Path of Building: Offline Build Planner [v1.4.71]

Made a big post, but then realized that it is easier to show, than to explain. So i decided to code a simulation tool to illustrate complicated things. Should have time for it within the next few days.

Hey there,
i just wanted to ask what to do to import ur acc / char if u have a name like mine "Aîphaton" cause he sees the i+^ as a ? and dosnt recognize it and if i insert my poessid or how its called i type in the code wich is at the "contend:" section of the cookie(chrome). I tried it with a ID from Firefox too and it dosnt work either.
Appreciate Help

Hey Openarl,

Thanks so much for this wonderful tool, first of all.

There seems to be an error in handling Static Strike's lightning conversion, though: in testing a build with Brain Rattler and Static Strike (where 100% of physical damage is already converted to Lightning Damage), the Physical to Lightning Support gem displayed a DPS increase as if some percentage of damage was not yet converted.

When I switched Static Strike for Lightning Strike, leaving all other variables the same, the damage increase displayed became what was expected, i.e. in line with the % Physical Damage Gained as Lightning Damage.

Thanks a bunch,
poeticEnnui

"

tremerrr wrote:

Made a big post, but then realized that it is easier to show, than to explain. So i decided to code a simulation tool to illustrate complicated things. Should have time for it within the next few days.

Fair enough, although I suspect you're heading down a well-trodden path.

"

Aîphaton wrote:

Hey there,
i just wanted to ask what to do to import ur acc / char if u have a name like mine "Aîphaton" cause he sees the i+^ as a ? and dosnt recognize it and if i insert my poessid or how its called i type in the code wich is at the "contend:" section of the cookie(chrome). I tried it with a ID from Firefox too and it dosnt work either.
Appreciate Help

You have to replace the character with its URL encoding; so your account name becomes:
A%C3%AEphaton

"

poeticEnnui wrote:

There seems to be an error in handling Static Strike's lightning conversion, though: in testing a build with Brain Rattler and Static Strike (where 100% of physical damage is already converted to Lightning Damage), the Physical to Lightning Support gem displayed a DPS increase as if some percentage of damage was not yet converted.

When I switched Static Strike for Lightning Strike, leaving all other variables the same, the damage increase displayed became what was expected, i.e. in line with the % Physical Damage Gained as Lightning Damage.

Believe it or not, this is working as intended; if there is a bug here, it's in the game, not the program.

When conversions from one damage type exceed 100%, the conversion values are scaled down to 100%. When this scaling occurs, any conversions that are implicit to the skill take priority; in this case, Static Strike's 60% conversion. Physical to Lightning's conversion, being sourced from a support skill, is also counted as a "skill conversion", so when combined with Static Strike it gives 110% skill conversion of physical damage. In theory that conversion should be scaled down to 100%, just as non-skill conversions get scaled; however, currently this does not occur!

You can easily verify this in-game using Static Strike and a level 1 Physical to Lightning on a character with no increases to elemental damage; you'll see that you end up with 120% of your physical damage as elemental, even though you only have 10% added (from Physical to Lightning).

Incidentally, Molten Strike and Ice Shot have the same issue; Frost Blades however does not despite having 60% conversion, because it uses global conversion instead of skill conversion, which is another bug. I reported both issues many months ago, but neither have been fixed yet.

So. the work is almost done (i still need to double check all calculations) and the results so far are:

I am getting +-20 - 25% expected dps results in every simulation, so random is highly impactful even in ~30s fight (which will rarely happen in game so the random spreads even more)
The main things i was concerned of were attack speed and bleed duration. Changing these doesn't affect dps in PoB, but due to the nature of bleed mechanics these should be considered. Both of these widen the window in which you can apply and reapply bleeds to get minimum lowdmg bleed duration (esp in short fights). And simulations with higher bleed duration show more consistency.
Overall, i overestimated impact of these parameters on the final output, so current bleed dps meter is somewhat more accurate than i thought. Besides, looks like it is hard to take them into consideration via simple formula.

Damn, i completely forgot about my main concern - chance to bleed and chance to bleed on crit. If you are using kondo's pride (bleed on crit) or something similar, adding chance to bleed on normal attacks reduces bleed dps by a lot, which is quite odd. Equipping voidheart or speccing into gladiator's bleed passives lowers the tooltip.
Gonna adress this when i have time because this is where calculations will start to differ significantly.

All set now.. nvm

So, after improving many things i am ready to provide some numbers. All these testings are made in a 10k seconds simulation to reduce random factor to a minumum. Dps varies from sim to sim +-<4%


#1
First setup shows diffrence in ~16% with expected dps.



#2
Lets increase APS to show that it affects bleed dps and is ignored by a tooltip. Setup stays the same, only aps changes.



#3
Lets increase bleed duration to show that it affects bleed dps and is ignored by a tooltip.Setup stays the same as in #1, only bleed duration changes.



#4
I crafted a custom ring with "causes bleeding on hit" mode, which lowered my tooltip by ~22% for some reason. Which is against any logic and testing confirms i am not insane. Same setup as in #1 only chance to bleed changes.


Hope that helps.

"

tremerrr wrote:

So, after improving many things i am ready to provide some numbers. All these testings are made in a 10k seconds simulation to reduce random factor to a minumum. Dps varies from sim to sim +-<4%

None of this is news to me; this topic has come up quite a few times before, although usually with regards to ignite (which of course functions identically, sans Emberwake).

Firstly, I should clarify what the bleed/ignite DPS shown by the program actually represents: It is the average size of each bleed or ignite instance that would be applied to the enemy. This is affected not only by damage modifiers, but also your crit chance and chance to inflict on crit or non-crit; if your chance to inflict on non-crit goes up then the size of the average instance goes down (as seen in #4), but you'll be more likely to actually apply an instance.

Now, as you correctly demonstrated, this number isn't representative of your average bleed/ignite DPS. But, as I suspect you've realised now, calculating the actual average DPS wouldn't be straightforward. I'm not sure how the math would even work; it should be a function of the average concurrent instances, but beyond that I have no idea.

The main problem with doing that math, however, is that any amount of time spent moving, or not attacking, will affect your average DPS in ways that cannot be accounted for; so the result of that calculation may not be much more accurate than the current output, and perhaps even less accurate under some conditions.

So, with that in mind, is it really worth going to all the trouble of producing that result, when, as you demonstrated, the current calculation isn't usually that far off? In my opinion, no.

Of course, that doesn't mean that the current calculation mode is perfect. The behaviour seen in #4 is unintuitive, although it does hint at something important: that it isn't really correct to get more chance to bleed/ignite on non-crit if you're able to sustain crit bleeds/ignites. It also doesn't factor in the impact of attack speed, but that's hard to account for.

Despite those quirks, the current calculation still works fine for optimising builds; anything that improves the Total DPS and Bleed/Ignite DPS should also improve your actual DPS. Of course the relative value of different modifiers might be off by some small amount, but you're never going to get the right anyway. There are other benefits too: it is also the same method used to calculate poison and Emberwake damage, so it gets bonus points for consistency, and it typically understates your DPS, which is preferable to overstating (which a more complex method certainly would).

All in all, I'm quite satisfied with the current calculation mode. That doesn't mean I'd never be open to changing it, but I'd need an alternative that's clearly superior.

any plans on adding support for minions now when they released the minion numbers?