[Resto] Healing Equivalency Points - Page 11

stassart · 03/12/09, 6:07 PM

I wanted to create my own thread on these forums, but it won't let me, so this seems the best place.

When browsing these and other shaman forums, I saw too many people theorizing on shaman HEP based on perfect conditions (100% replenishment, dropping mana tide every cooldown on the dot, 5 min fights, spamming a single spell, or just a rotation of two spells). I did not feel this would provide a realistic method for calculating HEP.

In college, one of my favorite quotes from a professor was "Theory is closer to reality, in theory, than in reality." In the spirit of that quote, I decided to create a program that would go through combat logs and determine based on actual data resto shaman Healing Equivalency Points.

I have released the program on Curse Gaming as shaman_hep:

shaman_hep

Provides a report giving the following statistics:

Total time in combat
Longest fight

For each spell:

Number of hits

Total healing (% of overall)
Effective healing (% of overall)
Number of crits (crit %)
Average non-crit, average crit, average effective, average raw
Overheals split into crit and non-crit, giving the number of overheals, the total amount overhealed, and the average amount overhealed
For chain heal, it gives the number of casts canceled and percentage (canceled, out of range, out of line of sight)
For chain heal, it gives the above statistics for each hop, and a percentage of chain heals that bounced to each hop
For earth shield, average time between procs
Stats for Glyph of Chain heal
What stats increase that spell, and by what amount

Total mana restored
Total mana restored in combat

For each source of mana:
Count
Total mana restored
In combat count
In combat mana restored
Average overall, average in combat
Overall in combat mp5 for this source
What water shield procs are from damage and which are from Improved Water Shield
Proc rate for Improved Water Shield
What stats increase the mp5 of this source, and by what amount

HEP Values for:
Spell Power
mp5 (you set the ratio of Spell Power to mana regen)
Haste rating
Crit rating
INT (actual)
INT (theoretical)

Features:

Uses actual data from combat logs.
Uses known values for spell coefficients.
Can set your talent points in Ancestral Knowledge and Nature's Blessing, and the program factors these talents into HEP values.
Associates Ancestral Awakening with the spell that proced it, and factors it into the scaling of that spell
Associates which water shield procs were triggered by damage

stassart · 03/12/09, 6:22 PM

The value of shaman_hep is that it can provide HEP values tailored to each shaman and their environment. The numbers are going to be different based on your role in raids and what ratio of spells you use.

In addition, the value of spell power to mana regen is configurable. The program defaults to a ratio of 1 mp5 = 5 SP, but you can change that to whatever ratio of healing output to mana regen you like. Based on that ratio it will determine the value of haste, crit, and int.

However, I thought it would be useful to provide a sample of what the program produces.

Here is my overall data:

