Jump to content

  • Curse Sites
Help
- - - - -

Damage Downtime as a Component in DPS Calculation?


  • Please log in to reply
52 replies to this topic

#1 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 23 February 2014 - 05:17 PM

What is Damage Downtime?: DDT is the time spent not attacking that is unrelated to an attack's central action window.

What increases DDT?: Dodges, Movement, Knockdowns, Interrupts, Fear, etc.

Why care about DDT?: DDT expresses a more realistic vantage point of Damage per Second and greatly alters the current understandings of balance between weapons, weapon choices, conditions, and other points of contention.

How do you calculate DDT?: There are two methods. The first method involves timing average runs through content while acknowledging the number of dodges, the amount of control, and anything else that would interrupt your ability to maintain a perfect weapon cycle. From this average you form a general percentage and then multiplying your DPS by this number. The second method involves using a specific enemy and simply doing the same thing but instead of a percentage you just form a solid average number by adding the multiple runs together and dividing them out of your DPS.

[Scenarios]

Current DPS Method:

You are in an open arena with a boss that is stationary. You have a melee constant attack rate of 1s for 1,000 damage stable and defeat this boss in 50s therefore it has 50,000 HP.

DDT Method:

You are in an open arena with a boss that teleports. You have a constant melee attack rate of 1s that hits for 1,000 stable damage but every ten seconds the boss teleports 3s away from you. You defeat the boss in 50s however the boss now has ( 50s- 12s ) 32,000 thousand HP. Since DPS is measured out over time your DPS actually drops because of this by the same amount.

DDT 2:

You are in the same open arena with the same boss that teleports but you now have a long-range weapon that hits for 800/s however you do not need to move in order to cover the distance that the boss teleports. Presuming the boss has not changed HP ( 32k ) it would now take 40s to defeat the boss. Your DPS increases.

DDT 3:

You are in the same open arena with the same boss but you now have a 1s bleed of 1000/s per strike with a melee weapon hitting for 1000/s; the boss still teleports away from you ever 10s forcing you to move for 3s downtime to strike the boss again. Presuming the boss has the same HP ( 32k ) it now takes 25s ( half the strikes ) to kill the boss. Your DPS doubles.

[ "Advanced" ]

These are simple examples but a complex and more realistic example would be:

You are armed with a melee weapon that does 1000/s and a boss releases a nova of some sort every 20s starting at second 5 that you must dodge or you will go into the downed state. You are in an open arena (so no wall-dodging to stay in place) and orientation after the 1s dodge takes about 1s.

Every 30s regardless of the dodge you take damage from an aura that forces you to use a heal skill ( which will have a 2s cooldown for this exercise ). The boss also has a random 2s fear that appears at least once per minute. The boss also obviously attacks so you may require some for healing between the attacks.

Incorporating nothing else ( so the boss is static and doesn't leap around ) the battle takes you 6:08 minutes. What is your DPS? How much HP does the boss have?

[ Using the basic DPS model the boss must have 360,000 HP and you must be doing 1000/s as it reads. ]

[ Using DDT we calculate the amount of times events occur and the downtime from those events: ]

3 novas / minute = 3 dodges / min. From this we find that in 6m there are 18 dodges and each dodge and reorientation takes 2s therefore 36s ( 10% of the battle ) is spent dodging.

You heal at least twice a minute to survive so that is an additional 12s ( 3% of the battle ).

You are feared at least once a minute for 2 seconds which also requires a reorientation time of 2 seconds so 4 seconds a minute, which over the course of the battle is 6 times for a total of 24s ( 7% of the battle ).

[ Analysis ]

Combining the above times and subtracting them from the total time of 6:08 yields 72 seconds, converting 6:08 to seconds yields 368s so that leaves 296s active seconds of combat which is just shy of 5 minutes by four seconds. Calculating the bosses HP it is now 296,000 which is 18% less than the previous basic modeling and your DPS is 804.35. To get the DPS uptime you just take the total HP or damage ( in this case 296k ) and divide out ( like I said prior ) over the total time ( 368s ) producing 296,000/368.

804.35 / 1000 = .80435 ( obviously ) which means that your real DPS is 20% lower than the original modeling. Not calculating DDT is exactly what causes the huge disparities between conditions and direct damage as well as ranged and melee weaponry. When calculated into the DPS these variations shrink considerably since, for instance in our above scenario, a ranged player does not lose a second orienting themselves to attack the boss because they can attack at whatever range so the DPS goes up for them, perhaps not to the level of melee but definitely not as exaggerated as current methodology would suggest. The same is true of conditions; since conditions have a 1:1 uptime ( once a second ) the same battle may take considerably less time using condition rotations depending on factors such as movement patterns.

The basic conclusion is that the more mobile the boss the greater the effect on melee player DPS and this effect, when ignored, produces an extreme variation from the originating modeling which inhibits player choice and wisdom since enemies that move around a lot are more likely to be slain faster with ranged or condition builds than with melee builds due to DDT.




It's a lot to read. For those of you who completed the gauntlet, your thoughts? I've begun incorporating this immediately of course.

Edited by gw2guruaccount, 23 February 2014 - 05:27 PM.


#2 master21

master21

    Vanguard Scout

  • Members
  • 447 posts

Posted 23 February 2014 - 05:49 PM

Damage downtime exists, but imo it's impossible to really calculate it on average.
Amount of timw which you out of fight to dodge/cc/heal/etc depends on party compisition and personal skill level + cooperation.

To really calculate it you need to take a video of a player and make this calculation for him. Even fights like COE p2 p3 where in theory you dodge every ~10 second you can do it in many many ways. You can dodge into wall, you can dodge somewhere else, or you can use evade skill like whirl etc. so now every dodge is loss of dps. Sometimes good placed reflect, stability or aegis just makes dodging not required. Sometimes even you have reflect/stability/etc but you still dodge because you don't believe in 100% that those effects will be applied (so cooperation problem)

But for the basics it's quite true. If boss have nasty 1 shot mechanic or moves too much it's sometimes better to not melee him.

