Jump to content

Persistence vs Rain of Godagh Field Damage Test


Recommended Posts

Hello folks. I'm someone who really enjoys playing rangers in this game, and I love bows. I'm an archer IRL and I'm quite happy extending my obsession with bent sticks into games. So, my ranger's main weapon is always a bow. When I last played through the game I used Rain of Godagh Field for the majority of it, as I found it to be the most effective bow. However I saw ottssffe's 'Bow Damage, a rough guide' post earlier in the year which found that Persistence had the highest damage, followed by Cloudpiercer, with Rain lagging behind and it made me wonder.

 

So, I decided to test it myself. I made a tank character with as much health as I could manage and with 18 DR and 104 Deflection, as this seemed like a reasonable amount for an example of a relatively tough PotD enemy. I wanted to run my test under the same conditions that my ranger would normally play, so I had Kana singing, alternating between Swift Handed Ila and Aefyllath Ues Mith Fyr (which causes some gaps in Ila but I like that burning lash). My character was wearing Gauntlets of Swift Action and both weapons were Durganised, Legendary (this was more to reach their full potential than to be an example of normal play) and had a corrosive lash. My ranger had Twinned Arrows active throughout. I started a fight in The Black Hound and my other characters blocked access to the back room while my ranger shot the tank. I ran each test for 30 seconds, ten times, with Durance healing the tank throughout. I then averaged the health damage done to the tank in each run. Persistence's average was 927.6, Rain's was 1019.7. Obviously 300 seconds of shooting each isn't long enough to be statistically significant or to guarantee that the variations in attack rolls are averaged out, but I think it's long enough to give a reasonable idea of each weapon's damage. So under the conditions described above, Rain seems to slightly out damage Persistence. (I also did a separate test for Persistence under the same conditions with Penetrating Shot active, however its average was only 917.2. This will have been highly influenced by the randomness of attack rolls but it seems to me that under these conditions Penetrating Shot isn't worth the talent point).

 

Now, while Rain had a slight edge, for rangers there's also their animal companion to consider. If Predator's Sense has been selected then Persistence's DoT effect can offer a substantial boost to the companion's damage. So I repeated the test again, this time upping the damage reduction of the tank to 28, just so I could keep him alive for long enough! With both ranger and animal companion (in this case a wolf) attacking, the average for Persistence was 1178.0 and for Rain it was 982.2. So, for rangers, who have selected Predator's Sense, Persistence appears the better pick. For other classes, Rain may be preferable. Of course, Persistence is available a lot earlier than Rain as well. And this only tests single target damage. For general use against mobs Stormcaller is probably still the best weapon for a ranger. Just in case anyone was wondering, I also ran the original test with Sabra Marie to see if its crit damage boost would help it compete. However, its average was only 825.5. Of course, it also has the bonus of being able to confuse enemies, which is excellent.

 

Anyway, I hope this was of interest for some of you :)

Link to comment
Share on other sites

Actually, ten runs of 30 seconds each might well be enough to be statistically significant, what were your standard deviations? 

 

Looks nice, but I would have concerns about the generalizability. Different weapons respond differently to different parameters, such as deflection and in this case particularly DR. After all, Hunting Bows are more susceptible to high DR than War Bows, so which will win out will depend very much on what you're trying to hit. The effectiveness of Penetrating Shot is also dependent on target's DR relative to your weapon damage (base + damage bonuses); perhaps counter-intuitively, it is no longer effective if DR gets too high.

 

For any future testing you might do, I can also strongly recommend using the console. Just grab one of your regular NPCs and set Constitution to 1000 and do Healparty, and you have yourself a handy target. Similarly, you can much more easily vary stats on the attacker as needed (keep in mind that the console sets the base values, before modifiers; also, I've never gotten it to work for hirelings, so you'd need regular NPCs/the protagonist for this), add experience, talents and abilities, items, etc. 

Link to comment
Share on other sites

Kaylon, my might was 21. My dexterity was 27.

 

Loren, thanks for the tips, I'll try using those console commands the next time I test something. My standard deviation was 94.09. I did consider doing a range of tests against various DRs, maybe I'll do that next against 0, 5, 10, 15, 20, 25 and 30. I'm just not sure when I'll get the time!

