THoK-Zeus Posted October 11, 2013 Share Posted October 11, 2013 Awesome thank you very much...holy hell 8% just from armor debuff...that's nutz What increase dps or % dps does using 50% rifle shot and 50% explosive probe work out to for you? I only ask because my fellow sniper doesn't think it affects him by more than 20-30dps (less than 1%) and trying to see if that's accurate....that way I can tell the other snipers to stop whining when I swtich to Firebug lol. I think it was 1,03% (so ~32 dps), so tell them to stop whining . Link to comment Share on other sites More sharing options...
odawgg Posted October 11, 2013 Share Posted October 11, 2013 I think it was 1,03% (so ~32 dps), so tell them to stop whining . Exactly what I wanted to hear Link to comment Share on other sites More sharing options...
Shandellon Posted October 11, 2013 Share Posted October 11, 2013 I'm sorry if it was already explained and if was, please just point me to a certain post, but how come hybrid gain for Gunslingers is higher than for DirtyFIghting? a) Both specs gain from not losing gcd on Flourish Shot equally. Yes, hybrid uses more k/e attacks, but this is already included if you compare it with dummies, as you throw armor debuff yourself. So actuall gain is only those few GCDs, not a direct dps increase b) While DF gains a straight dps increase below 30% due to increased damage from Shrap Bomb, Vital Shot and Wounding Shots, all that hybrid gains is a chance to use Quickdraw, wich actually rarely fits into rotation, being too costly Link to comment Share on other sites More sharing options...
THoK-Zeus Posted October 11, 2013 Share Posted October 11, 2013 (edited) I'm sorry if it was already explained and if was, please just point me to a certain post, but how come hybrid gain for Gunslingers is higher than for DirtyFIghting? a) Both specs gain from not losing gcd on Flourish Shot equally. Yes, hybrid uses more k/e attacks, but this is already included if you compare it with dummies, as you throw armor debuff yourself. So actuall gain is only those few GCDs, not a direct dps increase b) While DF gains a straight dps increase below 30% due to increased damage from Shrap Bomb, Vital Shot and Wounding Shots, all that hybrid gains is a chance to use Quickdraw, wich actually rarely fits into rotation, being too costly I updated my post with concerns about the hybrid numbers (about 1 page behind your post now). Now a post for the comparision: a) The values for Flourish Shot are similar calculated for both specs but have different values (3,21% and 2,7%). One reason for this diplacement is that red'october used 8 Shatter Shots (=3,92% of the time) for his Hybrid parse and 7 Shatter Shots (=3,43% of the whole fighting time) for his Dirty Fighting Parse. The calculations for his parses are correct, but i would rather go and change the hybrid value to 3,43% of the fighting time (which will decrease the hybrid value). These scalars are now generic numbers and should represent the increase with the minimum. b) i think there's a mistake in the execution calculations for the hybrid, seems a *2 is too much there (see my post before). Edited October 11, 2013 by THoK-Zeus Link to comment Share on other sites More sharing options...
Shandellon Posted October 12, 2013 Share Posted October 12, 2013 Ok, let me then reevaluate for Gunsligers/Snipers. I will use my own parses : Hybrid - http://www.torparse.com/a/454203/time/1381476565/1381476872/0/Damage+Dealt DF - http://www.torparse.com/a/453627/time/1381438364/1381438669/0/Damage+Dealt Pre-calculations : Previously estimatet values were ~5% additional damage, so I will use this to calculate time, instead of iterating myself - to save time. I know its not 100% accurate, but results shouldn't vary as much to justify adding myself extra work (in reality, I believe they won't change at all, due to discrete nature of a poblem, rather than continous) So, we need 70% of parse time to get into a burn phase. Then assuming our dps increases by 5% we need addtional 28.5% parse time to finish it off. I will therefore cut last 1.5% and analize timeframe 70%-> 98.5%.DirtyFIghting : Where do we gain: a) FlourishShot b) ShrapBomb, VitalShot, WoundingShots (internal) damage increase by 15% c) Quickdraw a) 6 Flourish Shots can be skipped. I will assume twice I will be high enough on energy to squizee a Charged Burst, other 4 would be Flurry of Bolts. That will be -7270 + 2*5233 + 4*1512 = +9244 damage change b) After cutting first 213s of fight we are left with 152'510 internal damage. 