Total log time processed: 12 days 7 hours 51 mins 45 secs 
Evaluating heals for "Trelis":
Time in combat: 8 hours 20 mins 6 secs 
Longest fight: 12 mins 3 secs 
Effective healing: 36080607
Total healing: 67085722
"Ancestral Awakening": hits: 1113, effective healing: 1303310 (3.61% of player's overall)
	Total (raw) healing: 1582272 (2.36% of player's overall)
	Crits: 0 (0.00%)
	Averages:
		Non-crit: 1421, crit: 0, effective: 1170, overall: 1421
	Overheals:
		Non-crit: count: 270, total: 278962, ave: 1033
		Crit: count: 0, total: 0, ave: 0
"Chain Heal": hits: 7801, effective healing: 14117946 (39.13% of player's overall)
	Total (raw) healing: 29575853 (44.09% of player's overall)
	Crits: 2818 (36.12%)
	Averages:
		Non-crit: 3253, crit: 4742, effective: 1809, overall: 3791
	Overheals:
		Non-crit: count: 2442, total: 8012329, ave: 3281
		Crit: count: 1637, total: 7445578, ave: 4548
	Number of Chain Heal casts started: 2881 (7.57% canceled)
	Initial target hits: 2663, crits: 926 (34.77%)
		Effective: 5535870 (39.21%), Total: 18300253 (61.88%)
		Ave: non crit: 5857, crit: 8775, effective: 2078, overall: 6872
		Overheals non-crit count: 1423, non-crit ave over: 4769
		Overheals crit count: 848, crit ave over: 7047
		Overhealed amount: 12764383 (69.75%)
	Second target hits: 2062 (77.43% bounced), crits: 756 (36.66%)
		Effective: 5136290 (36.38%), Total: 7150979 (24.18%)
		Ave: non crit: 2935, crit: 4387, effective: 2490, overall: 3467
		Overheals non-crit count: 573, non-crit ave over: 1589
		Overheals crit count: 446, crit ave over: 2474
		Overhealed amount: 2014689 (28.17%)
	Third target hits: 1675 (62.90% bounced), crits: 601 (35.88%)
		Effective: 2369984 (16.79%), Total: 2895705 (9.79%)
		Ave: non crit: 1460, crit: 2207, effective: 1414, overall: 1728
		Overheals non-crit count: 297, non-crit ave over: 815
		Overheals crit count: 223, crit ave over: 1271
		Overhealed amount: 525721 (18.16%)
	Fourth target hits: 1401 (52.61% bounced), crits: 535 (38.19%)
		Effective: 1075802 (7.62%), Total: 1228916 (4.16%)
		Ave: non crit: 731, crit: 1112, effective: 767, overall: 877
		Overheals non-crit count: 149, non-crit ave over: 479
		Overheals crit count: 120, crit ave over: 680
		Overhealed amount: 153114 (12.46%)
	Glyph of Chain heal: effective: 1075802 (2.98% of total effective)
	Processed by cast: 2881, time: 2630, less accurate: order: 3, count: 3
	HPS increased by:
		1 SP = 0.309325
			(0.2173 + 0.3021 + 0.1694 + 0.0845)
		1 Haste rating = 0.21717132675298
		1 crit rating = 0.163082294779396
			(0.0324 + 0.0600 + 0.0436 + 0.0270)
		1 int = 0.0539071679224853
			(0.0107 + 0.0198 + 0.0144 + 0.0089)
"Earth Shield": hits: 2533, effective healing: 4159612 (11.53% of player's overall)
	Total (raw) healing: 5203052 (7.76% of player's overall)
	Crits: 813 (32.10%)
	Averages:
		Non-crit: 1774, crit: 2645, effective: 1642, overall: 2054
	Overheals:
		Non-crit: count: 413, total: 545883, ave: 1321
		Crit: count: 279, total: 497557, ave: 1783
	Earth shield average time between procs: 8.77 secs
	HPS increased by:
		1 SP = 0.0509985848164905
		1 crit rating = 0.0143184171097279
		1 int = 0.00473298046585869
"Earthliving"_hot: hits: 3039, effective healing: 1766329 (4.90% of player's overall)
	Total (raw) healing: 1959035 (2.92% of player's overall)
	Crits: 0 (0.00%)
	Averages:
		Non-crit: 644, crit: 0, effective: 581, overall: 644
	Overheals:
		Non-crit: count: 538, total: 192706, ave: 358
		Crit: count: 0, total: 0, ave: 0
	HPS increased by:
		1 SP = 0.18763672260612
		1 crit rating = 0
		1 int = 0
"First Aid"_hot: hits: 5, effective healing: 4130 (0.01% of player's overall)
	Total (raw) healing: 4130 (0.01% of player's overall)
	Crits: 0 (0.00%)
	Averages:
		Non-crit: 826, crit: 0, effective: 826, overall: 826
	Overheals:
		Non-crit: count: 0, total: 0, ave: 0
		Crit: count: 0, total: 0, ave: 0
	HPS increased by:
"Forethought Talisman"_hot: hits: 416, effective healing: 364261 (1.01% of player's overall)
