New Updates: 1. Added Heart of Narikor to both dropdowns with the other 'extra' augs to pick from. 2. Added Mana Charge as an ADPS option. I still need to do a little testing in game to make sure it's working like I have setup. It's just hard considering the long reuse timer. But right now it will be trigger by pretty much everything. So, if you turn on all the equipment augs don't be surprised when all 3 charges get used up in one cast. 3. Changed how Claw refresh works. See what you guys think of the results. It's possible that it's bad in an all new different ways but so far I think it's working out. What I did, after thinking about it some more, is go with an EQ style solution. After all the posts on aug proc formulas I went with a similar approach based on recast time to reduce how often to apply the benefit for spells with shorter timers. I'm doing a simple Math.ceil(10 / recastTime) to get a number of 1 for long recast spells and as high as 7 when it's cloudburst. Then I'll apply the benefit at those intervals. So, Dicho gets 6% every 1 claw cast while cloudburst gets 6% * 7 every 7 casts. Repeating the tests I did earlier I got the same results for the good cases and the bad case now gets 10 Dicho cast which I think is fair.
Ok cool. I think that makes sense and my crazy long parses are coming pretty close the the value in the description at base 100 so I think it's good enough. Thanks for looking into it.
Are you calculating it on an item with a proc mod? And have you factored in Twinproc (or are doing it on Test without Twinproc?) [24351] Sympathetic Force of Magic IX Target: Self Resist: Beneficial, Blockable: Yes Focusable: Yes Casting: 0s Duration: 3.3h (1950 ticks), Extendable: Yes, Dispelable: Yes 1: SPA 383 Base1=100 Base2=23718 Max=0 Calc=100 --- Cast: Force of Magic IX on Spell Use (Base1=100) e.g. Cast Time 2s=15% 3s=20.0% 4s=30.1% 5s=40.1% 2: SPA 138 Base1=0 Base2=0 Max=0 Calc=100 --- Limit Type: Detrimental 3: SPA 348 Base1=10 Base2=0 Max=0 Calc=100 --- Limit Min Mana Cost: 10 4: SPA 142 Base1=70 Base2=0 Max=0 Calc=100 --- Limit Min Level: 70 5: SPA 311 Base1=0 Base2=0 Max=0 Calc=100 --- Limit Type: Exclude Combat Skills 6: SPA 137 Base1=0 Base2=0 Max=0 Calc=100 --- Limit Effect: Current HP Has a chance to strike your opponent with a magic attack each time you cast a damaging spell. I think that should be correct. It doesn't account for proc mods from items or Twinproc of course.
I've never been able to get the Base * Base1 / 10 * Normalizer * Item Proc Mod formula to align with the results I've seen in game, even prior to the changes to make procs fire independently (also prior to casters having twinproc). The proc rates have always been pretty significantly higher. For example, here's a parse from before the proc changes with Ancient Hedgewizard's Brew: Combat Dummy Beza on 6/30/2015 Sancus - 10990 --- Coronal Rain Rk. II - 2029 --- Rain of Cutlasses Rk. II - 2029 --- Remorseless Servant Rk. II - 2214 --- Spear of Molten Shieldstone Rk. II - 2027 --- Storm of Many Rk. II - 2691 Produced by GamParse v1.5.0.2 Accounting for twincasts, I had 11345 casts, and 4266 procs, which is a 37.6% proc rate. Ancient Hedgewizard's Brew uses SPA 383 just like Arcane Fusion, so it should work the same. The spell that would have the highest proc rate would be one of the rains. Using the formula, it should be: 0.06 * 100 / 10 * (0.167 * (4 - 1)) = 30.06%, which is significantly lower than the proc rate the weave had, which included three spells with lower cast times than 4s. For weapon procs prior to the proc changes, each individual proc had a decreased chance to proc as you added procs. With that rotation I had a 24-25% chance to proc each of my three procs (all base1 = 100, Magna Fulgia for a weapon). With two procs I had a ~29-30% chance to proc each proc, and with one proc a ~37% chance to proc. After the proc changes, each proc acted as if it were the only proc, meaning they all had ~37% chance to proc with that weave regardless of the number of procs I had.