15% of that equals to 22'876.5 additional damage c) There is no gain. Quickdraw is a fixed part of rotation, we won't use any addtional Total damage from parse : 1'009'192 (3302.71dps) Damage gain : a + b + c = 9224 + 22'876.5 + 0 = 32'100.5 Adjusted time : 305*98.5% = 300.4s New dps : 3466.35This is 4.955% dps gain Hybrid : Where do we gain: a) FlourishShot b) Quickdraw a) 6 Flourish Shots can be skipped. I will assume once I will be high enough on energy to squizee a Charged Burst (hybrid is more energy fragile than DF), other 5 would be Flurry of Bolts. That will be -5244 + 4893 + 5*1313 = +6214 damage change b) We are left with 28.5% of the fight, so 87.5s. Technically, that is enough for 8 Quickdraws, but this will be disastrous for energy management, wont fit into rotation either. However, since we are about to finish combat anyway, energy is not that crucial. I will assume 1 Quickdraw per rotation. Lets make it 5, instead of Flurry of Bolts. Taking Quickdraw numbers form DF parse we end up with : 5*(-1313+4897) = +17'920 damage change Total damage from parse : 1'062'942 (3455.09dps) Damage gain : a + b = 6214 + 17'920 = 24'134 Adjusted time : 307*98.5% = 302.4s New dps : 3594.82This is 4.04% dps gain Link to comment Share on other sites More sharing options...
THoK-Zeus Posted October 12, 2013 Share Posted October 12, 2013 (edited) Ok, let me then reevaluate for Gunsligers/Snipers. I will use my own parses : Hybrid - http://www.torparse.com/a/454203/time/1381476565/1381476872/0/Damage+Dealt DF - http://www.torparse.com/a/453627/time/1381438364/1381438669/0/Damage+Dealt Pre-calculations : Previously estimatet values were ~5% additional damage, so I will use this to calculate time, instead of iterating myself - to save time. I know its not 100% accurate, but results shouldn't vary as much to justify adding myself extra work (in reality, I believe they won't change at all, due to discrete nature of a poblem, rather than continous) So, we need 70% of parse time to get into a burn phase. Then assuming our dps increases by 5% we need addtional 28.5% parse time to finish it off. I will therefore cut last 1.5% and analize timeframe 70%-> 98.5%.DirtyFIghting : Where do we gain: a) FlourishShot b) ShrapBomb, VitalShot, WoundingShots (internal) damage increase by 15% c) Quickdraw a) 6 Flourish Shots can be skipped. I will assume twice I will be high enough on energy to squizee a Charged Burst, other 4 would be Flurry of Bolts. That will be -7270 + 2*5233 + 4*1512 = +9244 damage change b) After cutting first 213s of fight we are left with 152'510 internal damage. 15% of that equals to 22'876.5 additional damage c) There is no gain. Quickdraw is a fixed part of rotation, we won't use any addtional Total damage from parse : 1'009'192 (3302.71dps) Damage gain : a + b + c = 9224 + 22'876.5 + 0 = 32'100.5 Adjusted time : 305*98.5% = 300.4s New dps : 3466.35This is 4.955% dps gain Hybrid : Where do we gain: a) FlourishShot b) Quickdraw a) 6 Flourish Shots can be skipped. I will assume once I will be high enough on energy to squizee a Charged Burst (hybrid is more energy fragile than DF), other 5 would be Flurry of Bolts. That will be -5244 + 4893 + 5*1313 = +6214 damage change b) We are left with 28.5% of the fight, so 87.5s. Technically, that is enough for 8 Quickdraws, but this will be disastrous for energy management, wont fit into rotation either. However, since we are about to finish combat anyway, energy is not that crucial. I will assume 1 Quickdraw per rotation. Lets make it 5, instead of Flurry of Bolts. Taking Quickdraw numbers form DF parse we end up with : 5*(-1313+4897) = +17'920 damage change Total damage from parse : 1'062'942 (3455.09dps) Damage gain : a + b = 6214 + 17'920 = 24'134 Adjusted time : 307*98.5% = 302.4s New dps : 3594.82This is 4.04% dps gain I don't really like the assumptions for Flourish Shot, but for hybrid its way too difficult to test it in reality. Guess we will have to take it like that, tough i would say you should have a Snipe/Charged Burst ready for every second Flourish Shot as then Adrenaline Probe comes off cooldown. That would be a slighty scalar increase for both hybrid/lethality. Also the increase in crit rating due to main stat and additional crit rating, should help a lot with energy problems, when you have full 78 gear. Next thing: For executes in this thread always 30% of the time was taken (and not 28,5% or sth. like that ). The reason for that is, that the time of the execute depends on the fight and the composition of the raid, but also execute damage is most of the time worth more then normal damage (Titan 6...). So far the calculations are good . I think there was a discussion about taking the average execute increase from all the dps specs and calculate an average execute time, but so far isn't done (or will never be done). Edited October 12, 2013 by THoK-Zeus Link to comment Share on other sites More sharing options...