	Total (raw) healing: 420957 (0.63% of player's overall)
	Crits: 0 (0.00%)
	Averages:
		Non-crit: 1011, crit: 0, effective: 875, overall: 1011
	Overheals:
		Non-crit: count: 101, total: 56696, ave: 561
		Crit: count: 0, total: 0, ave: 0
	HPS increased by:
"Healing Stream Totem"_hot: hits: 863, effective healing: 71785 (0.20% of player's overall)
	Total (raw) healing: 220391 (0.33% of player's overall)
	Crits: 0 (0.00%)
	Averages:
		Non-crit: 255, crit: 0, effective: 83, overall: 255
	Overheals:
		Non-crit: count: 592, total: 148606, ave: 251
		Crit: count: 0, total: 0, ave: 0
	HPS increased by:
		1 SP = 0.00706546929316338
		1 crit rating = 0
		1 int = 0
"Healing Wave": hits: 225, effective healing: 1321042 (3.66% of player's overall)
	Total (raw) healing: 2398674 (3.58% of player's overall)
	Crits: 82 (36.44%)
	Averages:
		Non-crit: 9243, crit: 13132, effective: 5871, overall: 10660
	Overheals:
		Non-crit: count: 67, total: 517619, ave: 7725
		Crit: count: 63, total: 560013, ave: 8889
	Ancestral Awakening: procs: 54, total: 170922, effective: 148606
	Effective healing w/ Ancestral: 1469648 (4.07% of overall)
	Total healing w/ Ancestral: 2569596 (3.83% of overall)
	HPS increased by:
		1 SP = 0.299213688888889
		1 Haste rating = 0.704555733333333
		1 crit rating = 0.298989853032492
		1 int = 0.0988316741332615
"Lesser Healing Wave": hits: 3124, effective healing: 6145019 (17.03% of player's overall)
	Total (raw) healing: 16140406 (24.06% of player's overall)
	Crits: 1124 (35.98%)
	Averages:
		Non-crit: 4382, crit: 6562, effective: 1967, overall: 5166
	Overheals:
		Non-crit: count: 1366, total: 5053071, ave: 3699
		Crit: count: 915, total: 4942316, ave: 5401
	Ancestral Awakening: procs: 506, total: 761893, effective: 617902
	Effective healing w/ Ancestral: 6762921 (18.74% of overall)
	Total healing w/ Ancestral: 16902299 (25.20% of overall)
	HPS increased by:
		1 SP = 0.163416453265045
		1 Haste rating = 0.393407106274008
		1 crit rating = 0.222301102308114
		1 int = 0.0734820592737399
"Living Ice Crystals": hits: 94, effective healing: 242925 (0.67% of player's overall)
	Total (raw) healing: 342099 (0.51% of player's overall)
	Crits: 33 (35.11%)
	Averages:
		Non-crit: 3092, crit: 4650, effective: 2584, overall: 3639
	Overheals:
		Non-crit: count: 22, total: 50019, ave: 2273
		Crit: count: 15, total: 49155, ave: 3277
	HPS increased by:
		1 SP = 0
		1 crit rating = 0.0032249519198325
		1 int = 0.00106601409380165
"Master Healthstone": hits: 10, effective healing: 58140 (0.16% of player's overall)
	Total (raw) healing: 65198 (0.10% of player's overall)
	Crits: 4 (40.00%)
	Averages:
		Non-crit: 5282, crit: 8375, effective: 5814, overall: 6519
	Overheals:
		Non-crit: count: 3, total: 7058, ave: 2352
		Crit: count: 0, total: 0, ave: 0
	HPS increased by:
		1 SP = 0
		1 crit rating = 0.00309996611728261
		1 int = 0.00102469979506004
"Riptide": hits: 1956, effective healing: 4430039 (12.28% of player's overall)
	Total (raw) healing: 6779566 (10.11% of player's overall)
	Crits: 654 (33.44%)
	Averages:
		Non-crit: 2967, crit: 4458, effective: 2264, overall: 3466
	Overheals:
		Non-crit: count: 562, total: 1171712, ave: 2084
		Crit: count: 379, total: 1177815, ave: 3107
	Ancestral Awakening: procs: 553, total: 649457, effective: 536802
	Effective healing w/ Ancestral: 4966841 (13.77% of overall)
	Total healing w/ Ancestral: 7429023 (11.07% of overall)
	HPS increased by:
		1 SP = 0.0394772494887526
		1 crit rating = 0.0566655457885702
		1 int = 0.0187309057453223
"Riptide"_hot: hits: 2738, effective healing: 2096069 (5.81% of player's overall)
	Total (raw) healing: 2394089 (3.57% of player's overall)
	Crits: 0 (0.00%)
	Averages:
		Non-crit: 874, crit: 0, effective: 765, overall: 874
	Overheals:
		Non-crit: count: 601, total: 298020, ave: 495
		Crit: count: 0, total: 0, ave: 0
	HPS increased by:
		1 SP = 0.0489111273435598
		1 crit rating = 0
		1 int = 0