@Sancus It's been many years, (even that post I quoted, which was itself two years ago, was a couple years after the tests I ran), but iirc I did 1 spell / 1 proc parses, without AAs on the Beta server, and while it didn't perfectly line up to the formula, it was fairly close, certainly not off by more than a couple percent. I recall wondering if it was off due to the effects of heroics, and I /think/ I was going to run a new test with no gear to confirm it, but then I don't remember if I did or not. I think beta ended and I got busy with live. If heroics do effect spell proc rates like they do melee proc rates that would go a long way towards explaining why they didn't match up, and why it may have gotten even more whacked lately with the huge increases in heroics. If they don't, then there is something else missing from it, or the information I was given was wrong / incomplete.
Well, I was testing with Sympathetic Force of Corruption X. It should be the same as Force of Magic right? I was on the Test server and only had Runed Belt of Boromas equiped and wrists for some extra mana. I don't think I have anything that would change proc rates. I used robe clickies for both wiz and enc for mana. Reset AAs and didn't buy anything really. Just the ones the ones that add mana and mana regen. After 3500 casts of Skyblaze and Rimeblast I ended up with like a 44.78% proc rate. Which is at least in the ballpark of the description. I still read it as 4.592% base10 for a spell that takes 3.75s to cast. I'm running a test right now with cascade/narendi's to see how that works. So, yeah for base100 I'm going with 45.92% at the moment. I could obviously be very wrong but it does seem to be basically the same formula that you posted it's just maybe they tweaked it a bit when they decided to let the proc rates not depend on each other. The modifier that works out to 0.167 still makes sense and the cast rate - 1 plus base1/100 works out ok. Maybe the 0.06 mod was just dropped. I don't know. The next thing I'll try is multiple procs but I did test that a little a few months ago and I'm pretty sure I saw the same thing that Sancus was seeing. The number of procs didn't seem to effect the rate anymore. But it can't hurt to test some more. For me, it doesn't have to be perfect but as long as the items/spells in the app itself have reasonable rates compared to what I'm seeing it's good enough for me. Worst case I'll just go be some small modification based on parses. At least for the wizard version of the app it's not like I have to account for many variations and it's easy enough to compare to in game results.
Runed Belt doesn't have a proc mod, and all the mainline sympathetics have a base1 of 100 as far as I know. You guys are going to make me parse this crap again.
lol sorry Edit - Actually, you might want to wait for this Narendi/Cascade test. It might be different. Edit- Edit- I hate this game. I thought it was doing around 21% but after watching it for a while I noticed I kept getting 'you must select a target for your spell'. If I count those it goes up to around 35%. But I did have to restart the test. Earlier I got disconnected and didn't notice auto-buy was on when I came back in. So, I started over and moved to Test dummies instead of guild hall dummies. Then this error started coming up. Forgot all about it this being an issue. bleh. Guess I'll move back to a GH.
Which this is actually pretty interesting in itself. If the proc rate works out like it used to on a Test dummy if you exclude the 'you need to select a target' errors. Then maybe whatever is causing that bug is related to the change. There could be something they broke that makes procs go off more than they should and we're seeing a symptom of the problem on Test dummies. Or that their 'fix' to make procs work independently is to do the old calculation, check if the proc succeeds or not, if it fails then do a new calculation on top of that for like a 2nd chance to proc. That then breaks on Test dummies but works everywhere else giving us the higher rate we're seeing. Might be able to guess at what it's doing with some more testing.