Shandellon Posted October 12, 2013 Share Posted October 12, 2013 I think there was a discussion about taking the average execute increase from all the dps specs and calculate an average execute time, but so far isn't done (or will never be done). Woh, that would be just plain wrong. It varies too much between specs for average to be good aproximation Anyway, I never would have dwell into this so much, I just saw in first post hybrid>DF which i thought is impossible, so needed to check it myself Link to comment Share on other sites More sharing options...
greg_biochem Posted October 12, 2013 Share Posted October 12, 2013 (edited) I'm not sure how, or even if, it would change things... but with Hope's parses, using only 6 Flourish Shots means there are 30+ seconds without an armor debuff on the dummy. While in my parses the armor debuff is up 100% of the time. Edited October 12, 2013 by greg_biochem Link to comment Share on other sites More sharing options...
MGNMTTRN Posted October 12, 2013 Share Posted October 12, 2013 (edited) Yes, though the error from the way that I calculated it is fairly slight. When I get a more authoritative post-2.4 parse from both Assault mirrors, I will update the calculation to respect armor pen. Basically, the way I calculated the armor pen used it as a multiplicative buff. The game actually calculates armor penetration as an additive buff (thus, HiB has an actual armor pen of 95% when an armor debuff is present on the target, whereas my calculations assume the armor pen is 110%). The error here is fairly small when it's just a single ability with low armor pen (as with many abilities in the game), but the error increases dramatically as the armor pen goes up. High Impact Bolt has a high enough armor pen that the error is probably a few tenths of a percent, and thus no longer within our stated margin. tl;dr: The current number is wrong for precisely the reason you stated, but it's not wrong by a lot. I will compute a more correct number once I have a solid post-2.4 parse. 95.15% of the damage is K/E. With 30% armor pen as part of the spec, we have 0.9515 * ((1 - 0.35 * (1 - 0.2 - 0.3)) / (1 - 0.35 * (1 - 0.3)) - 1) = 8.82%.. The armor debuff can be calculated similarly. 74.97% of wild-berry's damage is affected by armor. Thus, that damage is reduced by 35% on a dummy. The armor debuff would account for the following DPS increase: 0.7497 * ((1 - 0.35 * (1 - 0.2)) / (1 - 0.35) - 1) = 8.07% This is not how armor penetration works. We'll use a Sniper as our test subject. I cleared my spec, flashed buffs and stim, and removed my SA relic. Then I specced just up to Interrogation probe and the game reported that it deals 6934 damage; over 7 ticks that's 990.57 damage per tick. When the probe is actually placed on a target dummy we see it only deals 644 non-crit damage, or 1-644/990.57 = 0.3498692672 damage was mitigated. Then I put my 20% armor debuff on the target dummy and record the new damage tick. It now ticks for 692 damage, which means 1-692/990.57 = 0.3014123182 damage is mitigated. The scalar resulting from the transition from 0% armor debuff to 20% armor debuff is 692/644 = 1.0745341615, not ((1 - 0.35 * (1 - 0.2)) / (1 - 0.35) ) = 1.1076923077. Then I specced 20 points into the Marksman skill tree for 10% armor penetration on all attacks. My Interrogation probe damage is still 6934 damage over 7 ticks, or 990.57 damage per tick. My actual damage with both the 20% armor debuff and 10% from tree was 714 damage per tick. This is means 1-714/990.57 = 0.2792028832 damage was mitigated. The scalar for adding 10% armor and 20% armor debuffs is 714/644 = 1.1087, not = ((1 - 0.35 * (1 - (0.2 +0.1))) / (1 - 0.35) ) = 1.1615384615. Interrogation probe was nice because it was a pure DOT and therefore had no spread on its damage. However, I am a patient guy... While I was speccing for 10% armor penetration in the Marksman tree, I also picked up the 20% armor penetration bonus on Ambush. I collected 101 non-crit samples and recorded the attack's reported spread. It was 6102-6451 