FFB - crit or haste ?

[underdogba]

Crit% is nota stat that dictates how many crits you will recieve in 100 casts... but simply what CHANCE each cast has to crit.

No, that is not true at all. Statistically those two things really are the same due to the Expected Value formula. Knowing the % chance each cast has to crit does in fact tell you on average how many crits you will receive in 100 casts.

Edit: If you want to argue otherwise, you're arguing against the validity of the Expected Value formula and against the Law of Large Numbers. Frankly that's a kooky argument. Comparing drop%'s to crit%'s is in principle the same, however crit%'s are subject to a much larger statistical sample size (i.e., # of casts), so will you naturally tend to see statistical distributions more in line with those predicted by crit%.

[choke]

This argument about FFB/fire spec seems to come up a lot, and while on the outside it would appear haste comes out on top, it actually doesn't. The reason is one talent, "Hot Streak", where 2 crits equal instant pyro blasts, which is obviously a massive amount of dps. If you're FFB/fire, stack as much crit as you possibly can, but don't completely ignore haste.

[AuthurDent]

It does mean that in 100 runs, it will drop. It does NOT mean you will be in that run, thus your view will be skewed(OMG, 220 runs and still no mount QQQQQQQQQQQ).

Wrong... there is a DISTINCT difference in a calculated percentage of drop rate, and an assigned chance to drop.

If there were 1000 runs and it dropped 10 times... it dropped 1% of the time.... that does NOT mean the item's %chance to drop was set to1%

If it WAS set to 1%... then EACH time you killed the boss... you hade 99 chances it would NOT drop... so you cannot say, "Well... next time I only have a 98% chance for it to NOT drop". Running it 100 times does not guarantee it will drop....

at 1%... you could EASILY* run it 300 times and never see it drop... simply because it's RNG not "averaged"...

This is exactly why the FFB build can vary so much in damage out/DPS... just because we "HAVE" 50% fire crit, doesn't mean we GET 50%... we sometimes get 68% and sometimes 42%.

[frugm]

kthxbye

I know this is off topic. However I do feel it's worth mentioning.

People writing kthxbye, No, It doesn't make you cool or an elitist. It Just makes you look like a retard. Please stop using it, Every time I see it, It just makes me want to punch a baby, Come on guys, Think of the babies.

[Nebbylol]

IF you have any and all dps questions just consult this website. www.maxdps.com

Crit is slightly better if you are wearing mostly best in slot items. I have raid buffed (no boomkin or eleshaman) 35% crit. Ignite does the second amount of dmg only lower to FFB. Haste is good if you are frost. Crit is good if you are fire/FFB.

[Mats]

I agree. His post proves nothing, except that he wants to make himself look better than the rest of us. (EDIT: This was a respond to frugm)

Other than that this thread has gone incredibly long (to be this forum) without people tossing everything else into the discussion. And looking at theorycrafting is always fun.

[AuthurDent]

A boss encounter and the resulting number of casts that initiate the "chance" calculations hardly falls under the "Law of Large Numbers"... while it's true that the larger the sample pool, the more accurate the average, In the numbers we are using in a boss encounter, RNG is much less reliable for an average.

I can easily have a 10% crit chance and get 2 crits out of 4 casts... wow I have a 50% crit chance?... I could also get NO crits in the same 4 casts and seemingly have 0% crit chance.

Supprting my argument is the fact that you can check Recount's logs and see that in some encounter with similar casts... say 70... you may see as much as 20% variance in crits over a few fights.

No, that is not true at all. Statistically those two things really are the same due to the Expected Value formula. Knowing the % chance each cast has to crit does in fact tell you on average how many crits you will receive in 100 casts.

Edit: If you want to argue otherwise, you're arguing against the validity of the Expected Value formula and against the Law of Large Numbers. Frankly that's a kooky argument. Comparing drop%'s to crit%'s is in principle the same, however crit%'s are subject to a much larger statistical sample size (i.e., # of casts), so will you naturally tend to see statistical distributions more in line with those predicted by crit%.

[Dokoran]

I tend to find crit more appealing, because it boosts ur dps almost as mutch as haste. But it also increase ur manareg wich i find is a problem in raids if I don't have a replinishment in my raidgroup. That is if you are using molten armor, with mage armor the manareg is ofcourse no problems, but 3% crit is allways 3%crit.