Link to comment
Share on other sites

It increases quadratically with Might, surely. Though it'll make a reasonable difference anyway. Especially against high DR targets, since the Wounding is not reduced by it. 

 

With that standard deviation you get a p-value of about 0.042 by the way (simple two-sided t-test), so it is significantly different. A bit more data is always nice of course, but this seems reliable enough. Though in this case it would ultimately be better to simply compute the damage output. All the variables are known (or knowable) as far as I can see, so the rest is just basic math. Also makes it easier to get to stuff like at what DR Penetrating Shot should be turned on and turned off again. Speed bonuses were 30% + 20% Ila + 20% for Rain, right? What were your armour penalty and INT?

Link to comment
Share on other sites

Fair play Kaylon, I'll try Persistence again with higher might when I'm next testing. Out of interest, since I know that you're experienced with doing PotD solos with rangers, if you were making a solo run ranger that was going to use both Persistence and Stormcaller, would you go for the Living Lands background (for more wounding damage with Persistence) or Deadfire Archipelago (for higher dex and more frequent Stormcaller shots)? 

 

Loren, the armour penalty was 0 (enchanted cloth) and Int was 8. You're right about the speed bonuses. And if you can work out the maths rather than doing the testing then please go ahead, that sounds much more efficient. My mathematical knowledge is basic, so I don't know how to.

Edited by Arranvin
Link to comment
Share on other sites

It would be doable, sure. Though I would need a couple more values. Lets see... in terms of shared/identical stats for both weapons we have:

- -50% recovery (Ila + Swift Action + Durganized)

- +33% (might), +55% (Legendary) damage

- +15 (Legendary), -10 (Twinned Arrows) accuracy

- +20% hit to crit, +30% crit damage (Durganized)

- +25% Fire (Aesfyllath), +25% Corrode (Lash) damage

- two hits per attack (Twinned Arrows)

 

For simplicity I'm assuming both chants are continuously active.

 

Weapon-specific we then also have:

- Persistance: +20% graze to hit, 25% damage as raw DOT

- Rain: -20% recovery

 

That should be the complete list of what I have (and need) from the above, I believe. So what is still missing? Primarily just Dexterity and Accuracy I think (either with or without Twinned Arrows and Legendary, so long as I know which; was it the same for both, or did you have any Weapon Focus?); ah, and Burn/Corrode DR, if they were different from the 18 DR for piercing. And were there any other relevant effects/abilities/whatever active on the ranger at the time? 

Edited by Loren Tyr
Link to comment
Share on other sites

Another question... You said the tank had 18DR - but was the general DR or the pierce DR? Because what matters is the pierce DR...

However it's possible that speedbuff + fire lash vs very high DR give an edge to the Rain of Godagh Field.

 

PS. Also since the ranger has very few means to increase his ranged damage (besides might/weapon enchantments) there's a reason you don't use the Stalker's Torc or the cloak with 10% ranged dmg? Also you should consider Apprentice's Sneak Attack too...

Edited by Kaylon
Link to comment
Share on other sites

I do actually use the Cloak of the Frozen Hunt. I forgot to include that in the run down. And the build does have Apprentice's Sneak attack, but in this test I wasn't applying any of the conditions that would allow it to qualify (and started shooting at the tank after more than two seconds had elapsed, since I attacked Tenfrith, let the recovery period end, and had Durance cast Consecrated Ground then let his recovery period end) so it wouldn't have made any difference. Thanks for the tips though :)

 

The 18DR was the Pierce DR.

Link to comment
Share on other sites

I've set up some scripts to calculate the expected damage output given various parameters; have been planning to do this anyway, so this gave me a nice excuse to do so.

 

However, I'm finding a discrepancy between my calculations and your results. The expected damage for 30 seconds according to the parameters above should be 1094.6 for Persistence and 1059.5 for Rain, still suggesting a small advantage for Persistence. The latter value is consistent with the mean value you found, but such a big gap for Persistence between expected and observed value (I didn't compute the variance yet, but going by your observed SD) would be very unlikely to be due to stochastic noise.