damage, or 6276.5 on average presuming a symmetric distribution. My actual non-crit average was 4809.009901, which means 4809/6276.5 = 0.1915 damage is mitigated. And the scalar for upgrading from 0% armor penetration to 20/10/20 armor penetration is presumably 4809/(6276.5*(1-0.35)) = 1.178755921, not ((1 - 0.35 * (1 - (0.2 +0.1+0.2))) / (1 - 0.35) ) = 1.2692307692. Though I didn't check this. Ambush non-crit values with 10/20/20 armor penetration: 4914 4895 4772 4768 4725 4763 4948 4785 4706 4872 4762 4766 4721 4734 4785 4712 4842 4697 4950 4871 4768 4766 4909 4754 4944 4718 4725 4715 4791 4928 4904 4803 4729 4899 4693 4741 4721 4742 4924 4849 4718 4907 4758 4919 4934 4728 4689 4752 4747 4773 4710 4896 4717 4719 4800 4716 4838 4691 4783 4692 4948 4834 4890 4705 4867 4942 4787 4806 4862 4713 4779 4773 4813 4914 4817 4706 4851 4948 4754 4846 4948 4814 4700 4792 4794 4859 4897 4886 4875 4794 4953 4804 4749 4847 4951 4952 4876 4706 4700 4938 4822 So what does model these scalars well? This formula does: http://imgur.com/rHrnpdZ. It makes use of KBN's 2.0 player armor formula, and you'll notice that if you apply those multiplicative armor debuffs to a 7539 armor rating, e.g. (1-0.1)*(1-0.2)*(1-0.2)*7539 = 4342.464 and then feed this value into the armor formula, you get the mitigation values that I posted up top. Also in my formula image I think I have a -1 to match KBN's post formulas, while in this post I removed his -1 to get scalars to multiply proportions by. So... watch out for that. EDIT: By the way, the denominator of this image's formula is fixed for transitions only from 0% arpen to some arpen. If you parse with 9% arpen or 30% arpen or whatnot, you'll have to change the denominator. In addition to collecting data on my Sniper I did tests on my Commando with ranged attacks with damage spreads, not DOT ticks with a precise value. From those 3 tests with 101 data points each I have additional values, though they share the 0% and 20% test results. I compiled a table showing the result of treating armor debuffs additively with my formula, treating them multiplicatively (which is what I'm claiming is the best model of their damage), of KBN's formula, and of some actual examples of damage done. base rating 7538 0.35 Armor debuffs 1 -SUM(debuff) PI(1-debuff) sum pi kbn actual 0.00 0.00 0.00 7538.00 7538.00 1.00 1.00 1.00 1.00 0.20 0.00 0.00 6030.40 6030.40 1.08 1.08 1.11 1.08 0.20 0.10 0.20 3769.00 4341.89 1.21 1.17 1.27 1.18 0.09 0.20 0.00 5351.98 5487.66 1.11 1.11 1.16 0.20 0.35 0.00 3392.10 3919.76 1.24 1.20 1.30 1.21 0.20 0.35 0.30 1130.70 2743.83 1.42 1.29 1.46 The 'pi' and 'actual' values are very close. Due to rounding and the somewhat random nature of 101 sample points, the results are about as perfectly modeled as I could hope. Pending review of this post I'd be happy to give a general formula for a vector of scalars that accounts for attacks with mixtures of armor penetrations, e.g. Vanguards or Commandos who have different armor penetrations on {HIB, Full Auto} vs {Hammershot, Demo Round}. I can also post a table of scalars where you can transition from any known armor penetration (like 9% for Infiltration Shadows, 30% for Focus Sentinels) to that value with the 20% debuff added. Off the top of my head it looks like it would be {Assault Commando HIB/Full Auto at 30%, all Vanguard HIB at 30%, Focus sentinels at 30%, Infiltration Shadows at 9%, and Scrappers Scoundrels at 30% for 100% uptime I'm guessing...}. I don't want to tackle Combat. I can also post my Commando data, but I'd do that in a separate post since this one is becoming lengthy. Sidenote: I made histograms of my Ambush and Charged bolts non-crit damage and the populations didn't cluster about a mean, they looked uniformly distributed. I did not see any systematic difference in frequency at points where you'd see mercs/gunslinger offhand attacks cluster. I can post my Charged Bolts images or data if anyone wants to see it. Edited October 13, 2013 by MGNMTTRN Link to comment Share on other sites More sharing options...
