Project 1999

Go Back   Project 1999 > Class Discussions > Tanks

Reply
 
Thread Tools Display Modes
  #1  
Old 05-19-2017, 03:09 PM
Launch Launch is offline
Skeleton


Join Date: May 2013
Posts: 19
Default An Empirical Warrior Dual Wield DPS Calculator

I couldn't find a fully fleshed out, empirical calculator for the mean damage per second of a dual wielding warrior (or any class for that matter) that properly interprets the impacts of all the important statistical variables, so I made one. Here it is.

Searching the forums yields a few older, less dynamic examples of this sort of effort like Yaolin's work from ~5 years ago. This is good work, and it along with the discussion in the corresponding thread served as a starting point for what I have put together. Mainly over a couple hours this morning - so there's a good chance I've done something that needs correcting.

Yaolin's spreadsheet allows for the comparison of any duel wield combination with the use of a relative value metric he calls "DPS Rating." This works quite well and scales properly with the most impactful parameters (attack speed, chance to hit) but doesn't handle proc damage (which scales differently and independently), or the more complicated relationship between a character's Strength and his damage per hit distribution. Finally, the DPS Rating doesn't have a tangible relationship to actual in-game damage per second - it just says, under some general conditions, how one set of weapons compares to another.

Initially my impetus for creating this was to properly value proc damage and answer a hypothetical question like this: is a 10/20 weapon with a 60DD proc better than a 11/20 weapon with no proc? Note that the damage per second of the proc is fairly easy to calculate - its the empirical calculation of the weapon's damage per second that's harder to get to. So I wound up with a full calculator with inputs for Level, Dexterity, Strength, Haste, and Chance to Hit, along with the parameters of a given Primary and Offhand weapon of course. Using great parse data from Sakuragi's Warrior Guide as a benchmark, the model comes pretty damn close for every single combination Sakuragi has documented.

I'll get to the key logic assumptions as I walk through an example output: a Level 60 Warrior with 255 Dexterity, 255 Strength, 100% Haste and a 70.5% chance to hit using a Sebilite Croaking Dirk (10/18) Primary and a Fist of Zek (16/30) Secondary.

Note that it's hard to find good data on Chance to Hit (and it scales with character vs. NPC level difference), but the best data I found was here. This example has a level 60 rogue landing hits ~75% of the time against a level 20 NPC; I've seen some commentary elsewhere saying hit rates range from 65% to 75%, and assuming a ~70% midpoint as the mean Chance to Hit lines up my model output with Sakuragi's parse data.

[You must be logged in to view images. Log in or Register.]

In Sakuragi's Warrior Guide he parses the damage per second of this exact weapon combination "at Level 60, 255 dexterity, 100% haste" at 63 DPS. With an assumed 70.5% chance to hit and an assumed 255 strength, my model calculates a total DPS of 62.73. Pretty damn close.

Here's a table of my model's calculation for each of the combinations that Sakuragi documents. The average DPS difference is ~0.25 over the 12 combinations. Note that these parses and modeled calculations of them exclude proc damage - it's pretty clear looking at the data that Sakuragi's DPS column doesn't include the damage from procs.

[You must be logged in to view images. Log in or Register.]

Using information from what looks like early P99 testing I structured the damage per hit distribution in a pretty straightforward way: 10% of hits take place at the Minimum Value, 20% at the "Magic Value" and 10% at the Maximum Value. The remaining 50% of hits take place at random values bounded by the Minimum and Maximum that are never equal to the Minimum, Maximum, or Magic Number. I didn't even have to try any other complicated distributions: this simple approach seemed to fit Sakuragi's data given an assumed 255 Strength and ~70% Chance to Hit. Note that in the above 10/18 Primary, 16/30 Secondary assumption that decreasing strength from 255 to 155 cuts the modeled DPS to 57.54 from 62.73. 100 strength is worth ~5 DPS, all else held constant.

Finally, I couldn't find a lot of information on Triple Attack Chance, but there is a mention in Yaolin's old thread that at level 60 half of Primary double attacks become triple attacks. With a 70.5% Chance to Hit, that results in the following for any given swing: 29.5% misses, 34.66% singles, 17.92% doubles, 17.92% triples in my calculations, at level 60.

