305 (or 485 when you toss hidden str into the mix).

Think of a naked base as ~100 dex, and a max of 485.
You're 50 to 125 points through a 385pt scale.
13-32% of the way to max. An increase of 1-2% accuracy for 20% of your max puts max dex at ~5-10% better for a max stat person.

Not to mention as you get more AA's, you specifically get accuracy aa's.

More of an increase than this would make those with max stats more obscene dps machines than they already are.
 
Alton,

Your points are actually very good. It may as well be that. However, can you explain how can you get to 485 max dex when you count hidden strength in? I know that you can get to 305 visible dexterity... how did you get to 485?

I guess my point was that, in the sea of things contributing to the HR calculation, dex is lost in the mix... +100 dex isn't that much after all.
 
I'm not sure what exactly you mean by "how," but you get the hidden strength AA and then wear gear that boosts your stats to 485 instead of 305. It doesn't show up on fomelo or in game because the bonuses past cap are hidden but you get benefits from them serverside.
 
You must log in or register to reply here.
Back
Top Bottom