Nature's Blessing adds 0.15 SP value to int.
In the longest fight 1 int worth of mana pool is worth: 0.1245 mp5


Total mana restored: 3148562
In combat mana restored 2693084
"Blessing of Sanctuary": count: 11, total mana restored: 5179 (0.16% of overall)
	In combat count: 11, mana restored: 5179 (0.19%)
	Ave: total: 470, in combat: 470
	Overall in combat mp5: 0.86
"Hymn of Hope": count: 15, total mana restored: 6789 (0.22% of overall)
	In combat count: 11, mana restored: 4965 (0.18%)
	Ave: total: 452, in combat: 451
	Overall in combat mp5: 0.83
"Judgement of Wisdom": count: 29, total mana restored: 2549 (0.08% of overall)
	In combat count: 29, mana restored: 2549 (0.09%)
	Ave: total: 87, in combat: 87
	Overall in combat mp5: 0.42
"Mana Restore": count: 189, total mana restored: 113400 (3.60% of overall)
	In combat count: 156, mana restored: 93600 (3.48%)
	Ave: total: 600, in combat: 600
	Overall in combat mp5: 15.60
"Mana Spring Totem": count: 17200, total mana restored: 731108 (23.22% of overall)
	In combat count: 9883, mana restored: 420073 (15.60%)
	Ave: total: 42, in combat: 42
	Overall in combat mp5: 70.00
"Mana Tide Totem": count: 169, total mana restored: 226063 (7.18% of overall)
	In combat count: 165, mana restored: 220671 (8.19%)
	Ave: total: 1337, in combat: 1337
	Overall in combat mp5: 36.77
	Mana Tide usage: 41.24% of combat
	MP5 increased by:
		1 int = 0.029694 mp5 (0.072000 theoretical)
"Replenishment": count: 19749, total mana restored: 1104033 (35.06% of overall)
	In combat count: 19056, mana restored: 1066294 (39.59%)
	Ave: total: 55, in combat: 55
	Overall in combat mp5: 177.68
	Replenishment uptime: 63.51%
	Mana pool based on replenishment: 22382
	MP5 increased by:
		1 int = 0.142891 mp5 (0.225000 theoretical)
"Restore Mana": count: 12, total mana restored: 51509 (1.64% of overall)
	In combat count: 12, mana restored: 51509 (1.91%)
	Ave: total: 4292, in combat: 4292
	Overall in combat mp5: 8.58
"Totemic Call": count: 657, total mana restored: 62769 (1.99% of overall)
	In combat count: 256, mana restored: 24996 (0.93%)
	Ave: total: 95, in combat: 97
	Overall in combat mp5: 4.17
"Water Shield": count: 2149, total mana restored: 845163 (26.84% of overall)
	In combat count: 2014, mana restored: 803248 (29.83%)
	Ave: total: 393, in combat: 398
	Overall in combat mp5: 133.85
	Procs from damage: 484
	MP5 from Improved Water Shield: 101.6819
	Procs from Improved Water Shield in combat: 1530
		out of 3650 procs possible, proc rate: 41.92%
	MP5 increased by:
		1 crit rating = 0.0528369399791159 mp5 
		1 int = 0.0174653526909062 mp5 


Total mp5 above character sheet (in raid): 448.76 while casting
--------------------
Healing output equivalency values:
1 SP = 1
1 mp5 = 5.0000 (configuration setting)
1 Haste rating = 1.0011
1 Crit rating = 0.9433
1 INT = 1.9471 (actual)
1 INT = 3.4468 (max theoretical)

stassart · 03/12/09, 7:36 PM

I created a lootrank for myself, but here it is if it is useful to others. I use a ratio of 1 SP to 0.2 Stamina, which is a matter of personal taste. I am still brain storming on ways to assign using empiric data, possibly processing combat logs and providing what deaths could have been avoided with how much additional stamina, and what amount of health was needed to avoid dying X% of the time. For non-tanks you may need to factor out any deaths that occurred after tank deaths (wipe). Either way that really wouldn't assign a HEP value to stamina, just whether you currently have enough or not and by how much.

[Edit by Trelis:

Removed old HEP, lootrank, gem, and enchant list because configured MP5 was too high and did not reflect reality.
]

MatsT · 03/12/09, 8:28 PM

Seems like a very nice tool, but i'm a bit curious about the HEP calculations? How do you for example calculate the value of crit? Do you factor in the increased risk of overhealing on crits? Is the haste HEP simply based on it's throughput value compared to spellpower or is mana effectiveness factored in somehow?

Also, your number for mp5 is way too high. I realize it's somewhat subjective what the correct number is and everyone is free to enter the number, but as a tip for everyone reading the thread (and yourself), if you look at the mana effectiveness you can get a kind of "upper bound" and it's a lot lower than 5.

Aanas · 03/12/09, 9:13 PM

Originally Posted by MatsT

Seems like a very nice tool, but i'm a bit curious about the HEP calculations? How do you for example calculate the value of crit? Do you factor in the increased risk of overhealing on crits? Is the haste HEP simply based on it's throughput value compared to spellpower or is mana effectiveness factored in somehow?

Also, your number for mp5 is way too high. I realize it's somewhat subjective what the correct number is and everyone is free to enter the number, but as a tip for everyone reading the thread (and yourself), if you look at the mana effectiveness you can get a kind of "upper bound" and it's a lot lower than 5.

Trelis's shaman_hep calculates the values. It calculates them on the basis of the combat log that the individual uses.

Originally Posted by stassart

I created a lootrank for myself, but here it is if it is useful to others.

The values that are presented are for the Shaman, Trelis. They are based on what he casts, as evaluated by shaman_hep's parsing of his combat logs.

This can evolve and change over time, based on his assignment, et cetera. I can imagine its usefulness, too, if one changes roles from raid to single-target, for instance. This seems to take the speculation out of the question, relying instead on evaluation.

Hamano · 03/12/09, 9:20 PM

Originally Posted by stassart

Alchemy (Flask of Pure Mojo +13 mp5 (65 HEP), Flask of the Frost Wyrm +37 SP (37 HEP)): 65/37 HEP

Why would you ever use [Flask of Pure Mojo] over [Flask of Distilled Wisdom]? It seems inferior in every single way unless u don't have 100% replenishment uptime (which will be near impossible in 3.1)

MatsT · 03/12/09, 9:54 PM

Originally Posted by Aanas

Trelis's shaman_hep calculates the values. It calculates them on the basis of the combat log that the individual uses.