I use the similarly documented duel wield and double attack chances of (Level + Skill) / 500 for Dual Wield Chance and (Level + Skill) / 600 for Double Attack Chance.

Obviously this all scales with NPC level (and AC) vs. Character level - but I would assume it does so linearly, making a higher DPS combination according to my model always the higher DPS combination. Here's a really high DPS combination, with procs included this time: two Scepters of Destruction.

[You must be logged in to view images. Log in or Register.]

Sakuragi's proc-excluded parses of two Scepters of Destruction was 73 DPS. My model calculates it at 72.97 excluding procs (again, pretty damn close), or 87.37 including them - if you were curious.

Disclaimer: my warrior is level 51 on Project 1999 and I haven't opened any EQ parsing software since 2004. I'm sure some raiding warriors can figure out if the output from my calculator is garbage or seems to make some sense.
Last edited by Launch; 05-19-2017 at 03:32 PM..
Reply With Quote
  #2  
Old 05-22-2017, 02:49 PM
Zentile Zentile is offline
Sarnak


Join Date: Feb 2015
Posts: 314
Default

Ambitious undertaking Axiomatic/Wazbom, I must say I'm really excited to see how it all ends up after feedback.

I have Yaolin's contact info in RL should you want to get a hold of him and walk through key assumptions. We discussed the % chance of triple attack one day in /gu and I believe his most recent thoughts were that 20% of double attacks becomes a triple attack (far below your 50% assumption).

Also, any thoughts on making this Monk/Ranger/Rogue usable?

Zentile
Reply With Quote
  #3  
Old 05-22-2017, 04:51 PM
Launch Launch is offline
Skeleton


Join Date: May 2013
Posts: 19
Default

Quote:
Originally Posted by Zentile [You must be logged in to view images. Log in or Register.]
Ambitious undertaking Axiomatic/Wazbom, I must say I'm really excited to see how it all ends up after feedback.

I have Yaolin's contact info in RL should you want to get a hold of him and walk through key assumptions. We discussed the % chance of triple attack one day in /gu and I believe his most recent thoughts were that 20% of double attacks becomes a triple attack (far below your 50% assumption).

Also, any thoughts on making this Monk/Ranger/Rogue usable?

Zentile

Thanks Z.

Adjustments for other classes should hinge exclusively on their respective Dual Wield / Double Attack skill values by level, and their "MaxSkill" value for each. The chance to Dual Wield or Double Attack on a given combat round is (Level + Skill) / MaxSkill, where MaxSkill is a fixed class-based number.

For warriors, this value is 500 for Duel Wield and 600 for Double Attack, thus rates for each are calculated as (Level + Skill) / 500 for Dual Wield Chance and (Level + Skill) / 600 for Double Attack Chance. For Monks, I think I've seen these values as 400 and 500 respectively in some other threads, but I need to confirm. Modifying my linked spreadsheet with this for each class should be pretty straightforward.

As for Triple Attack Rate (20% of DAs in your /gu convo vs. my assumption of 50% of DAs), you can see the discussion of it in Yaolin's original thread and where I got the 50% guess - but this should be pretty easy to measure for a 60 warrior and a simple parse.

Edit: Nevermind, looks like I misread the brief mention on Triple Attack Rate:

Quote:
IIRC the triple attack chance is 1/2 of the double attack chance and fires only on successful double attacks. Warrior DA chance should be (250+60)/600 -> about 50%, so basically it should be: 1/2 single attack, 3/8 double attack, 1/8 triple attack.
I've got it firing too frequently; will update.
Last edited by Launch; 05-22-2017 at 04:56 PM..
Reply With Quote
  #4  
Old 05-22-2017, 05:20 PM
Launch Launch is offline
Skeleton


Join Date: May 2013
Posts: 19
Default

Quote:
Originally Posted by Launch [You must be logged in to view images. Log in or Register.]
Thanks Z.

Adjustments for other classes should hinge exclusively on their respective Dual Wield / Double Attack skill values by level, and their "MaxSkill" value for each. The chance to Dual Wield or Double Attack on a given combat round is (Level + Skill) / MaxSkill, where MaxSkill is a fixed class-based number.

