Elitist Jerks Computer simulation analysis of resilience vs hit points

 03/06/07, 10:05 AM #1 Tantelus Glass Joe   Tantelus Gnome Mage   Mal'Ganis Computer simulation analysis of resilience vs hit points I wrote a basic simulation to determine the benefit of resilience vs hit points. Basically the program simulates damage, variable between 50%-100% of a controllable base damage, and does iterations till the "character" dies. A crit is calculated as double damage. Resilience effect is calculated as a -1% crit rating reduction and a 2% reduction to damage done when crit for 39.4 rating points. Limitations: This program does not take into account non linear scaling of damage with crits due to effects like flurry. Nor does it calculate the benefit of a crit based on your opponent, for instance a frost mage specced 40/0/21 will crit harder than 2x. If anyone has any recommendations I'm willing to modify my program. Longevity is measured in the mean lifetime (MLT), units are number of iterations. Data: Basic stats: 10000 HP 0 RES MLT: 166 seconds ```Hit point difference @20% crit: HP MLT 10000 166 10500 175 11000 183 12000 200 Resilience difference @20% crit: Res MLT 000 166 050 169 100 172.5 150 175 200 178 300 183 Hit point difference @30% crit: HP MLT 10000 153 10500 161 11000 169 12000 184 Resilience difference @30% crit: Res MLT 000 153 050 157 100 160 200 166 250 169 300 171``` So the quick and dirty is this: At 20% crit, 150 resilience is equal to about 500 hit points or 1 resilience is equal to about 0.33 stam At 30% crit, 125 resilience is equal to about 500 hit points or 1 resilience is equal to about 0.4 stam. Overall it seems more beneficial to find socketed gear and toss in gems rather than get pvp gear with resilience. Last edited by Tantelus : 03/06/07 at 11:39 AM. Reason: table fixed.
 03/06/07, 10:20 AM #2 Malan Mike Tyson     Malan Tauren Shaman   No WoW Account So where is the crossover point at which the resilience becomes better? (if ever?)
 03/06/07, 10:30 AM #3 • frmorrison Protector     Ashstrike Human Paladin   No WoW Account The note about resilience, the PvP gear has decent resilence and good stamina, while most PvE gear doesn't have good stamina. Sockets can make up for the lower base stamina though. However, your simulation likely doesn't take into account that PvP fights are sometimes about luck, and when resilence stops that big crit that can help you win. What is resilience's cost vs stamina in the item budget?
 03/06/07, 10:39 AM #4 Tiiki Rogue About Town   Tiiki Troll Rogue   The Venture Co (EU) And am I right in thinking that 1 resilience costs the same as 1.5 stam in terms of itemisation (stacking one stat excepted). In fact, essentially, isn't it the fact that you can have 'lots' of Stam and 'lots' of Resilience on a particular item for less budget that 2 x 'lots' of Stam that makes Resilience useful? I.e. giving two survivability stats to pump instead of just one. Other than that, it seems to me it's clearly overpriced. Of course it also dampens spike damage pretty well at high levels, which makes you much easier to heal. Who knows if the 184 seconds at 12,000 HP wasn't 100 seconds of not very much followed by an instagib crit chain? The best thing to add to your simulation would be throwing out a CSV of current HP over time that could be graphed in Excel, perhaps?
 03/06/07, 10:43 AM #5 Tantelus Glass Joe   Tantelus Gnome Mage   Mal'Ganis I'm not entirely sure what the cost of resilience vs stam is but stam is definitely cheaper. If you compare on gems, 10 resilience is an epic gem whereas 12 stam is a blue gem. It is conceivable that 1 resilience = 1.5 stam for item cost. There's no cross over point where 1 stam equals 1 resilience for survivability. One thing to consider however is damage taken. Say a person is taking 30000 damage in an arena, that clearly exceeds their hit point. In this case, resilience has granted the extra survivability for free, whereas hp just creates a larger pool to heal.
 03/06/07, 10:45 AM #6 Tantelus Glass Joe   Tantelus Gnome Mage   Mal'Ganis The results you see are an average of 1000 simulations. I actually have the graph of damage done over time displaying as the program runs, I could try calculating the standard deviation from the mean for high resilience and high stam. That should give a sense of the spiky-ness of the damage being done.
03/06/07, 10:58 AM   #7
Avair
The Howard Roark of Shipwrights

Avair
Human Rogue

No WoW Account
 One thing to consider however is damage taken.
This is not an insignificant consideration. Serious Arena teams are going to be healing damage, so any calculations on healing should take that into account. If resilience increases the value of the heals (much like Quick Recovery does) than it should be accounted for.