 

I have first verified the correctness of my computations with some tests of my own. I didn't quite replicate the above, I omitted the Chanter since it is more convenient to test outside of combat. I also had slightly different DR values: 17.5 Pierce, 14 Corrode (Fine Plate). I modified Sagani and Pallegina (the target) to get the same Might, Dex, accuracy and deflection as above, gave Sagani the Gauntlets and Cloak and Durganized Legendary bows with Corrode Lash. Expected damage under these settings is 740.9 for Persistence and 641.9 for Rain over 30 seconds. I tested each for 10 minutes, which gave me mean damage per 30 seconds of 729.9 and 646.5 respectively; very close to the expected values, and thus pretty much confirming the correctness of my calculations. Adding the Burning Lash and speed-up from Ila are simple changes to the script and therefore are unlikely to contain errors (though I did recheck, obviously). The small gap in Aefyllath coverage I omitted from the calculations also can't be the explanation, since it would affect both bows; if anything, it affects Rain more since the Lash damage accounts for a greater proportion of the total expected damage (22.6%, vs 17.0% for Persistence).

 

Anyway, this suggests that either we missed some relevant parameter or effect that affected your simulations but that we didn't think of here; or there was some sort of glitch. I actually did get one of those during my own test, with some accuracy effects not applying to Rain when they should (though the drop in DPS wasn't *that* big). In any case, the computations are shown to be accurate, so I'll wrap them up in some more convenient functions and whip up some more results (effect of DR, Penetrating Shot, etc.). 

  • Like 1
Link to comment
Share on other sites

Made a plot with some results, just to get an idea. This is with the same settings as used above in my own test (ie. no Chanting), varying DR. For simplicity, I set DR to the same value for all damage types. I computed Persistence and Rain both with and without Penetrating Shot active. I also included the damage output for Persistence with the Wounding effect removed, to show how much of a difference that makes.

post-163298-0-06322400-1470705237_thumb.png

Edited by Loren Tyr
  • Like 2
Link to comment
Share on other sites

With 118acc vs 104def you should have on average 1%miss, 35%grazes, 50%hits and 14%crits with ROGF and 1%misses, 28%grazes, 57%hits and 14%crits with Persistence.

 

The average damage per hit would be:

- for Persistence 12.5*(.28*1.48+.57*1.98+.14*2.78)=22.4pierce, 5.6corrode, 5.6burn, 7.5raw and after DR 7pierce, 1.1corrode, 1.1burn and 7.5raw (16.7 total)

- for RoGF 16.5*(.35*1.48+.50*1.98+.14*2.78)=31.3pierce, 7.9corrode, 7.9burn and after DR 13.5pierce, 1.6corrode, 1.6burn (16.7 total)

 

The delay between shots would be:

- for Persistence (1+1.66*(1-.38))/1.51=1.34s

- for RoGF (1.5+2.5*(1-.66))/1.51=1.56s

 

Over 30s that would make 747dmg for Persistence vs 642 for RoGF with Twinned Arrows. The advantage should be clearly on Persistence side, however my numbers are much lower than the ones from the test - maybe you used Fast Mode...

Link to comment
Share on other sites

With 118acc vs 104def you should have on average 1%miss, 35%grazes, 50%hits and 14%crits with ROGF and 1%misses, 28%grazes, 57%hits and 14%crits with Persistence.

 

The average damage per hit would be:

- for Persistence 12.5*(.28*1.48+.57*1.98+.14*2.78)=22.4pierce, 5.6corrode, 5.6burn, 7.5raw and after DR 7pierce, 1.1corrode, 1.1burn and 7.5raw (16.7 total)

- for RoGF 16.5*(.35*1.48+.50*1.98+.14*2.78)=31.3pierce, 7.9corrode, 7.9burn and after DR 13.5pierce, 1.6corrode, 1.6burn (16.7 total)

 

The delay between shots would be:

- for Persistence (1+1.66*(1-.38))/1.51=1.34s

- for RoGF (1.5+2.5*(1-.66))/1.51=1.56s

 

Over 30s that would make 747dmg for Persistence vs 642 for RoGF with Twinned Arrows. The advantage should be clearly on Persistence side, however my numbers are much lower than the ones from the test - maybe you used Fast Mode...