The values that are presented are for the Shaman, Trelis. They are based on what he casts, as evaluated by shaman_hep's parsing of his combat logs.

This can evolve and change over time, based on his assignment, et cetera. I can imagine its usefulness, too, if one changes roles from raid to single-target, for instance. This seems to take the speculation out of the question, relying instead on evaluation.

If you had read the post, you would know that you have to manually set a value for the spellpower vs mp5 ratio, and since other stats often give a combination, it influences their value as well.

stassart · 03/12/09, 9:58 PM

Originally Posted by Hamano

Why would you ever use [Flask of Pure Mojo] over [Flask of Distilled Wisdom]? It seems inferior in every single way unless u don't have 100% replenishment uptime (which will be near impossible in 3.1)

Part of the point of writing that program was to evaluate the real value of stats, rather than making assumptions. Using real values in the current live game in my raid, Flask of Pure Mojo is quite a bit superior to Flask of Distilled Wisdom.

I am not going to guess what the values will be in 3.1. When I do my first 3.1 raid I will grab the combatlog and run the data, and see what the new values are. I expect crit to be more valuable in 3.1 and it sounds like replenishment will have a higher uptime making int closer to its theoretical than it currently is.

[EDIT BY TRELIS:

Using a calculated MP5, rather than my configured value of MP5 that was too high, Hamano is correct in that Flask of Distilled Wisdom is better than Flask of Pure Mojo.
]

stassart · 03/12/09, 10:08 PM

Originally Posted by MatsT

Seems like a very nice tool, but i'm a bit curious about the HEP calculations? How do you for example calculate the value of crit? Do you factor in the increased risk of overhealing on crits? Is the haste HEP simply based on it's throughput value compared to spellpower or is mana effectiveness factored in somehow?

I look at the average effective non-crit and the average effective crit, I assume that on average if a non-crit were to have crit its effective heal would have changed from the effective non-crit value to the effective crit-value.

For haste I only base it on its throughput value not its effect on mana, as haste does not make you less mana efficient (HPM does not change - the amount of healing output per mana), it just decreases your oom time (mps increases). I guess it depends on which you care more about healing/mana or mana/sec.

MatsT · 03/12/09, 10:37 PM

Originally Posted by stassart

For haste I only base it on its throughput value not its effect on mana, as haste does not make you less mana efficient (HPM does not change - the amount of healing output per mana), it just decreases your oom time (mps increases). I guess it depends on which you care more about healing/mana or mana/sec.

Haste does not make you less mana efficient, but spellpower does make you more mana efficient in addition to adding throughput. You have to account for this difference on one of the places, and i'm not convinced you are based on the results you are getting.

Kaytikat · 03/13/09, 6:25 AM

Originally Posted by stassart

Part of the point of writing that program was to evaluate the real value of stats, rather than making assumptions. Using real values in the current live game in my raid, Flask of Pure Mojo is quite a bit superior to Flask of Distilled Wisdom.

I'm struggling to see how this is the case when your program's output clearly shows that you've arbitrarily assigned mp5 a value of 5.0. In fact, it appears that you have simply guessed that value and are now basing your entire HEP on it. Perhaps you can provide some justification for that valuation?

The problem with mp5, of course, is that if you're not going OOM then additional mp5 has a HEP of 0, and you need to rate upgrades based upon mp5 delta between the old and new item (ignoring the "worthless" amount of extra mp5 you don't need).

Handyhoof · 03/13/09, 8:56 AM

I still recommend using WWS for empirical combat data and using that in one of the spreadsheets. It's the most used tool by raids, making the data already available without needing to log yourself. Being able to determind Active Casting % is the most important factor in how much mana you actually use per fight. If your mana pool + regen*fight duration > mana needed, any further MP5 has an HEP of 0. I like that you dove into building your own model, but it has its flaws, even as close to reality as you wanted to get.

For example: according to my last WWS I was only using mana to heal 36% of the time. At that level, my MP5 HEP is 0. If I were to crank that up to 90% because we sat 2 healers, I would run OOM halfway through the fight at that level of usage, which would drive my MP5 HEP up to 6 (vs. Spellpower at 1.58)

Daidalos · 03/13/09, 11:21 AM

Originally Posted by Handyhoof

I still recommend using WWS for empirical combat data and using that in one of the spreadsheets. It's the most used tool by raids, making the data already available without needing to log yourself. Being able to determind Active Casting % is the most important factor in how much mana you actually use per fight. If your mana pool + regen*fight duration > mana needed, any further MP5 has an HEP of 0. I like that you dove into building your own model, but it has its flaws, even as close to reality as you wanted to get.