For warriors, this value is 500 for Duel Wield and 600 for Double Attack, thus rates for each are calculated as (Level + Skill) / 500 for Dual Wield Chance and (Level + Skill) / 600 for Double Attack Chance. For Monks, I think I've seen these values as 400 and 500 respectively in some other threads, but I need to confirm. Modifying my linked spreadsheet with this for each class should be pretty straightforward.

As for Triple Attack Rate (20% of DAs in your /gu convo vs. my assumption of 50% of DAs), you can see the discussion of it in Yaolin's original thread and where I got the 50% guess - but this should be pretty easy to measure for a 60 warrior and a simple parse.

Edit: Nevermind, looks like I misread the brief mention on Triple Attack Rate:



I've got it firing too frequently; will update.
Here is a link to a version with fixed Triple Attack Rates.

Using Chance to Hit as the sole variable to fit the modeled output to Sakuragi's parse data, an assumed hit rate of ~73% now minimizes error across his all of his parsed combinations. Pleasantly, this matches closer to the 74% measured in the best parse data I could find on Chance to Hit here.

Here's a screenshot of my original example with the corrected Triple Attack logic and 73% hit rate:

[You must be logged in to view images. Log in or Register.]

A level 60 warrior now will: miss 27% of the time, single ~36% of the time, double ~25% of the time, and triple ~12% of the time via this calculator.

And another Model vs. Parse Table:

[You must be logged in to view images. Log in or Register.]
Reply With Quote
  #5  
Old 05-22-2017, 05:42 PM
Lobus Lobus is offline
Kobold


Join Date: Jun 2016
Posts: 136
Default

bookmarked
Reply With Quote
  #6  
Old 05-22-2017, 05:58 PM
Vexenu Vexenu is offline
Fire Giant


Join Date: Feb 2012
Posts: 547
Default

Impressive.
Reply With Quote
  #7  
Old 05-22-2017, 09:23 PM
Zentile Zentile is offline
Sarnak


Join Date: Feb 2015
Posts: 314
Default

You probably need some sort of drop down box/selector to pick 1hb vs. 1hs vs. 1hb so it looks-up the right column in the "Skills Chart" tab. Doesn't really matter that much, but Infestation is 1hp and used by many warriors on this server.

Also would be cool to model a DoT proc vs. DD proc in your DPS calc too.

Minor suggestions in an otherwise kickass and long needed tool. Great work Axiomatic/Wazbom!

Zen
Reply With Quote
  #8  
Old 05-23-2017, 05:01 AM
Muggens Muggens is offline
Fire Giant


Join Date: Aug 2013
Location: Norway
Posts: 637
Default

Powerful post!
Reply With Quote
  #9  
Old 05-23-2017, 09:07 AM
Launch Launch is offline
Skeleton


Join Date: May 2013
Posts: 19
Default

Quote:
Originally Posted by Zentile [You must be logged in to view images. Log in or Register.]
You probably need some sort of drop down box/selector to pick 1hb vs. 1hs vs. 1hb so it looks-up the right column in the "Skills Chart" tab. Doesn't really matter that much, but Infestation is 1hp and used by many warriors on this server.
I originally intended to figure out how to use the weapon skill differences like this, but I've decided that the impacts of weapon skills are bucketed into the complicated character level vs. NPC level difference problem.

My understanding from reading around old posts is that Weapon Skill dictates Chance to Hit - and obviously this is a function of Character level (and skill) vs. NPC level (and AC). Note that the "ATK" rating of a character is an ensemble of two things: (1) chance to hit and (2) the damage distribution of a given hit. Changing Strength varies the ATK value and so does wielding weapons of which the character has different skills.

For example I'd bet that a level 60 warrior with 240 skilled 1HP or a 250 skilled 1HS won't see any chance to hit or DPS difference against a level 35 NPC, but I suspect he does against a level 65 NPC. That lumps the weapon skill impact into my broader caveat:

Quote:
Obviously this all scales with NPC level (and AC) vs. Character level - but I would assume it does so linearly, making a higher DPS combination according to my model always the higher DPS combination.
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -5. The time now is 03:46 AM.


Everquest is a registered trademark of Daybreak Game Company LLC.
Project 1999 is not associated or affiliated in any way with Daybreak Game Company LLC.
Powered by vBulletin®
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.