You're counting wrong. The "You must first select a target for this spell" error from test dummies comes from when it tries to heal you (Replenish On) but you are already full HP/Mana. I ran a 5,000 cast parse of Cloudburst/Hedgewizard Proc last night without any gear or AAs and got the following: 4,978 successful casts 22 Fizzles 758 Procs 758 / 4978 = 0.1523 or a 15.23% proc rate. The formula estimates .06 * 100 / 10 * .25 = .15 = 15% proc rate so my observed rate was just ever so slightly above this. Now it is close, very close, and since it is a random effect we would expect it to be off and not dead on, however generally speaking things in EQ tend to estimate just -under- what they should, not -over- like this is. IIRC this was the issue I could never reconcile with procs in the past. They always come out just a hair -above- whatever the math says they should, whereas generally percentage-based effects in EQ average out just a hair -lower- than whatever the math says they should. Granted it could be nothing at all and it might just need a longer parse, (5,000 casts is enough to show the formula isn't wildly wrong, but it isn't going to be like 6 sigma accuracy or something), but it could also show there is something (small) off about the formula, or perhaps there is a small + to proc chance. It wouldn't be heroics in this instance since I had no gear on. I'm about to start a 5k of Skyblaze. That's going to take about 12.5 hours. That will give more information for comparison.
I'm counting the same way when using Test dummies vs dummies in the GH and the numbers are different in the GH if that really is just the replenish message. I got 33.45% after around 1500 casts vs 20% on test dummies for Narendi/Rimeblast Cascade. Nothing else changed.
It's really not the replenish message. Was just doing a short test. If I take the belt off and have zero procs the error message message goes away. Whether I have full mana or half mana when replenish hits you never see an error. I did like 40 or so casts. Not a ton but still. Then when I put the belt on I start seeing the error pop up almost immediately. Sometimes I have full mana and I get the error and sometimes I have well less than full mana and the error still shows up. It seems like maybe there is a conflict between refresh and procs. Maybe you can't proc at the same time that a refresh goes off. I wouldn't use Test dummies for this.
My guess is that the proc check for replenish and the damage aug both succeed at the same time. It then executes the procs with replenish going first. Replenish changes the current target to heal you and doesn't or isn't able to change the current target back before the damage code executes. It then executes and fails. The error message would then actually make sense.
This is more fun than playing the game! I don't want to ignore a hint. It's no coincidence that the proc rate works out like the old formula suggests if you discount the replenish error. It really does act like it calls the old code and if it succeeds then everything works like it used to and there's no conflicts. Then they do something in addtion. When comparing the ethereals tests I did with the crappy spell test both seem to proc around 1.66 times more often than the old formula. Maybe they just through the 0.167 multiplier back on top of everything. hmm
Alright, I think I know what it is. You're right about the Replenish/Target issue interfering with procs for some reason, my mistake, though I really don't know why it would. It's a single target spell that the NPC casts, so I don't see how it would affect the players' targetting in any way. Turning it off /does/ prevent it from interfering, which is why I said to do so. It is off for me right now as I'm running Skyblaze/Potion, (another 3ish hours to go) but I went to double check my Cloudburst parse and it was on for that, meaning my results /were/ off. Using that, I went back and recalculated which gave me ~25.13% for Cloudburst, (I say ~ because I won't say something where a bug is involved is entirely trustworthy) and though Skyblaze isn't finished, at the time I took a look, it was currently 1401 / 3069 = 45.65%. With that in mind, I went back and tweaked the formula slightly and I believe I may have found values that are correct. It does require more testing (with both a third spell cast time AND with something that uses other than 100 for Base1).
Cool. That number is pretty close to what I've been seeing too. Sorry you had to jump back into boring parsing. I haven't verified Cloudburst or anything with a different multiplier yet but I'll see if I can look at that soon too. I'm hoping we're going to see 25%. Just the straight multiplier value. Which would mean what I'm currently doing in the app doesn't have to change just yet. Which is just this: base1 / 100 * multiplier Now when we get to enc dicho it could fall apart. I think that listed a base1 of 160. Looking at the base 16% chance numbers in the description give me some hope that it will hold up. Skyblaze would come to like 73% or so in that case. It's going to be fun testing that one with only a handful of procs every minute...
You guys rock, fantastic work Kizant. This community is awesome. If and when you make the Mage version, we will be immensly grateful.