For example: according to my last WWS I was only using mana to heal 36% of the time. At that level, my MP5 HEP is 0. If I were to crank that up to 90% because we sat 2 healers, I would run OOM halfway through the fight at that level of usage, which would drive my MP5 HEP up to 6 (vs. Spellpower at 1.58)

While I am sympathetic to the "theory is the same as reality , in theory" mindset for HEP I'm not convinced of the usefulness of Stass's approach either. I do what Handy suggests and just look at wws/recount so see what my replenishment uptime, overhealing, % of healing from what spell is and plug it into my sheet. I also use multiple HEPs sets since I think averaging them together just leads to being mediocre at everything set.

I do think the idea of a mod that would continuously analyze your combat log in specific zones that would accumulate stats sounds like a cool idea. While this is something you get a feeling for yourself after a few clears it would be nice to see numbers to back it up. From this you could create an ideal set for each zone/encounter. For example creating the ideal set for Malygos 25 man or a naxx10 KT set based on past statistics.

stassart · 03/13/09, 3:22 PM

Originally Posted by Kaytikat

I'm struggling to see how this is the case when your program's output clearly shows that you've arbitrarily assigned mp5 a value of 5.0. In fact, it appears that you have simply guessed that value and are now basing your entire HEP on it. Perhaps you can provide some justification for that valuation?

The problem with mp5, of course, is that if you're not going OOM then additional mp5 has a HEP of 0, and you need to rate upgrades based upon mp5 delta between the old and new item (ignoring the "worthless" amount of extra mp5 you don't need).

Thank you guys for the feedback.

Yes, I arbitrarily assigned mp5 a value of 5.0. You can change that value to whatever you like. It looks like in Daidalos's spreadsheet he arbitrarily assigned mp5 a value of (200%) 2.0, or do you have some justification for picking that value?

I think what I am going to do is calculate the most mana used in a fight and the length of that fight to determine what mp5 you would need for the most mana intensive fight. Then I will provide a value for int and mp5 until you reach that cap, and a separate value after you reach the needed cap (mp5 will be 0 after the cap).

I cannot think of a way to actually not split it into a before-mana-cap and after-mana-cap values for int and mp5. Although I am very open to ideas on how this could be accomplished.

stassart · 03/13/09, 3:23 PM

Originally Posted by MatsT

Haste does not make you less mana efficient, but spellpower does make you more mana efficient in addition to adding throughput. You have to account for this difference on one of the places, and i'm not convinced you are based on the results you are getting.

This is a valid point, I will work on providing two numbers for haste, the current one and one that reduces it by the impact it has on mps based on the weight set for mp5.

stassart · 03/13/09, 3:47 PM

Originally Posted by stassart

Thank you guys for the feedback.

Yes, I arbitrarily assigned mp5 a value of 5.0. You can change that value to whatever you like. It looks like in Daidalos's spreadsheet he arbitrarily assigned mp5 a value of (200%) 2.0, or do you have some justification for picking that value?

I think what I am going to do is calculate the most mana used in a fight and the length of that fight to determine what mp5 you would need for the most mana intensive fight. Then I will provide a value for int and mp5 until you reach that cap, and a separate value after you reach the needed cap (mp5 will be 0 after the cap).

I cannot think of a way to actually not split it into a before-mana-cap and after-mana-cap values for int and mp5. Although I am very open to ideas on how this could be accomplished.

Actually, thinking on it further. I could have the user provide a spell power value to use to calculate a ratio. I know the user's mana pool based on Replenishment in the combatlog, I can calculate the most mana needed in a fight. If the user provides a spell power value, I could calculate the ratio of spell power to mp5 needed to hit the amount of mana needed for the most mana intensive fight.

I will need to provide an option to also look at mana drain effects as it is quite likely that one of the mana drain fights is the most mana intensive. But some users may wish to ignore the effect if they get hit by a ton of Mana Detonations on Kel'thuzad or something.

I think I like that approach better than providing two numbers.

Daidalos · 03/13/09, 4:15 PM

Originally Posted by stassart

Thank you guys for the feedback.

Yes, I arbitrarily assigned mp5 a value of 5.0. You can change that value to whatever you like. It looks like in Daidalos's spreadsheet he arbitrarily assigned mp5 a value of (200%) 2.0, or do you have some justification for picking that value?

I think what I am going to do is calculate the most mana used in a fight and the length of that fight to determine what mp5 you would need for the most mana intensive fight. Then I will provide a value for int and mp5 until you reach that cap, and a separate value after you reach the needed cap (mp5 will be 0 after the cap).

I cannot think of a way to actually not split it into a before-mana-cap and after-mana-cap values for int and mp5. Although I am very open to ideas on how this could be accomplished.