03/06/07, 11:02 AM   #8
Tiiki

Troll Rogue

<Moo>
The Venture Co (EU)
 One thing to consider however is damage taken. Say a person is taking 30000 damage in an arena, that clearly exceeds their hit point. In this case, resilience has granted the extra survivability for free, whereas hp just creates a larger pool to heal.
What exactly do you mean by this?

And I believe, correct me if I'm being really stupid, that the TBC itemisation formula was basically:

1 non-stam stats = 1 of ANY rating/resilience/etc = 2 AP = 1.5 stam?

I.e. it was balanced so 1 hit rating = 1 crit rating because you need more crit rating (22.1 vs 15.8) for 1% at 70.

03/06/07, 11:05 AM   #9
levk
King Hippo

Byashi
Gnome Warrior

No WoW Account
 Originally Posted by Avair This is not an insignificant consideration. Serious Arena teams are going to be healing damage, so any calculations on healing should take that into account. If resilience increases the value of the heals (much like Quick Recovery does) than it should be accounted for.
Is this true? This is the first time I hear this.

 03/06/07, 11:09 AM #10 Shiftir Von Kaiser   Shiftir Night Elf Druid   Stormscale What he means is resilience would slow the incoming flow of damage thus making the outgoing flow of heals more valuable. Not increasing the size of heals in any way, just increasing the effective "time to live" granted by each heal of X amount.
 03/06/07, 11:12 AM #11 Tantelus Glass Joe   Tantelus Gnome Mage   Mal'Ganis Ok so here are the results of the standard deviation calculations at 30% crit rate: ```HP RES MLT SD 10K 0 102 4% 11K 0 112 4% 10K 100 106 3.7% 10K 200 110 3.4%``` SD here is the percent standard deviation among the 1000 simulations and represents the variation in how fast you die. As you can see (and probably expected) increased resilience makes death time more predictable while increased HP does nothing. So yes, you're less likely to get omgcrit near the end of your hit points and therefore more likely to get a heal. As for the healing efficacy, I would calculate the healing bonus factor as follows: Effective healing = (Raw heal) * ((HP value of your resilience rating)/(Total HP)+1) This way the % of your effective health coming from resilience would be like +healing on your healer. Last edited by Tantelus : 03/06/07 at 11:36 AM. Reason: table fixed, thx.
 03/06/07, 11:17 AM #12 Tantelus Glass Joe   Tantelus Gnome Mage   Mal'Ganis Tiiki, what I mean is that when you're looking at taken damage you have to consider what effect resilience had to that damage. Say your listed taken damage is 30000. With 250 resilience (using my conversion factor at 30% crit rate), it will seem as if you had 11K HP for every 10k HP you had. Therefore the true damage done to you will have been 33000, though only 30k of it had to be healed, since resilience mitigated about 3k damage. Had you gone the higher HP route, you would have had to heal the extra 3k damage as well.
 03/06/07, 11:18 AM #13 Viktus Von Kaiser   Viktus Undead Warlock   Archimonde What numbers did you use for the base damage?
03/06/07, 11:18 AM   #14
levk
King Hippo

Byashi
Gnome Warrior

No WoW Account
 Originally Posted by Shiftir What he means is resilience would slow the incoming flow of damage thus making the outgoing flow of heals more valuable. Not increasing the size of heals in any way, just increasing the effective "time to live" granted by each heal of X amount.
The only time this would be an issue is if the heal you land on a target isn't enough to make them live to see the next heal. At the very best this is single target HPS throughput problem at worst it's lack of avoidance (broken CC or whatnot). I can't imagine resilience on your whole team is a good way to solve it.

03/06/07, 11:24 AM   #15
Tantelus
Glass Joe

Gnome Mage

<TDC>
Mal'Ganis
 Originally Posted by Viktus What numbers did you use for the base damage?
I tried a variety of different base damage and it didn't make a difference as far as the numbers go. The decimal point shifted obviously but the outcome was the same. Base damage doesn't seem to make a difference in terms of mean lifetime, in that respect every class benefits from resilience equally. I can't remember what the base damage was for the OP, the 2nd post with the standard deviations used a base damage of 100.

 Elitist Jerks Computer simulation analysis of resilience vs hit points

 Similar Threads Thread Thread Starter Forum Replies Last Post Dragonspine Trophy Analysis Aldriana Public Discussion 494 11/02/08 12:20 AM T2 - T2.5 - T3 upgrade spreadsheet analysis Malan Public Discussion 17 08/29/06 11:16 AM Combat Log Analysis draghkar Public Discussion 31 06/01/06 9:11 AM Important ingame analysis. Slug Public Discussion 33 01/13/06 11:36 AM