 

For the set-up Arranvin used the numbers are a bit different actually. Damage per shot after DR comes to 8.4 Pierce + 8.4 Raw + 4.3 Corrode + 4.3 Burn = 25.3 total for Persistance, and 14.9 Pierce + 6.2 Burn + 6.2 Corrode = 27.3 total damage for Rain. At the very least you seem to be missing the +20% hit-to-crit conversion, but I'm also not sure whether you are accounting for the 20% minimum damage after DR reduction. I have attack durations at 1.39 and 1.55 seconds respectively, and 60 * 25.3 / 1.39 = 1094.6 and 60 * 27.3 / 1.55 = 1059.5 damage output for 30 seconds I reported above as well (I'm using the underlying unrounded values for computation, obviously). My computations are not for fast mode, just normal mode; and Arranvin couldn't have used Fast mode either, since he tested during combat.

 

I computed the expected damage for my own tests in the same way as well, which as noted matched the observed results very well and suggests that my computations are sound. 

Link to comment
Share on other sites

Top quality. From your graph appears that Pen shots are worth only in a small gap of dr, and also in that case seems to add very little dmg... Would you reccomand to skip the talent entirely?

 

Well, it does depend on various factors as well, so it is difficult to draw a general conclusion. Just for further illustration, I have expanded the figure slightly, adding a line for the Persistence without Wounding with Penetrating Shot. You can see that there the gap is much more pronounced. By a rough estimate, for that bow in the best case it is equivalent to something like a +20% damage bonus, maybe +25% (again, also dependent on various factors what the net effect of those is, but just as a general indication). This, I think you'll agree, is certainly worth it. The main reason why it's not doing nearly so much for Persistence is that the Wounding is not affected by DR, so that part is only hindered by the -20% speed penalty. 

 

For further illustration, see also the second attached plot where I stripped away the Durgan and Gauntlet +15% attack speed bonuses. Reason for this is that if you're already fast, the -20% penalty has more impact than if you're slower (due to the additive nature of how those combine). Recovery duration is computed as C/1.2 * (attack duration), where C = 1 - 2*S and S is the attack speed component (so S=0.5 is 0 recovery). When attacking with a ranged weapon, S essentially starts out as -0.5. Bonuses increase S, penalties decrease it, so ranged weapon + leather armour would get you S = -0.5 - 0.3 = -0.8. 

 

For the above scenarios with Persistance, we have a 15% attack speed bonus from Durganized and from Gauntlets. These are, as it happens, multiplicative with each other, so combine into 1.15*1.15 = 1.3225, so a +32.25% attack speed bonus. This results in S = -0.5 + 0.3225 = -0.1775 and C = 1.355. With the Penetrating Shot penalty, S = -0.1775 - 0.2 = 0.3775 and becomes C = 1.755; this is about a 30% increase in recovery duration. 

But now look at the situation where that initial +32.25% bonus isn't there (or almost equivalently, it is there but you are also wearing Leather armour). Without Penetrating Shot, S = -0.5 and C = 2. With Penetrating Shot, S = -0.5 - 0.2 = -0.7 and C = 2.4; this is now only a 20% increase in recovery duration, and therefore the negative impact on DPS is smaller as well.

 

Anyway, to amble back in the direction of your question: I certainly wouldn't recommend not taking Penetrating Shot (or Vulnerable Attack, same kind of deal) in a general sense. There are definitely plenty of scenarios where it can increase damage quite a bit (and more than other talents would). But it is quite conditional on other factors though, so it won't always be useful. Going by the data in this thread, in combination with Persistence in particular it probably isn't going to help all that much (though it does a little). And also good to keep in mind is that at very high DR it will eventually start becoming counterproductive (because at that point you'll often hit the MIN damage cut-off even with the -5 DR so the DR reduction doesn't help (much) anymore, but the speed penalty does still slow you down).That's something to keep an eye on, so you can turn it off against those enemies.

post-163298-0-17991500-1470715139_thumb.png

post-163298-0-88398200-1470715394_thumb.png

Link to comment
Share on other sites

Thanks very much for all your hard work Loren, that's really helpful. I can't explain the discrepancy between your results and mine. I'll take another look at my save to see if there are any other factors that I may have forgotten. I may even run the test again when I get the time. Not that I think it's necessary, I'm sure your results will be correct, I'm just curious about what has caused the discrepancy.