#3 Andemius

Andemius

    Sylvari Specialist

  • Members
  • 669 posts
  • Guild Tag:[herd]
  • Server:Desolation

Posted 23 February 2014 - 05:53 PM

Interesting, but I'm not sure what use it has beyond trying to guess enemy heath pools.

Surely its irrelevent whether your team is max dps or a mix of everything, if the boss has the same functions which cause "DDT", ie needing to evade.

#4 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 12:58 AM

Why would this get moved out of questions? It's a question. @_@

I'll never understand the mods ... @_@

#5 Acolyte

Acolyte

    Fahrar Cub

  • Members
  • 13 posts

Posted 24 February 2014 - 02:07 AM

Quite an interesting topic to bring up, considering how the community tends to assume ideal situations when theory crafting (i.e. static bosses/constant attack cycles). The principle of DDT modifiers to dps calculations clearly places a larger importance on condition and range use (as you demonstrated in the scenarios above), and I love the idea of shaking up the all melee/physical dps meta. Unfortunately though, I have a feeling the prevalence of the condition cap, and the dps loss from being out of range of party boons if you aren't stacking would easily out weigh the benefit of using range or non direct damage party setups (not to mention the abused functions of skills against walls). As it stands now, you can just melt away the health of anything you face (in dungeons) too fast for people to need to think about anything else than pure damage numbers in rotation. That being said, I'd still love to see the community gather more information on boss health and skill rotations/recharges to do some more specific math with, perhaps there are a few niche situations that benefit from a new approach to dps.

A little less on topic:
GW1 pve content was rewarding because it required an understanding of enemy party compositions/AI. When mobs used the same skills as you, you couldn't just c space through anything without wasting a ton of time/dps (just think of how much things would change if enemy mobs could dodge too). Foes in GW2 seem to typically have an auto attack and 1 or two generic skills that recharge as fast as your dodge does. There are countless things that could be done to improve the GW2 combat system, but until players start facing mobs that do more than los into a corner and swing a big hammer every 10 seconds, there wont be enough depth to care about anything else than big damage numbers.

I appreciate the write up, it was the first time I've seen anything like it on the forums.

#6 Beyond Freedom

Beyond Freedom

    Vanguard Scout

  • Members
  • 247 posts
  • Location:In your dreams.
  • Profession:Thief

Posted 24 February 2014 - 07:21 AM

View Postgw2guruaccount, on 24 February 2014 - 12:58 AM, said:

Why would this get moved out of questions? It's a question. @_@

It's not even a complete sentence...

#7 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 01:37 PM

View PostBeyond Freedom, on 24 February 2014 - 07:21 AM, said:

It's not even a complete sentence...
This is why I didn't want it here.

#8 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 01:56 PM

View PostAcolyte, on 24 February 2014 - 02:07 AM, said:

Quite an interesting topic to bring up, considering how the community tends to assume ideal situations when theory crafting (i.e. static bosses/constant attack cycles). The principle of DDT modifiers to dps calculations clearly places a larger importance on condition and range use (as you demonstrated in the scenarios above), and I love the idea of shaking up the all melee/physical dps meta. Unfortunately though, I have a feeling the prevalence of the condition cap, and the dps loss from being out of range of party boons if you aren't stacking would easily out weigh the benefit of using range or non direct damage party setups (not to mention the abused functions of skills against walls). As it stands now, you can just melt away the health of anything you face (in dungeons) too fast for people to need to think about anything else than pure damage numbers in rotation. That being said, I'd still love to see the community gather more information on boss health and skill rotations/recharges to do some more specific math with, perhaps there are a few niche situations that benefit from a new approach to dps.

Well it depends on the boss. Team composition itself may begin to change depending on the boss and how he moves or the layout of the floor meaning that instead of having your bufftastic heroes constantly work towards the enemy they may work grouped up away from the enemy to capitalize on whatever tactic is best. Also the condition cap is neither here nor there in small-man teams. I would personally suggest using it as supplemental damage rather than the goal for most classes simply because if you have a five second bleed attacked to a ranged weapon ( and that's sort of rare ) if you consider the party buffer having considerable DDT with some DDT added to all characters for maintaining those buffs in the party there shouldn't be an extreme ( if present ) constant overlap or looping problem with any given condition. If there is because it is supplemental and not the main focus you're covered anyway since you still do direct damage or you can change your rotation and equipment to compensate.

I have found through watching videos and whatnot that including all movement and elements anywhere from 30~50% of most boss fights is DDT ( with most of the videos of course being melee and the highest DDT being against bosses that either had a gimmick [think: Destroyer in CoE] or were highly versatile [think: Anything that teleports] ). That's a lot! It ranges anywhere from time spent buffing the self to time spent dodging and healing and as in the example above it adds up but the example above does not incorporate many elements on purpose such as any form of real channeling, being downed ( with the time it takes to rez a person presuming they are not "Dead" ), being constrained, boss cheese and specific tactics that cause imprisonment ( Alpha, Jormag, Spiders, etc. ), adds and trash, etc.

By not including these calculations we have greatly diminished our ability to understand and capitalize on the best methods for defeating specific enemies and dealing with specific situations.

#9 raspberry jam

raspberry jam

    Vigil Crusader

  • Members
  • 4791 posts

Posted 24 February 2014 - 03:09 PM

View Postgw2guruaccount, on 23 February 2014 - 05:17 PM, said:

It's a lot to read. For those of you who completed the gauntlet, your thoughts? I've begun incorporating this immediately of course.
Nearly completely pointless. Your idea of "DDT" is simply a multiplier on the static DPS. The size of the multiplier depends on the content, since the content decides how much you have to dodge and so on.