Shandellon Posted October 13, 2013 Share Posted October 13, 2013 I'm not sure how, or even if, it would change things... but with Hope's parses, using only 6 Flourish Shots means there are 30+ seconds without an armor debuff on the dummy. While in my parses the armor debuff is up 100% of the time. "Does it change anything" depends on your actuall question a) Does it lower/increase theoretical dps b) Does it change anything in dummy->boss math a) Not keeping Flourish Shot on dummy 100% of the time actually increases dps by insignificant amount. I did not clip a single debuff, so total time debuff was present was 270 out of 300 seconds. There is a gap at start (no point debuffing while only FF is active + doting, thats about 5 secods), there are 1-2 seconds gaps between debuff running off and refreshing (where only affected thing that could happen is single Shock Charge tick) and there is no point refreshing it for the last 15 seconds, as that GCD would not have enough time to pay for itself (dps gain from armor reduction would be lower than dps gain from using ChargedBurst instead of Flourish). But this is strictly for higher dummy numbers. Notice that only last part applies to DF, as any non-debuff gaps in middle of parse are for refreshing dots, thus are not affected by presence of debuff anyway b) It would slightly change damage dealt in last ~15 seconds (no other time period matters, as explained above). So yes, technically all kinetic/energy damage in that period should be scaled up. Taking numbers from log, there was total of 30'364 k/e damage after last Armor reduction wore off. In actuall combat we would notice a damage gain of 2100 damage. This is less than 10% gain from other effects calculated in my original post. So you are right, it does change a little (I should have taken this into consideration earlier), its additional 7dps Instead of 4.04% gain for hybrid we now have 4.24% Using same algoritm for DF, we now have 5.11% instead of 4.95% Link to comment Share on other sites More sharing options...
oaceen Posted October 13, 2013 Share Posted October 13, 2013 Just a question... what relics are you using. I assume a SA of some level but I couldn't place the 2nd one from the logs. Thanks. sorry, just saw this question. i was using KD SA and UW BA Link to comment Share on other sites More sharing options...
paowee Posted October 22, 2013 Share Posted October 22, 2013 Just stumbled to your thread KBN. I like! Thanks for the great contributions! Link to comment Share on other sites More sharing options...
Emperor-Norton Posted October 22, 2013 Share Posted October 22, 2013 (edited) nevermind. Edited October 31, 2013 by Emperor-Norton Link to comment Share on other sites More sharing options...
odawgg Posted October 23, 2013 Share Posted October 23, 2013 (edited) To add to the armor debuff discussion, I did some testing about a year ago and I'm pretty sure it's not 20% off the initial 35% mitigation, rather it's 20% off the armor rating of 7538 (plug that into the DR formula and you get 35%), so 7538*(1-.2) = 6030.40 which you then plug into the DR formula... Then you have talents that ignore armor and works in same way, so for example, if you have an ability that ignores 30% of armor. The armor rating would be 7538*(1-.3) = 5276.60 which you plug into DR formula. To find the same ability with armor debuff you would take that value 2454.9 *(1-.2) =4221.28 and plug that in... So the scalar would be the ratio of after debuff/before debuff. I'll post a spreadsheet for Pyro that I did when I get home, and of anyone has imperical evidence negating my claims, please share, but I recall confirming this back in the day (although my memory is bad at times) Edited October 26, 2013 by odawgg Link to comment Share on other sites More sharing options...
odawgg Posted October 24, 2013 Share Posted October 24, 2013 (edited) .....continued I took data from this parse: Merc Pyro Parse * Updated: And here's my spreadsheet, tried to show all my work...only thing I left out was the damage reduction equation: Spreadsheet Edit: I will also try to do some actual testing to back up this theory....WTB the data I collected a year ago, I could end up being way off here but my gut is telling me this is right. Edited October 27, 2013 by odawgg Link to comment Share on other sites More sharing options...