 

If you wouldn't mind, I'd be really interested to see a visual representation of how increased might would affect the damage for both bows versus increased dex.

Link to comment
Share on other sites

Very nice stuff. It would be very cool to use your script and knowledge and put such a neat graphic onto every wiki page of a weapon. ;)

 

I could even make an online tool which would take your char's stats and tells you what weapon you should use - but I don't know if it's worth the effort and if it can reliably tell you what's "best" - because the game is more complicated than that. But determining the highest DPS weapon for one single char under certain circumstances should be easy enough.

One could even put in the whole bestiary and make charts for a character with a certain weapon and how it performs against all the different enemies (of course some things like the Redeemer would make this a bit more difficult). If I only had time for that. ;)

Edited by Boeroer

Deadfire Community Patch: Nexus Mods

Link to comment
Share on other sites

 

With 118acc vs 104def you should have on average 1%miss, 35%grazes, 50%hits and 14%crits with ROGF and 1%misses, 28%grazes, 57%hits and 14%crits with Persistence.

 

The average damage per hit would be:

- for Persistence 12.5*(.28*1.48+.57*1.98+.14*2.78)=22.4pierce, 5.6corrode, 5.6burn, 7.5raw and after DR 7pierce, 1.1corrode, 1.1burn and 7.5raw (16.7 total)

- for RoGF 16.5*(.35*1.48+.50*1.98+.14*2.78)=31.3pierce, 7.9corrode, 7.9burn and after DR 13.5pierce, 1.6corrode, 1.6burn (16.7 total)

 

The delay between shots would be:

- for Persistence (1+1.66*(1-.38))/1.51=1.34s

- for RoGF (1.5+2.5*(1-.66))/1.51=1.56s

 

Over 30s that would make 747dmg for Persistence vs 642 for RoGF with Twinned Arrows. The advantage should be clearly on Persistence side, however my numbers are much lower than the ones from the test - maybe you used Fast Mode...

 

For the set-up Arranvin used the numbers are a bit different actually. Damage per shot after DR comes to 8.4 Pierce + 8.4 Raw + 4.3 Corrode + 4.3 Burn = 25.3 total for Persistance, and 14.9 Pierce + 6.2 Burn + 6.2 Corrode = 27.3 total damage for Rain. At the very least you seem to be missing the +20% hit-to-crit conversion, but I'm also not sure whether you are accounting for the 20% minimum damage after DR reduction. I have attack durations at 1.39 and 1.55 seconds respectively, and 60 * 25.3 / 1.39 = 1094.6 and 60 * 27.3 / 1.55 = 1059.5 damage output for 30 seconds I reported above as well (I'm using the underlying unrounded values for computation, obviously). My computations are not for fast mode, just normal mode; and Arranvin couldn't have used Fast mode either, since he tested during combat.

 

I computed the expected damage for my own tests in the same way as well, which as noted matched the observed results very well and suggests that my computations are sound. 

 

Yeah, I was tired and watched the olympics at the same time... However I don't understand how do you get so much lash dmg...

 

After corrections I have:
- for Persistence 12.5*(.28*1.48+.47*1.98+.24*2.78)=25.2pierce, 6.3corrode, 6.3burn, 8.4raw and after DR 8.2pierce, 2.1corrode, 2.1burn and 8.4raw (20.8 dmg total)
- for RoGF 16.5*(.35*1.48+.40*1.98+.24*2.78)=32.6pierce, 8.2corrode, 8.2burn and after DR 14.6pierce, 3.7corrode, 3.7burn (22dmg total)
 
And for the delays (gloves+durgan+chant):
- for Persistence (1+1.66*(1-.52))/1.51=1.19s
- for RoGF (1.5+2.5*(1-.79))/1.51=1.34s
 
And over 30s Persistence would do 1048.7dmg, while RoGF 985.1dmg... 
 
PS. Corrected speed...
Edited by Kaylon
Link to comment
Share on other sites

Like I said, I'm no sure if you've factored in the minimum damage component. But aside from that, the 2.1 / 3.7 Lash damage after DR is the amount I get when I plug in DR = 18, whereas for Burn/Corrode it was 8. Under 18 DR, the expected damage overall over 30s drops to 903.2 for Persistence and 870.1.