That was my personal weight I was using for whatever I was doing. Its a variable (everything in blue is a user input) This arbitrary and will vary depending on many things. This is why its an input. Mp5 has no direct correlation to Spell power so mp5->spell power conversion is going to based on your needs. You can argue the total healing done in a certain amount of time can determine weights but this is going to be just as arbitrary. Also this is why I personally make different sets.

Kaytikat · 03/13/09, 6:33 PM

The more I think about this the more I want to leave behind the entire concept of a HEP for mp5 (and regeneration as a whole).

Mana regeneration exists to serve one purpose only: to make sure you have enough mana to cast the spells you need to cast. So long as that criteria is met the HEP of *all* regeneration is 0.

So, to value regen we simply need to know the amount of regeneration required to make Time to OOM > Fight Length.

To calculate Time to OOM you'd need to know the amount of static mp5 present on a player's gear, as well as which buffs were active, but the rest (such as mana pool size and mp5 returned by replenishment) are directly available in the combat log.

Ideally, you could then present a series of fight lengths with the amount of total mp5 from all sources (gear, buffs, replenishment, etc) required to cast the same pattern of casts as found in that combat log. You could then create appropriate Pawn scales for each fight length of 1mp5 = 1, 1 int = x to see easily the amount of mp5 present on an item and whether you have a high enough amount of mp5 to last for the anticipated fight length.

For example (numbers entirely made up):

Fight mana consumption: 834mp5
Required regen (based upon configurable margin of error): 950mp5
Required regen from gear: 325mp5  (this is required regen minus replenishment, buffs and static mp5 from mana pool)

10min fight: 1 int = 0.234mp5
5min fight: 1 int = 0.445mp5
1min fight: 1 int = 0.7mp5

Originally Posted by Handyhoof

At that level, my MP5 HEP is 0. If I were to crank that up to 90% because we sat 2 healers, I would run OOM halfway through the fight at that level of usage, which would drive my MP5 HEP up to 6 (vs. Spellpower at 1.58)

Actually, I'd contend that when you have enough regen the HEP of mp5 is 0 and when you don't the HEP is infinity. And this is the big problem with trying to produce a HEP for mp5!

Think about it. If you run OOM halfway through and you go out and add enough mp5 to get to 75% of the fight, you're still going OOM and wiping the raid! Think back to pre-nerf M'uru where many restos were actually spamming CH1 (at a loss to HPS equivalent of about several hundred +healing) to ensure they increased the Time to OOM enough to survive the whole fight.

hvidgaard · 03/17/09, 1:15 PM

It was proposed earlier and I still think, for the reasons mentioned above this post, the idea of ManaPoints and HealingPoints have some strong merit. Trying to include regen stats in a HEP scale is, IMO, doing it wrong - once you don't run OOM regen stats loose all value.

If you take ManaPoints to be the amount of mana you get from the gear in a 5min (perhaps 10min is better to avoid the inflation from the initial mana?) fight, and HealingPoints being the normalized values of all the stats that increase the healing done. Haste will be an obvious problem tho, as it will increase the burst potential, but it will not actually increase the healing done with a finite amount of mana. So haste can improve your healing done, if you increase your total manapool.

The idea with those 2 different scales is that you have a estimate of ManaPoints needed to get trough a fight based on the length of the fight, tanks gear and amount of raid damage, and then get as many healing points needed.

Daidalos · 03/17/09, 1:53 PM

Originally Posted by hvidgaard

It was proposed earlier and I still think, for the reasons mentioned above this post, the idea of ManaPoints and HealingPoints have some strong merit. Trying to include regen stats in a HEP scale is, IMO, doing it wrong - once you don't run OOM regen stats loose all value.

If you take ManaPoints to be the amount of mana you get from the gear in a 5min (perhaps 10min is better to avoid the inflation from the initial mana?) fight, and HealingPoints being the normalized values of all the stats that increase the healing done. Haste will be an obvious problem tho, as it will increase the burst potential, but it will not actually increase the healing done with a finite amount of mana. So haste can improve your healing done, if you increase your total manapool.

The idea with those 2 different scales is that you have a estimate of ManaPoints needed to get trough a fight based on the length of the fight, tanks gear and amount of raid damage, and then get as many healing points needed.

There is never going to be an agreed apoun weight of HEP vs MEP (not without getting downranking back.. :-( ). You are of course free to assign your own values, but I highly doubt you will be to convince everyone that your weight will be universal in any regard. I allow HEP to MEP conversion in my sheet for the simple reason of selecting your ideal gear. Obviously if your HEP to MEP ratio is very high you will basically just be getting caster dps gear. I am already getting to the point of wearing some ele gear on some fights (aside form 4 pct7 for the resto bonus) simply due to never being in danger of going oom, but I still look at regen sets since I am not expecting Uldar to be the loot pinata naxx is.