Ench Dicho currently eats an additional charge every time its proc fires. So the 16 counters it provides is really only 8 (Dicho strike has a really high proc rate). Something to keep in mind when you add that. Posting it here in case someone has an explanation for why that's happening, because it's obnoxious. [49271/6672] Dichotomic Reinforcement 6 Max Hits: 16 Matching Spells 1: Cast: Dichotomic Reinforced Strike 6 on Spell Use (Base1=16) e.g. Cast Time 2s=4% 3s=5.3% 4s=8.0% 5s=10.7% 2: Limit Max Level: 110 (lose 100% per level) 3: Limit Effect: Current HP 4: Limit Type: Detrimental 5: Limit Spells: Exclude Group - Dichotomic Reinforced Strike 6: Limit Min Mana Cost: 100 7: Increase Current Mana by 2811 8: Increase Base Spell Damage by 3959 9: Limit Max Level: 110 (lose 100% per level) 10: Limit Effect: Current HP 11: Limit Type: Detrimental 12: Limit Min Mana Cost: 100 [49277/6673] Dichotomic Reinforced Strike 6 Target: Single Range: 300' Resist: Lowest -25 Mana: 0 Focusable: Yes Reflectable: No Casting: 0s, Recast: 6s 1: Decrease Current HP by 21944 2: Decrease Current Hate by 12%
So I did some testing too, all of which aligns pretty strongly with everyone else's results. First, an Ethereal test on my Wizard (yeah, kinda old news at this point): Combat Dummy Azia on 3/24/2017 Wizard - 4884 --- Ethereal Rimeblast Rk. III - 2442 --- Ethereal Skyblaze Rk. III - 2442 Produced by GamParse v1.5.2.3 4884 spells cast, 23 fizzles, and 248 twincasts. I had Ancient Hedgewizard Brew, Runed Belt of Boromas, Fleshburner of Boromas, Tome of Obulous, and Living Golem Heart. Factoring in twinprocs, their proc rates were 45.52%, 46.22%, 45.54%, 45.95%, and 45.73%, and they average to a 45.79% proc rate. If we're assuming the formula is now Base1/100*(.167*(Cast-1)), that's well within the margin of error of a 45.92% proc rate. Next, I did a shorter test with Storm (well actually longer, but fewer casts), just to confirm it was around 25%: Combat Dummy Beza on 3/24/2017 Sancus - 1053 --- Storm of Many Rk. III - 1053 Produced by GamParse v1.5.2.3 1053 casts, 4 fizzles, 43 twincasts. I had Nightfear's Halo, Bone Shards of Frozen Marrow, Runed Belt of Boromas, Sodkee's Sympathetic Stone, Ancient Diamond Spellcharm, Ancient Hedgwizard Brew, and the Sacred Prayer Shawl of the Duke (which has Base1=80). The Base1=100 procs had proc rates of 24.91%, 25.18%, 22.52%, 23.54%, 25.73%, and 25.09% respectively. There's some variation there because this test has fewer casts, but they average out to 24.53%. The Brell's aug had a proc rate of 18.22%, which is a bit below the 20% projected by the new formula but is well within the margin of error. I probably should've run a test with the sands line or something for more data points, but I was curious about Storm specifically. Finally, I did a test chaining 3 spears: Combat Dummy Beza on 3/24/2017 Sancus - 10017 --- Spear of Blistersteel Rk. III - 3339 --- Spear of Molten Shieldstone Rk. III - 3339 --- Spear of Molten Steel Rk. II - 3339 Produced by GamParse v1.5.2.3 10017 casts, 18 fizzles, 530 twincasts. I had the same augs/procs as before. The Base1=100 procs (in same order as before) had proc rates of 41.46%, 42.45%, 41.92%, 42.46%, 41.22%, and 42.16%. That averages to 41.95%, which is very close to the 41.75% the formula would suggest. The Brell's aug had a proc rate of 32.94% with 3468 procs, which is very close to the 33.4% proc rate the forumla suggests (a bit off, but within the MoE). All of that aligns pretty well with the formula kizant suggested (Base1/100 * Normalizer * Item Proc Mod, or however you want to write it). Also regarding the Replenish bug, it becomes even more apparent if you try to use the Mage RS spell on dummies. If the replenish cast lines up with your RS cast, you'll get the message about lacking a target and the actual pet won't spawn. It's always been a rather weird bug.