 

I'm also getting other attack durations though. Durgan +15%, Gauntlet +15% and Rain +20% speed enchantment all fall in the same attack speed category, so those are multiplicative and yield a bonus of 1.15*1.15 = +32.25% and 1.15*1.15*1.2 = +58.7% respectively. Add the +20% bonus from Ila and the -50% penalty from not dual-wielding, and we end up with +2.25% for Persistence and +28.7% for Rain, ie. S = 0.0225 and S = 0.287. For Persistence the attack/animation duration T = 1, so the recovery duration R = (1-2*S)/1.2 * T = 0.995/1.2 = 0.796 and the duration of a full attack cycle is 0.2 + 1.796/1.51 = 1.389 seconds. For Rain T = 1.5, R = (1-2*S)/1.2 * T = 0.426/1.2*1.5 = 0.5325 and the full attack cycle 0.2 + 1.5325/1.51 = 1.546 seconds.

 

I can't quite place how you are computing the attack durations, though the unscaled idle time seems to be missing. I have previously estimated it at about 0.2; there's a slight uncertainty in it (and it doesn't seem entirely constant), but for practical purposes it is close enough: timing 10 attacks each for Persistence and Rain under the above speed bonuses I get 13.8 seconds and 15.5 seconds respectively. 

Link to comment
Share on other sites

Yes, I factored the MIN damage.

 

For Persistence you have 1.15*1.15*-1+.2=52.2% speed, thus you're left with 47.8% recovery. And for RoGF you have 1.15*1.15*1.2*-1+.2=78.7% speed, thus you're left with 22.3% recovery. The base recovery of Persistance is 1.66s which means you're left with 1.66*.478=.79s and the base recovery of RoGF is 2.5s which means you're left with 2.5*.223=.56s. The attack+recovery phase of Persistence 1+.79=1.79s is reduced by the dex bonus to 1.79/1.51=1.19s and for RoGF 1.5+.56=2.06s which is further reduced by dex to 2.06/1.51=1.36s.

Edited by Kaylon
Link to comment
Share on other sites

Thanks very much for all your hard work Loren, that's really helpful. I can't explain the discrepancy between your results and mine. I'll take another look at my save to see if there are any other factors that I may have forgotten. I may even run the test again when I get the time. Not that I think it's necessary, I'm sure your results will be correct, I'm just curious about what has caused the discrepancy.

 

If you wouldn't mind, I'd be really interested to see a visual representation of how increased might would affect the damage for both bows versus increased dex.

 

See attached. This is the same default settings as the first plot (ie. without Chanting). 

 

I also ran a quick test for damage output with Chanting (10 minutes each). Expected damage against the 17.5/14 DR armour are 981.7 and 961.5 for Persistence and Rain respectively for 30 seconds. Observed values are slightly higher, 1002.8 and 992.6; this is well within the margin of error though. So still not sure what was different in your initial tests.

post-163298-0-12598000-1470751623_thumb.png

post-163298-0-60159500-1470751623_thumb.png

Link to comment
Share on other sites

Yes, I factored the MIN damage.

 

For Persistence you have 1.15*1.15*-1+.2=52.2% speed, thus you're left with 47.8% recovery. And for RoGF you have 1.15*1.15*1.2*-1+.2=78.7% speed, thus you're left with 22.3% recovery. The base recovery of Persistance is 1.66s which means you're left with 1.66*.478=.79s and the base recovery of RoGF is 2.5s which means you're left with 2.5*.223=.56s. The attack+recovery phase of Persistence 1+.79=1.79s is reduced by the dex bonus to 1.79/1.51=1.19s and for RoGF 1.5+.56=2.06s which is further reduced by dex to 2.06/1.51=1.36s.

 

I figured, it's just due to the 18DR vs 8DR difference on the lashes, then. 

 

It's 21.3% for Rain actually; but otherwise we now get the same attack and recovery duration, so that clears that discrepancy up as well. You do still need the +0.2s idle time (or wherever it's coming from, it's annoyingly difficult to trace in the code) to compute DPS though.

Edited by Loren Tyr
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...