Sprout · 03/17/09, 3:20 PM

Originally Posted by Daidalos

There is never going to be an agreed apoun weight of HEP vs MEP (not without getting downranking back.. :-( ). You are of course free to assign your own values, but I highly doubt you will be to convince everyone that your weight will be universal in any regard. I allow HEP to MEP conversion in my sheet for the simple reason of selecting your ideal gear. Obviously if your HEP to MEP ratio is very high you will basically just be getting caster dps gear. I am already getting to the point of wearing some ele gear on some fights (aside form 4 pct7 for the resto bonus) simply due to never being in danger of going oom, but I still look at regen sets since I am not expecting Uldar to be the loot pinata naxx is.

I would disagree with you a little on your conclusions. With current gear sets, the i213/i226 BiS items have been normalized to fall into 3 catagories

Crit/mp5
Haste/MP5
Crit/Haste

All other stats are exactly the same or very very close, including spellpower. You can argue gems/enchants are different as well, but most of those fall into the "gem for X or spell power’ argument where X is Crit, haste or Int (since int is better than MP5). Effectively, the same 3 stats. Plus as a general rule its alot easier to swap gear than gems/enchants.

If we assume this to be true, you can then use Stassart's tool to create optimal gear sets per fight. The key would be to use the tool to set the baseline mana regen requirements. Once you have that you can then gear to that level, and go with crit/mp5 and haste/mp5 on gear until you reach the baseline Then you can go crit/haste pieces or maybe SP with a trinket.

One small note on this is as the raid geared up and learned the fight the mana requirements would inevitably fall and you would need to create a new baseline.

I know that in our heads we all do this right now to some extent or another (or at least did in Sunwell). But as someone posted above, if we can use this tool to give us a more quantitative number set to work with then we will be that much better.

One last thing I am very excited about with this tool is getting quantitative data on CH bounces and over heal. It often takes me a long time to evaluate this on a new fight since I am learning the fight at the same time and not focused on watching the bounce pattern/numbers.

hvidgaard · 03/18/09, 7:44 AM

Originally Posted by Daidalos

There is never going to be an agreed apoun weight of HEP vs MEP (not without getting downranking back.. :-( ). You are of course free to assign your own values, but I highly doubt you will be to convince everyone that your weight will be universal in any regard. I allow HEP to MEP conversion in my sheet for the simple reason of selecting your ideal gear. Obviously if your HEP to MEP ratio is very high you will basically just be getting caster dps gear. I am already getting to the point of wearing some ele gear on some fights (aside form 4 pct7 for the resto bonus) simply due to never being in danger of going oom, but I still look at regen sets since I am not expecting Uldar to be the loot pinata naxx is.

I think you missed my point. By having different scales for HEP and MEP you can fine tune your gear for a given fight. MEP enough (obviously depends on the amount of incoming damage and length of the fight) and then as much HEP as you can pack. They are not meant to be exclusive to each other, but being 2 different scales that complement eachother.

MatsT · 03/18/09, 12:00 PM

There is no set value of mana that you "need" for any given fight. Even if you are going out of mana halfway through the fight, 100 spellpower is still better than 10 mp5. In most cases you can simplify the calculations and get some numbers by assuming your spells had a 0 second casting time. Looking at how much more healing 1 spellpower and 1 mp5 gives over a 5 minute period provided you are not limited by the amount of spells you have time to cast puts a cap on the effectiveness of mp5.

If you are not limited by mana at all, mp5 has a HEP of 0. If you are limited by mana, then it has the value of this cap. In reality it's not this simple since the gear you choose will also affect how much you are limited by mana. What i do is put the HEP of mp5 to the cap, and then for fights where i know mana isn't a problem i replace some obvious slots with items that gives throughput at a reasonable mana cost.

Daidalos · 03/18/09, 1:12 PM

Originally Posted by hvidgaard

I think you missed my point. By having different scales for HEP and MEP you can fine tune your gear for a given fight. MEP enough (obviously depends on the amount of incoming damage and length of the fight) and then as much HEP as you can pack. They are not meant to be exclusive to each other, but being 2 different scales that complement eachother.

I'm not sure what you mean by fine tune. I agree with MatsT points and in addition changes in dps, strategy, or people just playing smarter will drastically change your "needed regen". I'm not disagreeing that after fight analysis is good I am just not buying that a combat log suddenly allows us to know exactly anything (other than what happened last time). I do think trending across multiple weeks of the same encounter may lead to observations that weren't entirely obvious I'm just not convinced that log parsing is superior to paying attention and thinking about a fight, but I'm certainly willing to be wrong about this.

MatsT · 03/18/09, 1:30 PM

I do think that combat log parsing may be very useful, mainly concerning overhealing. For example spellpower will only increase your healing on heals that don't overheal, and you can determine crit ratings HEP more precisely if you know with what frequency and amount crits will overheal compared to noncrits. Chain Heal jumps is another thing that parsing could give a lot of useful information regarding.