However, this is because the DPS calculation itself is primitive. If you think of the time spent dodging etc. as the time where you cannot activate skills, you will understand that the average minimum effective cooldown (for each skill) is a function of the length of the average chunk of downtime. What this means is, to present a simple example:

Assume that you can land one swing every 10 seconds, because the rest of the time you need to dodge and/or run to teleporting bosses etc.
Let's say that you can pick either a weapon that has a 1000-damage skill with a cooldown of 10 seconds, or a weapon that has a 700-damage skill with a cooldown of 5 seconds. Theoretically, the first weapon would have a DPS of 100 (from that skill) and the second one would have a DPS of 140 (from that skill). Second weapon seems 40% better, right? However since your effective cooldown will be 10 seconds, the DPS of the first weapon is still 100 while the DPS from the second one drops to 70. Suddenly, the first weapon is 30% better than the second.
Of course one swing every 10 seconds is an extreme case, and weapons with only one skill does not exist. But the same principle can be applied in all situations, by simple summation of all cases.

tl;dr: consider the granularity of the damage packet flow, do not just apply a multiplier.

#10 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 04:00 PM

View Postraspberry jam, on 24 February 2014 - 03:09 PM, said:

tl;dr: consider the granularity of the damage packet flow, do not just apply a multiplier.
Actually it does. Your example has a major flaw in it and that's a static cooldown. DDT does not work off a static cooldown at all.

If a boss moves out of range of melee every 5 seconds but does not move out of range of a ranged weapon the DDT of the ranged weapon is zero and the DDT of the melee weapon is however long it takes to re-orient ( 2s ) so over the same period of time, let's say 18s, the ranged weapon with 800 dmg suffers zero DDT while the melee weapon with 1,200 damage suffers 6s DDT or 6x1,200 Dmg loss. Calculated out 800 x 18 = 14,400 and 12,00 x (total time - DDT = 12) = 14,400.

The coefficient differentiation between two weapons being .4 or greater is very rare in game so this example is not a proof of anything. However a weapon with 33% damage base doing exactly equivalent damage due to DDT is imperative to recognize.

Edited by gw2guruaccount, 24 February 2014 - 04:03 PM.


#11 Veji

Veji

    Vanguard Scout

  • Members
  • 479 posts

Posted 24 February 2014 - 04:15 PM

WoW theorycrafters battled with this in days yore.  They came up with two terms:  DPS and realistic DPS.  DPS is when you stand and deliver without having to move and you just deliver your skill set onto the boss without interruption.  "Realistic" DPS is the DPS you do in any given raid or group scenario, where you have to account for movement, dodging, etc.

Once again, this game has no "Recount" or anyway of showing either types of DPS and the results of individual encounters or group encounters, so its relative to the flying spaghetti monster and his/her thoughts on quantum singularities.  Its a no-brainer that a melee DPS is going to lose DPS time when they get CC'd or have to dodge an attack.  If i had to theorycraft myself, i'd say this is in relation to balancing out melee and ranged DPS, and why ranged DPS is perceived to do less damage than melee DPS, as ranged attackers don't have to move as much as a melee DPS.  The speedrunners have mitigated this by finding ways to "corner" bosses, as to maximize melee DPS and thus shun ranged DPS.

#12 master21

master21

    Vanguard Scout

  • Members
  • 447 posts

Posted 24 February 2014 - 04:20 PM

Just introduce dps meter to gw 2 and it will easly show real realistic dps.

#13 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 04:24 PM

View PostVeji, on 24 February 2014 - 04:15 PM, said:

WoW theorycrafters battled with this in days yore.  They came up with two terms:  DPS and realistic DPS.  DPS is when you stand and deliver without having to move and you just deliver your skill set onto the boss without interruption.  "Realistic" DPS is the DPS you do in any given raid or group scenario, where you have to account for movement, dodging, etc.

Once again, this game has no "Recount" or anyway of showing either types of DPS and the results of individual encounters or group encounters, so its relative to the flying spaghetti monster and his/her thoughts on quantum singularities.  Its a no-brainer that a melee DPS is going to lose DPS time when they get CC'd or have to dodge an attack.  If i had to theorycraft myself, i'd say this is in relation to balancing out melee and ranged DPS, and why ranged DPS is perceived to do less damage than melee DPS, as ranged attackers don't have to move as much as a melee DPS.  The speedrunners have mitigated this by finding ways to "corner" bosses, as to maximize melee DPS and thus shun ranged DPS.
If you have a stop watch and use the game's damage log you're more than capable of deciphering both. The real value of DDT is not that it's a "no brainer" but as you get more complex you begin to be able to make sounder decisions. For instance let's take our above example and add a bleed of 10s to each cut, the DDT can be reduced by the bleed damage modifying the difference between two melee weapons even. The key here is not to prove whether ranged or melee is better it's just a tool. WoW has different mechanics so old methodologies don't work anymore. It would be similar to comparing chess to checkers not necessarily in complexity just in means of calculation.

By the way DDT can be used to monitor and improve your own behavior. Counting your DDT when you dodge "unnecessarily" and the like can also tell you how effective of a combatant you actually are.

View Postmaster21, on 24 February 2014 - 04:20 PM, said:

Just introduce dps meter to gw 2 and it will easly show real realistic dps.
This would definitely make all Theorycrafting easier. Or rather sort of moot.

Edited by gw2guruaccount, 24 February 2014 - 04:26 PM.