odawgg Posted October 24, 2013 Share Posted October 24, 2013 (edited) .....continued Yes, I have confirmed that armor debuff is taken off of Armor Rating, not the straight percentage. I unequipped my SA relic so power was stable. I'm rounding to 2 decimals in this thread, but all the calculated info is based off unrounded numbers. I used data from this parse: Arsenal Dummy Test Now look at the Tracer Missile data: Average Damage: 3921.24 Crit %: 44.8 Surge %: 72.96 Therefore Average "Base" Damage is 3921.24/(1+.448*.7296) = 2955.27 Calculated damage for Tracer Missile (also what appears on the tooltip when you hover over TM with my gear stats): Coefficient: 1.75 Bonus Damage: 1660.60 St Health Percent: .125 (min) - .23 (max) St Health: 2685 Skill Tree Modifier: 1.12 (12% cumulative buff) Ability Damage: 3630.67 (min) - 3931.39 (max) Therefore Average Ability Damage = 3781 Ok so now we factor in both the 35% armor pen from HVGC and the 20% armor debuff. If it was straight percentage, we would be looking at a armor debuff + HVGC modifier of 1-(.35*(1-.35)*(1-.2)) = .818. So the average damage "should" be 3781*.818 = 3093. Not looking like 2955.27 we got from the data yet. Ok so instead lets assume it's off the armor rating. Now we would be looking at a armor rating after armor debuff + HVGC modifier of 7538*(1-.35)*(1-.2) = 3919.76. Plug this into the Damage Reduction formula found Here and you get an armor mitigation of 21.88%. So the damage scalar/modifier is 1-.2188 = .78. So the average damage comes to 3781*.78 = 2953.79. Ahhhhhhh that's much closer to the 2955.27 from the data I don't need to do any further testing to have confidence in it. 357 TMs is a decent sample set and it's already bringing back the memories of a year ago when I conducted this study.... What I do still need to test is Unload with Arsenal and Rail Shot with Pyro....because I want to make sure the talent trees armor ignore is also multiplicative and not additive. In particular I'm not sure if the 2 talents in the pyro tree that both give 30% armor ignore to rail shot is additive for 60% (making it 1-.6 = .4) or if it's multiplicative (1-.3)*(1-.3) = .49. I will try to test those soon. Edited October 26, 2013 by odawgg Link to comment Share on other sites More sharing options...
MGNMTTRN Posted October 24, 2013 Share Posted October 24, 2013 (edited) The data definitely supports using ~7500 armor rating and applying KBN's armor rating formula for players to calculate damage reduction. However, some of that data was from my Interrogation Probe tooltip; that part worked beautifully, its DOT ticks came out to exactly what they should have. My Commandos's Full Auto parses are absolutely not parsing as their tooltip indicates they should, and neither were my Vanguard's HIBs. Unfortunately I'm traveling right now and I expect the conclusion to be reached will be 'SWTOR tooltips are useless' but if someone else wants to investigate it here's the data. 20 minute parse All full autos were cast with the 30% COF proc, which on my Commando had an updated tooltip spread of 2780-3280 First 100 non-crit Full Autos 2755 2426 2661 2740 2738 2619 2645 2665 2799 2683 2802 2648 2721 2778 2823 2715 2635 2605 2723 2829 2428 2469 2738 2683 2812 2492 2811 2794 2648 2713 2496 2436 2633 2787 2663 2587 2576 2815 2844 2541 2429 2703 2514 2674 2705 2778 2731 2687 2679 2662 2662 2675 2712 2517 2554 2707 2502 2644 2510 2770 2549 2752 2762 2840 2649 2728 2713 2762 2607 2774 2488 2721 2619 2473 2582 2775 2573 2734 2428 2615 2582 2736 2421 2666 2491 2548 2564 2775 2555 2567 2764 2814 2476 2734 2564 2771 2546 2665 2723 2765 2745 So center of tooltip spread = (2780+3280)/2 = 3030, average of those 100 samples = 2657.495049505 . That's a damage reduction of 1 - 2744/3030 = 1 - 0.905 = 0.094. Whereas Full Auto should get armor rating 7539*(1-0.35)*(1-0.3)*(1-0.2) = 2744 and then damage reduction should be applied to 2744/(2744+240*55+800) = 0.164 proportion of the outgoing damage. That's 3030*(1-0.164) = 2533, which was significantly different from my expected 2657.495049505. // 10/25 edit: changed some values from 2744 to 2657.495049505, because I mixed up armor rating and average damage done in several cases. Post's accusation is still correct: damage done was too high As noted, I had an extremely similar problem with predicting Vanguard HIB damage from Assault spec using tooltips, even when 1) not speccing for any of the bonuses to burning or bleeding targets 2) trying to incorporate those effects into the model. I have the HIB data, but I recommend someone else conduct retests anyway. It couldn't be a cap on armor penetration because both my Commando parses of Full Auto (with 0.35, 0.3, 0.2 arpen) and Assault Vanguard (0.45, 0.3) had values higher than a combination of tooltip + arpen application would have predicted. It's probably just a tooltip problem... Once I get back home on ~Sunday I'll probably parse with speccing for the Full Auto arpen and without and see if that helps. Same for Vanguard HIBs. If someone wants to do the data collection before me the paradigm would be 1) collect HIB/FA data from a Commando before speccing into arpen in Assault tree 2) collect data after speccing for arpen in Assault tree. If the pi model given in my previous post predicts that average increase in non-crit from (0.2, 0.35) to (0.2, 0.3, 0.35) armor penetration, then we can know the tooltip is wrong and I can move on with my life. Also I can't log into SWTOR right now, but if you really wanted to test what happens at arpens exceeding 100%... you can spec for 30% HIB arpen in the Assault tree, 30% HIB+Full Auto arpen in the Assault base, get 35% from AP/HVGC stance, and 20% from the Grav Round debuff for a sum of 1.15 armor penetration. It would be a cool way to test/do a corner case for the Pi model. Edited October 26, 2013 by MGNMTTRN Link to comment Share on other sites More sharing options...