#14 MazingerZ

MazingerZ

    Golem Rider

  • Curse Premium
  • Curse Premium
  • 2274 posts
  • Profession:Guardian
  • Guild Tag:[CYRL]
  • Server:Tarnished Coast

Posted 24 February 2014 - 04:41 PM

View PostVeji, on 24 February 2014 - 04:15 PM, said:

WoW theorycrafters battled with this in days yore.  They came up with two terms:  DPS and realistic DPS.  DPS is when you stand and deliver without having to move and you just deliver your skill set onto the boss without interruption.  "Realistic" DPS is the DPS you do in any given raid or group scenario, where you have to account for movement, dodging, etc.

This...  The only time this ever mattered was when high DPS specs required a certain level of uptime to reach maximum efficiency, such as Combat Rogues and Bandit's Guile.  Prior to Mists, the cycle would reset if you had to switch targets.  So while Combat Rogues had potentially high DPS, that DPS would suffer in fights that had quick target changes.  Same thing for combo points, but most rogues had a dump for that (SnD).

Most theory-crafting in DPS only deals with ideal conditions because there are fights where you do have to 'burn the boss' largely as a DPS check (which in WoW is both gear-related and skill-related) because unless you're dealing with fights that require you to cycle-up over a long period (relatively, more than say 5 seconds) to reach max DPS, its generally a negligible different between max and realistic.

Edited by MazingerZ, 24 February 2014 - 04:42 PM.

It's okay to enjoy crap if you're willing to admit it's crap.
Every patch is like ArenaNet walking out onto the stage of the International Don't Kitten Up Championship, and then proceeding to shiv itself in the stomach 30 times while screaming "IT'S FOR YOUR OWN GOOD! IT'S FOR YOUR OWN GOOD!"

#15 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 06:15 PM

View PostMazingerZ, on 24 February 2014 - 04:41 PM, said:

Most theory-crafting in DPS only deals with ideal conditions because there are fights where you do have to 'burn the boss' largely as a DPS check (which in WoW is both gear-related and skill-related) because unless you're dealing with fights that require you to cycle-up over a long period (relatively, more than say 5 seconds) to reach max DPS, its generally a negligible different between max and realistic.

This isn't the case in GW2 though. Using the previous example the total damage expected out of the melee was 21,600, the actual damage dealt was 14,400. 33% is not small and it's based on in-game mechanics. I think it's time we graduated from WoW comparisons and mathematics because the games are different. GW2 doesn't use gear-checks as a means to lower your DPS and it doesn't use skill-checks as a means to take up damage slots. Using the same modeling for a totally different game is something like saying using basic math from a 1980s rogue will give you the same understanding of a multi-faceted complex system like GW1.

Let me express it this way:

GW2 uses "time sinks" to effect your DPS with most management being reliant on distance and disruption.

WoW ( and it's clones ) use "stat sinks" to effect your DPS with most management being reliant on skill choices and counters as well as gear choices and stats.

Using them interchangeably will produce disastrous results.

#16 raspberry jam

raspberry jam

    Vigil Crusader

  • Members
  • 4791 posts

Posted 24 February 2014 - 07:02 PM

View Postgw2guruaccount, on 24 February 2014 - 04:00 PM, said:

Actually it does. Your example has a major flaw in it and that's a static cooldown. DDT does not work off a static cooldown at all.

If a boss moves out of range of melee every 5 seconds but does not move out of range of a ranged weapon the DDT of the ranged weapon is zero and the DDT of the melee weapon is however long it takes to re-orient ( 2s ) so over the same period of time, let's say 18s, the ranged weapon with 800 dmg suffers zero DDT while the melee weapon with 1,200 damage suffers 6s DDT or 6x1,200 Dmg loss. Calculated out 800 x 18 = 14,400 and 12,00 x (total time - DDT = 12) = 14,400.

The coefficient differentiation between two weapons being .4 or greater is very rare in game so this example is not a proof of anything. However a weapon with 33% damage base doing exactly equivalent damage due to DDT is imperative to recognize.
All skills have static cooldowns. You are right in that over time the difference between two weapons become a simple multiplier, but so what? The interesting thing is why there is a difference. Comparing a melee weapon to a ranged works like my example, with the exception that nothing is increasing the cooldowns. The summation across all packet possibilities is still carried out in the same way.

#17 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 07:26 PM

View Postraspberry jam, on 24 February 2014 - 07:02 PM, said:

All skills have static cooldowns. You are right in that over time the difference between two weapons become a simple multiplier, but so what? The interesting thing is why there is a difference. Comparing a melee weapon to a ranged works like my example, with the exception that nothing is increasing the cooldowns. The summation across all packet possibilities is still carried out in the same way.
That's not how it works.

Let's say you're frozen in a block of ice by Jormag for 8s. Does it matter if your skill has a 2s cooldown if you're frozen in a block of ice? No. DDT does not directly correlate to a cooldown model. DDT is a part of the DPS model and it is not a replacement for the DPS mode.

All of my models for now have used the same Cooldown ( 1/s ) just for ease of digestion but you can easily swap and interchange all the numbers you want for comparison. So a bow that fires at a rate of .5s and a sword that swings at a rate of 1s have nothing to do with DDT at all. The fact that our teleporting enemy doesn't get out of range of the bow is what makes the DDT zero, cooldown is not effected, and the fact that it gets out of range of melee is what increases the DDT to 2s, it has nothing to do with how often the sword swings, ad infinitum.

#18 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 07:39 PM

I want to add this to explain why you can't incorporate this into a cooldown model. "Chilled" increases the cooldown of skill and decreases your movement speed but it does not decrease the active speed of behavior or game mechanics. For instance when your skill is fully charged it will fire at the same rate regardless of being Chilled or not. While you can use chilled with DDT for movement and realignment you cannot swap them interchangeably since Chilled directly impacts the cooldown of skills and DDT only is calculated when not a non-combat function is in place. Chilled is considered a condition and affects a cooldown which is considered an uninterrupted combat function.

Edited by gw2guruaccount, 24 February 2014 - 07:39 PM.


#19 MazingerZ

MazingerZ

    Golem Rider

  • Curse Premium
  • Curse Premium
  • 2274 posts
  • Profession:Guardian
  • Guild Tag:[CYRL]
  • Server:Tarnished Coast

Posted 24 February 2014 - 08:06 PM

View Postgw2guruaccount, on 24 February 2014 - 06:15 PM, said:

This isn't the case in GW2 though. Using the previous example the total damage expected out of the melee was 21,600, the actual damage dealt was 14,400. 33% is not small and it's based on in-game mechanics. I think it's time we graduated from WoW comparisons and mathematics because the games are different. GW2 doesn't use gear-checks as a means to lower your DPS and it doesn't use skill-checks as a means to take up damage slots. Using the same modeling for a totally different game is something like saying using basic math from a 1980s rogue will give you the same understanding of a multi-faceted complex system like GW1.

Let me express it this way:

GW2 uses "time sinks" to effect your DPS with most management being reliant on distance and disruption.

WoW ( and it's clones ) use "stat sinks" to effect your DPS with most management being reliant on skill choices and counters as well as gear choices and stats.

Using them interchangeably will produce disastrous results.

At this point, as I've stated numerous times, you are then better off removing the entire statistical block from GW2 because if its a game that's focusing on skill, the stat blocks are just there to muddy the waters.  Turn into a DMC ARPG where hitting the monster successively 5 times kills it.  Link got the occasional sword and weapon upgrade, but his damage I don't think was ever counted in terms of numbers, just number of successive strikes.

Based on everyone else's constant babbling about skill, 'Zerker and Ascended, the only reason the statistical blocks are in the game at this point are to feed the treadmill.

Statistical blocks are a hold over from P&P days when you had to roll dice, and held over up through most MMOs where you still have to accrue stats to overcome passive barriers to kill the target as well as a gating mechanic to accurately demonstrate the difference in power from boss to boss (IE: like when WoWInsider used to have two bosses wail on one another on those private back-ended WoW servers, IIRC).

In GW2 you have a level system that is clearly there as a gating mechanic, but no statistical system designed to actually indicate a difference in power between bosses (the downranking system wouldn't exist otherwise)... Basically, the downranking system was a way of saying: "We want you to grind to level 80, but we think its silly that the world isn't posing a constant challenge no matter your level."

Same thing with stats.  I can believe to a certain point their intent was to provide a certain level of customization to characters, but given the overall shallowness of the systems and the Reign of 'Zerker, they'd be better off trying not to balance around stats (if it wasn't feeding their gear treadmill) and instead focus on mechanics and skill challenges, instead of making massive health pools 'the challenge.'

Edited by MazingerZ, 24 February 2014 - 08:06 PM.

It's okay to enjoy crap if you're willing to admit it's crap.
Every patch is like ArenaNet walking out onto the stage of the International Don't Kitten Up Championship, and then proceeding to shiv itself in the stomach 30 times while screaming "IT'S FOR YOUR OWN GOOD! IT'S FOR YOUR OWN GOOD!"

#20 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 24 February 2014 - 08:29 PM

View PostMazingerZ, on 24 February 2014 - 08:06 PM, said:

Points
Stats in GW2 play a supplemental role instead of a primary role which is where it deviates from older systems; one way to think of it is that instead of having a single management system it has a multi-faceted system based entirely on skill weight and personal ability. This key difference is why skill weight and personal capability are so much more important than statistical distribution and thus why equipment plays a lot less of a role in player culpability. The second portion I wish to talk about is the statement of indication power differences; that actually isn't true, when you level down you don't lose the tier separating ( meaning that 3 stat armor stays 3 stat armor even against a level 2 boss ) for armor or weapons and you also don't lose the right to use excessively leveled sigils, runes, and other perks such as your traits. Obviously these make a huge difference in power and it explains why some classes can AFK lower level bosses at 80 but never could otherwise.

GW2 is not the epitome of skill-based gaming but it does come a lot closer to that level than other popular and older MMOs. Player choice in the past was based less on the actions of the player and more on the static components ( builds ) of the player while in this game you really could go and kill some boss in all whites if you really, really wanted to suffer that kind of punishment.

Circling around DDT is an extension of this change. You wouldn't use anything like this in WoW or if you did it would be awkward and not consistently useful.

Edited by gw2guruaccount, 24 February 2014 - 08:29 PM.


#21 raspberry jam

raspberry jam

    Vigil Crusader

  • Members
  • 4791 posts

Posted 25 February 2014 - 07:27 AM

View Postgw2guruaccount, on 24 February 2014 - 07:26 PM, said:

That's not how it works.

Let's say you're frozen in a block of ice by Jormag for 8s. Does it matter if your skill has a 2s cooldown if you're frozen in a block of ice? No. DDT does not directly correlate to a cooldown model. DDT is a part of the DPS model and it is not a replacement for the DPS mode.

All of my models for now have used the same Cooldown ( 1/s ) just for ease of digestion but you can easily swap and interchange all the numbers you want for comparison. So a bow that fires at a rate of .5s and a sword that swings at a rate of 1s have nothing to do with DDT at all. The fact that our teleporting enemy doesn't get out of range of the bow is what makes the DDT zero, cooldown is not effected, and the fact that it gets out of range of melee is what increases the DDT to 2s, it has nothing to do with how often the sword swings, ad infinitum.
Hmm. Perhaps you do not understand. The fact that being frozen bumps your effective cooldown to 8 seconds even though the skill itself has a cooldown of 2 seconds, and that extra-character effects (such as teleporting enemies or being frozen in a block of ice) will have an effect on the actual DPS, is my point.