ceazare Posted October 26, 2013 Share Posted October 26, 2013 I did some testing too and I can also confirm that 20% armor debuff is on Armor Rating. All further armor penetration talents I tested have been multiplicative. Base Armor Rating: 7538 Damage Reduction: 35 % After 20 % armor debuff Armor Rating 7538*(1-0.2) = 6031 Damage Reduction: 30.1 % Add 15 % armor penetration Armor Rating: 7538*(1-0.2)*(1-0.15) = 5126 Damage Reduction: 26.8 % Link to comment Share on other sites More sharing options...
odawgg Posted October 26, 2013 Share Posted October 26, 2013 (edited) oh yeah whoops, My 3507 armor rating was because I was plugging it into a sin tanking armor equation that was multiplying armor rating by 2.15 That had to be pretty confusing to the readers lol. 7538 armor rating was what I meant, I'll edit my post, doesn't change the results though Edited October 26, 2013 by odawgg Link to comment Share on other sites More sharing options...
cxten Posted October 26, 2013 Share Posted October 26, 2013 With 2.5, it's possible that inspiration should be added to this, as it's now like the 20% armor debuff (everyone should get it, assuming you have a sentinel in your group). At the least this is a 0.75% (= 15% * 15/300) buff for non-sentinels across the board. Previously, since each added sentinel brought a new inspiration, it would have actually been proper to add the damage increase of the 3 other people in the sentinel's group to the sentinel's adjusted dps. That means the sentinel would have roughly a 2.25% damage increase (or rather, 2.25% of the average dps across classes, say). Link to comment Share on other sites More sharing options...
odawgg Posted October 27, 2013 Share Posted October 27, 2013 (edited) I did some testing too and I can also confirm that 20% armor debuff is on Armor Rating. All further armor penetration talents I tested have been multiplicative. Add 15 % armor penetration Armor Rating: 7538*(1-0.2)*(1-0.15) = 5126 Damage Reduction: 26.8 % I thought so, thank you for testing, have you had a chance to test any abilities that receive multiple armor pens through talents? I am thinking they are multiplicative too but I'm not certain. When I get time I will test rail shot in pyro as it has 2 talents. I updated my Spreadsheet assuming multiplicative armor pen talents, 4.89% damage increase from armor debuff for mercenary pyro with this parse anyway. Having a harder time deciding how to treat the sub 30 burn, I know there's been discussion on it and maybe yall decided on a common ground to base it off of. Group composition as well as the encounter mechanics play a big role. Edited October 27, 2013 by odawgg Link to comment Share on other sites More sharing options...