It has to do with how often the sword swings. In a static comparison, the sword that swings once per second might outdamage the bow, but with the increased time between actual swings, the bow might outdamage the sword. So far, both methods yield the same result. However imagine that the sword can use some special skill once per three seconds. Having a time of two seconds between each actual swing would mean that every other swing would use the skill, that is, a four second interval between each autoattack and between each use of that special skill. The question is whether that is enough to outdamage the bow? Let's assume that it is not. Using a granularity sum, you could deduce whether intentionally waiting for the skill to recharge and make one swing every three seconds (if possible), using the skill in every swing, would outdamage the bow or not. If the damage difference between autoattack and the skill is large enough, then swinging every three seconds, if possible, will do more damage than swinging every two seconds.

Granularity sums is a complete effective-DPS model. Applying a multiplier is not.

#22 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 25 February 2014 - 05:57 PM

View Postraspberry jam, on 25 February 2014 - 07:27 AM, said:

Hmm. Perhaps you do not understand. The fact that being frozen bumps your effective cooldown to 8 seconds even though the skill itself has a cooldown of 2 seconds, and that extra-character effects (such as teleporting enemies or being frozen in a block of ice) will have an effect on the actual DPS, is my point.

It has to do with how often the sword swings. In a static comparison, the sword that swings once per second might outdamage the bow, but with the increased time between actual swings, the bow might outdamage the sword. So far, both methods yield the same result. However imagine that the sword can use some special skill once per three seconds. Having a time of two seconds between each actual swing would mean that every other swing would use the skill, that is, a four second interval between each autoattack and between each use of that special skill. The question is whether that is enough to outdamage the bow? Let's assume that it is not. Using a granularity sum, you could deduce whether intentionally waiting for the skill to recharge and make one swing every three seconds (if possible), using the skill in every swing, would outdamage the bow or not. If the damage difference between autoattack and the skill is large enough, then swinging every three seconds, if possible, will do more damage than swinging every two seconds.

Granularity sums is a complete effective-DPS model. Applying a multiplier is not.

Granularity sums can only generate data based on a model with consistent and relative effects. Inconsistencies make for weaker system analysis or finer and finer analysis which is basically what DDT does without overhead. Variable rates cause extreme problems for those models because so far our models are only based on singular component ( enemy ) behavior not enemy and player behavior. DDT covers both seamlessly.

Your two expressions for instance don't work in relation to proper modeling; for instance a "de facto cooldown" of 8s only works if it's always 8s, or if we're specific to that instance and don't continuously track the behavior, that you're frozen but let's say it's possible you can be frozen for anywhere between 1s and 8s; unless you break down the granularity to an extreme point it is nearly impossible to calculate without some form of complex function that takes into account probability modeling in relation to many, many factors. This is because when you combine the two elements of cooldown and downtime you must account for and differentiate between actual cooldown rates and the downtime itself. For instance if you have a skill that has a recharge of 30s but you are frozen for 8s of it and chilled afterwards for 8s the downtime itself is still 8s, it does not incorporate the recharge rate ( because you can use other skills in that that ) nor does it include frozen ( which only effects recharging skills and therefore doesn't represent real player downtime, just the downtime of the skill, which in turn may not matter in relation to rotation anyway ) however using a cooldown model the actual rate of effect is ( presuming you used the skill and were immediately frozen after ) 14.457 / 24s for proper representation without genuine incorporation of all skills; the more complex the rotation the more calculations you must make however with the DDT model your actual downtime may not even increase by a meaningful amount since if you can manage to maintain and sustain through the chill duration your rotation whether it be through an active weapon swap or some other function you may be in a lossless scenario for actual uptime beyond the actual duration of being frozen.

It's not just a multiplier, it's a real-time tracking device that can be converted into a multiplier or divisor without all the overhead that your sums require for total completeness. By ignoring that overhead you get a simple realistic uptime ratio without the headache of breaking down many elements. Furthermore it is used in real-time both ways; so it expresses not only skill uses ( which is all we used before ) but player behavior, enemy behavior, and environmental awareness to award stronger calculations without excessive work.

I suppose I should say that your method would work but it isn't worth as much due to effort since you'd be breaking down in just the scenario above distance to target, cooldown on 5 skills, duration of condition ( presuming it's greater than 0 either way i.e. +/-10% ) player behavior, complete condition analysis ( does the player have any other elements that would detract or add to the dilemma? ), and at least one analysis of present environmental conditions.

It's quite a bit easier to just add up your downtime and make a summary report for time accountability.

#23 Konzacelt

Konzacelt

    Sylvari Specialist

  • Members
  • 556 posts
  • Location:a prairie under a clear, blue sky
  • Guild Tag:[PD]
  • Server:Sanctum of Rall

Posted 25 February 2014 - 06:20 PM

View Postgw2guruaccount, on 23 February 2014 - 05:17 PM, said:

wall of text

Why do you keep posting think-tank threads?  ANet is already plying the playerbase for ideas on their official forums, it's redundant to do the same here.  It's also a little sad.

#24 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 25 February 2014 - 06:52 PM

View PostKonzacelt, on 25 February 2014 - 06:20 PM, said:

Why do you keep posting think-tank threads?  ANet is already plying the playerbase for ideas on their official forums, it's redundant to do the same here.  It's also a little sad.
What does a question about math have to do with Anet?

Oh, and I also didn't post this here, it got moved here.

#25 raspberry jam

raspberry jam

    Vigil Crusader

  • Members
  • 4791 posts

Posted 25 February 2014 - 08:20 PM

View Postgw2guruaccount, on 25 February 2014 - 05:57 PM, said:

Granularity sums can only generate data based on a model with consistent and relative effects. Inconsistencies make for weaker system analysis or finer and finer analysis which is basically what DDT does without overhead. Variable rates cause extreme problems for those models because so far our models are only based on singular component ( enemy ) behavior not enemy and player behavior. DDT covers both seamlessly.