ceazare Posted October 27, 2013 Share Posted October 27, 2013 I thought so, thank you for testing, have you had a chance to test any abilities that receive multiple armor pens through talents? I am thinking they are multiplicative too but I'm not certain. When I get time I will test rail shot in pyro as it has 2 talents. I updated my Spreadsheet assuming multiplicative armor pen talents, 4.89% damage increase from armor debuff for mercenary pyro with this parse anyway. Having a harder time deciding how to treat the sub 30 burn, I know there's been discussion on it and maybe yall decided on a common ground to base it off of. Group composition as well as the encounter mechanics play a big role. Sorry, my claim above was actually wrong. I tested each talent against an armor debuff without other similar talents interfering. I spent some time today on dummy and tested everything that has to do with armor debuffs / armor ignores and armor penetrations on mercs and snipers. It turns out that ignores add, so if you take Advanced Targetting (Rail Shot and Unload ignore 30 % of target's armor) Superheated Rail (Rail shot ignores 30 % of target's armor), the two talents will stack additively and rail shot will ignore 60 % of target's armor. Armor Debuff will stack multiplicatively. I also tested High Velocity Cylinder on top of all that and to my surprise its 35 % armor penetration stacks additively to armor ignores (for a whopping 95% armor penetration on Rail Shots). All of the above percentages refer to Armor Rating. Original Armor Rating: 7538 Original Damage Reduction: 35 % Equiv Armor Rating when 60 % ignored: 7538 * (1 - 0.3 - 0.3) = 3015 Damage Reduction: 17.72 % Equiv Armor Rating when 60 % ignored with High Velocity Cylinder: 7538 * (1 - 0.3 -0.3 - 0.35) = 377 Damage Reduction: 2.62 % Add Armor Debuff on top of it all: 7538 * (1 - 0.95) * (1-0,2) = 302 Damage Reduction 2.11 % Sample size: 50 rail shots for each talent/skill added I tested similar talents and skills on sniper and got same results. Armor penetrations and ignores talents stack additively, but multiplicatively wih an armor debuff. Link to comment Share on other sites More sharing options...
odawgg Posted October 28, 2013 Share Posted October 28, 2013 Sorry, my claim above was actually wrong. I tested each talent against an armor debuff without other similar talents interfering. I spent some time today on dummy and tested everything that has to do with armor debuffs / armor ignores and armor penetrations on mercs and snipers. It turns out that ignores add, so if you take Advanced Targetting (Rail Shot and Unload ignore 30 % of target's armor) Superheated Rail (Rail shot ignores 30 % of target's armor), the two talents will stack additively and rail shot will ignore 60 % of target's armor. Armor Debuff will stack multiplicatively. I also tested High Velocity Cylinder on top of all that and to my surprise its 35 % armor penetration stacks additively to armor ignores (for a whopping 95% armor penetration on Rail Shots). All of the above percentages refer to Armor Rating. Original Armor Rating: 7538 Original Damage Reduction: 35 % Equiv Armor Rating when 60 % ignored: 7538 * (1 - 0.3 - 0.3) = 3015 Damage Reduction: 17.72 % Equiv Armor Rating when 60 % ignored with High Velocity Cylinder: 7538 * (1 - 0.3 -0.3 - 0.35) = 377 Damage Reduction: 2.62 % Add Armor Debuff on top of it all: 7538 * (1 - 0.95) * (1-0,2) = 302 Damage Reduction 2.11 % Sample size: 50 rail shots for each talent/skill added I tested similar talents and skills on sniper and got same results. Armor penetrations and ignores talents stack additively, but multiplicatively wih an armor debuff. You're awesome for testing all that...gonna save me many hours this week Thank you Link to comment Share on other sites More sharing options...
Falver Posted October 29, 2013 Share Posted October 29, 2013 (edited) With 2.5, it's possible that inspiration should be added to this, as it's now like the 20% armor debuff (everyone should get it, assuming you have a sentinel in your group). At the least this is a 0.75% (= 15% * 15/300) buff for non-sentinels across the board. Previously, since each added sentinel brought a new inspiration, it would have actually been proper to add the damage increase of the 3 other people in the sentinel's group to the sentinel's adjusted dps. That means the sentinel would have roughly a 2.25% damage increase (or rather, 2.25% of the average dps across classes, say). I would rather not see inspiration in the calculations personally. Seems kinda niche-y to add to anyone other than a sentinel. Anyways, I'm a little too tired to really look through the math but, what's the consensus on the new math? The scalars still reasonable and acceptable? Edited October 29, 2013 by Falver Link to comment Share on other sites More sharing options...
Shandellon Posted October 29, 2013 Share Posted October 29, 2013 what's the consensus on the new math? The scalars still reasonable and acceptable? Reasonable yes, but not accurate. Almost every number differs from what is stated in first post. They are however in 1%-2% range (in percentage points) Link to comment Share on other sites More sharing options...
Recommended Posts