Your two expressions for instance don't work in relation to proper modeling; for instance a "de facto cooldown" of 8s only works if it's always 8s, or if we're specific to that instance and don't continuously track the behavior, that you're frozen but let's say it's possible you can be frozen for anywhere between 1s and 8s; unless you break down the granularity to an extreme point it is nearly impossible to calculate without some form of complex function that takes into account probability modeling in relation to many, many factors. This is because when you combine the two elements of cooldown and downtime you must account for and differentiate between actual cooldown rates and the downtime itself. For instance if you have a skill that has a recharge of 30s but you are frozen for 8s of it and chilled afterwards for 8s the downtime itself is still 8s, it does not incorporate the recharge rate ( because you can use other skills in that that ) nor does it include frozen ( which only effects recharging skills and therefore doesn't represent real player downtime, just the downtime of the skill, which in turn may not matter in relation to rotation anyway ) however using a cooldown model the actual rate of effect is ( presuming you used the skill and were immediately frozen after ) 14.457 / 24s for proper representation without genuine incorporation of all skills; the more complex the rotation the more calculations you must make however with the DDT model your actual downtime may not even increase by a meaningful amount since if you can manage to maintain and sustain through the chill duration your rotation whether it be through an active weapon swap or some other function you may be in a lossless scenario for actual uptime beyond the actual duration of being frozen.

It's not just a multiplier, it's a real-time tracking device that can be converted into a multiplier or divisor without all the overhead that your sums require for total completeness. By ignoring that overhead you get a simple realistic uptime ratio without the headache of breaking down many elements. Furthermore it is used in real-time both ways; so it expresses not only skill uses ( which is all we used before ) but player behavior, enemy behavior, and environmental awareness to award stronger calculations without excessive work.

I suppose I should say that your method would work but it isn't worth as much due to effort since you'd be breaking down in just the scenario above distance to target, cooldown on 5 skills, duration of condition ( presuming it's greater than 0 either way i.e. +/-10% ) player behavior, complete condition analysis ( does the player have any other elements that would detract or add to the dilemma? ), and at least one analysis of present environmental conditions.

It's quite a bit easier to just add up your downtime and make a summary report for time accountability.
But making a straight sum of the downtime does not take skill cooldowns into account. And as I have shown, the granularity of cooldowns can have a huge effect.

I want to help you. Set up a complete, explicit example of your calculations and I'll show you where your method breaks down.

#26 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 25 February 2014 - 08:30 PM

View Postraspberry jam, on 25 February 2014 - 08:20 PM, said:

But making a straight sum of the downtime does not take skill cooldowns into account. And as I have shown, the granularity of cooldowns can have a huge effect.

I want to help you. Set up a complete, explicit example of your calculations and I'll show you where your method breaks down.

There's a reason I don't use skill cooldown and it is referenced in the summation above: Rotations. The cooldown on an individual skill itself doesn't matter if the rotation covers the entire duration or more. For instance let us use a rotation of 8s for all 5 skills, if skill A is used and takes 2 seconds to use skill B will begin after skill A is at 6s, if skill B takes the same 2s then skill A is at 4 and skill B is at 6, if C takes 5 seconds then A is recharged and B is recharged while C is starting it's cooldown, if D only takes 1s then C has 7 left, everything else recharged, etc.

The reason I don't incorporate this is because this is generally how it runs real-time. So measuring actual "up-time" makes more sense than trying to incorporate cooldowns across all 5 skills, esp. in situations where the cooldown is 1s or lower ( auto-attack for instance ) and soaks up the lapses between cooldown on other skills. Since you can still maintain a 100% uptime there's no value to trying to convert cooldown effects due to the lack of a global cooldown in the game.

A real example:

Engineer with Grenade kit throws a Shrapnel Bomb (2) which has a 5s cooldown. Normal "autoattack" level grenades (1) take less than 1s to toss therefore 4 are thrown between Shrapnel Bombs; there is no DDT because there is a constant wave of attacks. Trying to integrate a cooldown reliant system will produce fractions of a second of inactivity based on cooldown since the grenade (1) has a cooldown of approximately .85~.92s.

#27 raspberry jam

raspberry jam

    Vigil Crusader

  • Members
  • 4791 posts

Posted 26 February 2014 - 10:01 AM

View Postgw2guruaccount, on 25 February 2014 - 08:30 PM, said:

There's a reason I don't use skill cooldown and it is referenced in the summation above: Rotations. The cooldown on an individual skill itself doesn't matter if the rotation covers the entire duration or more. For instance let us use a rotation of 8s for all 5 skills, if skill A is used and takes 2 seconds to use skill B will begin after skill A is at 6s, if skill B takes the same 2s then skill A is at 4 and skill B is at 6, if C takes 5 seconds then A is recharged and B is recharged while C is starting it's cooldown, if D only takes 1s then C has 7 left, everything else recharged, etc.

The reason I don't incorporate this is because this is generally how it runs real-time. So measuring actual "up-time" makes more sense than trying to incorporate cooldowns across all 5 skills, esp. in situations where the cooldown is 1s or lower ( auto-attack for instance ) and soaks up the lapses between cooldown on other skills. Since you can still maintain a 100% uptime there's no value to trying to convert cooldown effects due to the lack of a global cooldown in the game.

A real example:

Engineer with Grenade kit throws a Shrapnel Bomb (2) which has a 5s cooldown. Normal "autoattack" level grenades (1) take less than 1s to toss therefore 4 are thrown between Shrapnel Bombs; there is no DDT because there is a constant wave of attacks. Trying to integrate a cooldown reliant system will produce fractions of a second of inactivity based on cooldown since the grenade (1) has a cooldown of approximately .85~.92s.
But most skills are more efficient than autoattack and it's thus better to save them until they can take effect.

Assume that you are effectively disabled (not actually disabled, just unable to hit the enemy) for let's say 7 seconds. During that time you will throw 1-2 shrapnel bombs if you mindlessly follow the rotation. These are essentially wasted, in fact not only that, but it will be about 3 more seconds until you can throw another shrapnel bomb.
Now assume that you throw a bomb the moment before you are disabled, at T=0. Then at T=5 you will throw another (wasted) one because you are following rotation. Then at T=7 you are able to hit the enemy but you are still not able to throw a shrapnel bomb because it's on cooldown. At T=10 you throw one more. At this point, the effective cooldown of your shrapnel bomb was 10 seconds, not 5.
To avoid that situation, a player would wait until T=7 to throw the bomb, thus by waiting he would actually reduce effective cooldown from 10 to 7 seconds.
(The disabling can take any form, for example an enemy becoming invulnerable, teleporting away, or the player killing an enemy and having to rush to/wait for the next one.)

Anyway, we might not be talking about the same thing. Granularity sums is a good way to calculate the optimal DPS in a situtation where you are occasionally prevented from damage output before you reach said situation or even start the game. Your system is just fine for calculating actual DPS post facto. It's just that at that point, the result lacks relevance.

#28 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 26 February 2014 - 01:23 PM

View Postraspberry jam, on 26 February 2014 - 10:01 AM, said:

But most skills are more efficient than autoattack and it's thus better to save them until they can take effect.

Assume that you are effectively disabled (not actually disabled, just unable to hit the enemy) for let's say 7 seconds. During that time you will throw 1-2 shrapnel bombs if you mindlessly follow the rotation. These are essentially wasted, in fact not only that, but it will be about 3 more seconds until you can throw another shrapnel bomb.
Now assume that you throw a bomb the moment before you are disabled, at T=0. Then at T=5 you will throw another (wasted) one because you are following rotation. Then at T=7 you are able to hit the enemy but you are still not able to throw a shrapnel bomb because it's on cooldown. At T=10 you throw one more. At this point, the effective cooldown of your shrapnel bomb was 10 seconds, not 5.
To avoid that situation, a player would wait until T=7 to throw the bomb, thus by waiting he would actually reduce effective cooldown from 10 to 7 seconds.
(The disabling can take any form, for example an enemy becoming invulnerable, teleporting away, or the player killing an enemy and having to rush to/wait for the next one.)

Anyway, we might not be talking about the same thing. Granularity sums is a good way to calculate the optimal DPS in a situtation where you are occasionally prevented from damage output before you reach said situation or even start the game. Your system is just fine for calculating actual DPS post facto. It's just that at that point, the result lacks relevance.


We're not talking about the same thing.

DDT is a time management tool.

Granularity ( as you present it ) in relation to cooldown has nothing to do with DDT.

Effective cooldown does not work since inaction does not incorporate inability. If a melee character cannot strike due to being too far away this effected both Granularity and DDT. If a melee character however holds his strikes until an opportune time this does not effect Granularity because there is no effective cooldown increase but it does effect DDT because of the real-time effect of "waiting".

Using the grenadier example DDT increases based on the downtime so the outcome of your Granularity model and DDT are effectively the same where at T7 the character is unbound and should act in whatever manner present, in Granularity as you presented it is presumed that there is a continuous volley but DDT does not presume player action it simply measures the time between effectively damaging actions therefore during that same 7 seconds if a bleed/condition were applied that lasted 10s and did a consistent 1s damage tick DDT would not increase but Granularity would by some percentage.

The idea behind DDT is that you can actually plan ahead so that you are constantly and perhaps consistently dealing damage while the idea behind Granularity is that you expect and react to specific scenarios; since DDT can be measured both ways, prior and after, while Granularity only makes sense to measure prior DDT can be used to monitor player behaviors and improve uptime as well as manage resources better while Granularity can only present composites of scenarios and functionally just works like design calculus.

The two tools have completely different uses.

#29 master21

master21

    Vanguard Scout

  • Members
  • 447 posts

Posted 26 February 2014 - 01:36 PM

View Postgw2guruaccount, on 24 February 2014 - 04:24 PM, said:

This would definitely make all Theorycrafting easier. Or rather sort of moot.

without dps metter every theoretical attempt for dps is just useless really.
All we know now is:
skill coef
formula for damage
and limited information cast time.

Even the easiest thing to calculate, damage done in xx seconds could be calculated "almost" because info about cast time is quite meh.

DPS meter is just must have. Anet probably has some stupid philosophy that "it will break community" because implementation of this is probably 1 hour for half brain developer.

Maybe if they stopped with "anti addons" politic it could be possible to make some sort of dps meter. Combat log is useless (why it's even there I don't know) so it should be flying "number" analyze meter. Game client get those numbers to show them, so all what we need really here is to sum them up and show somewhere.

#30 gw2guruaccount

gw2guruaccount

    Sylvari Specialist

  • Members
  • 892 posts

Posted 26 February 2014 - 01:39 PM

View Postmaster21, on 26 February 2014 - 01:36 PM, said:

without dps metter every theoretical attempt for dps is just useless really.
All we know now is:
skill coef
formula for damage
and limited information cast time.

Even the easiest thing to calculate, damage done in xx seconds could be calculated "almost" because info about cast time is quite meh.

DPS meter is just must have. Anet probably has some stupid philosophy that "it will break community" because implementation of this is probably 1 hour for half brain developer.

Maybe if they stopped with "anti addons" politic it could be possible to make some sort of dps meter. Combat log is useless (why it's even there I don't know) so it should be flying "number" analyze meter. Game client get those numbers to show them, so all what we need really here is to sum them up and show somewhere.
I agree with you. Working against Anet's "Keep Players in the Dark" mantra is not easy but I would dare say it is a challenge that is worth undertaking. The more we understand the game the more we can play it effectively without feeling like total toolz for buying into popular and majorly unsupported mathematical claims.




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users