5% crit with molten armor actually. Unless you're complete fail and don't have the molten armor glyph....

[Hexxidecimal]

It does mean that in 100 runs, it will drop. It does NOT mean you will be in that run, thus your view will be skewed(OMG, 220 runs and still no mount QQQQQQQQQQQ).

Not true neccsarily, unless of course the 1% is programmed to happen one in every 100 runs on purpose.

If it is a one percent drop rate, that means, as far as I know, that every kill that mob has a 1% chance to drop that item, and it's quite possible for 100 runs to happen and that 1% never come up.

What you are thinking of is a fraction more than a %. If it drops without question every 1 out of a hundrend time it's drop rate is 1/100 of it's chance to NOT drop.

% are by nature are pretty chaotic and never a gaurntee, they are just the odds. What they generally do is is take the average and break it down for us, but are not really a mirrior of reality. In reality with a 1% drop rate it is COMPLETELY possible to get the item two times in a row than not again for another 198 runs. The average works out to be 1%.

That isn't to say they aren't accurate, they just aren't accurate on a per case basis, but over a long string of cases they are.

Like I could have 10% crit, which means by your figurings (assuming I am understanding you correctly) I shouldn't be able to in 10 casts get 2 crits off, nor should I be able to in 10 casts not get at least 1 crit off. You sometimes won't see the average that the 10% represents until you have a large enough pool of casts.

[Vileeh]

No, that is not true at all. Statistically those two things really are the same due to the Expected Value formula. Knowing the % chance each cast has to crit does in fact tell you on average how many crits you will receive in 100 casts.

Edit: If you want to argue otherwise, you're arguing against the validity of the Expected Value formula and against the Law of Large Numbers. Frankly that's a kooky argument. Comparing drop%'s to crit%'s is in principle the same, however crit%'s are subject to a much larger statistical sample size (i.e., # of casts), so will you naturally tend to see statistical distributions more in line with those predicted by crit%.

So, if I had a deck of cards, and I want you to pull out the ace of clubs. If you pull 52 times, and still not get the ace of clubs the world will stop to exist?

Or maybe, if you make a pan with spaghetti, and one of the spaghetti's is red, the rest white, 100 in total. After 100 times of grabbing one out and putting it back in you MUST have had the red one once?

GET
REAL
BRUV

[Hexxidecimal]

So, if I had a deck of cards, and I want you to pull out the ace of clubs. If you pull 52 times, and still not get the ace of clubs the world will stop to exist?

Or maybe, if you make a pan with spaghetti, and one of the spaghetti's is red, the rest white, 100 in total. After 100 times of grabbing one out and putting it back in you MUST have had the red one once?

GET
REAL
BRUV

Not really sure what your stance here is with your comments so I’ll make an assumption, sorry if I am wrong.

The analogies you are making are flawed if you are trying to compare them to drops rates or crit rates.. Drop rates from mobs or even crit rates are in no way like a deck of cards where there is an inevitable eventuality within a range of topped off numbers.

The first problem is it’s not like a boss has a certain number of other loot to be passed out and once those are gone they are gone. You can not increase your chances of a drop by eliminating other loots choices from the list. The list of possible loot drops do not in reality decay.


There is only the average. The average is your best indicator for accuracy but is not perfect. The word “average” itself insinuates that inconsistencies will happen.

[SBS]

So, if I had a deck of cards, and I want you to pull out the ace of clubs. If you pull 52 times, and still not get the ace of clubs the world will stop to exist?

Or maybe, if you make a pan with spaghetti, and one of the spaghetti's is red, the rest white, 100 in total. After 100 times of grabbing one out and putting it back in you MUST have had the red one once?

GET
REAL
BRUV

I don't know if you're joking or not, but I'll answer anyways.

Dropchance can not be equated with a deck of cards. The loot is not pre-programmed so that a random horsie hides in 1 of 100 sacks of loot.

I'll give you an example of how it does work though:
You kill the boss -> the game chooses a random number from 1 - 100 -> if you get 92, horsie drops. The dice does _not_ get smaller every time and your different baron-runs or w/e is not linked together with a shared pool of loot.

[peacemaker]

Please don't believe anything posted in this thread unless you find it elsewhere. Most posters completely fail to grasp the basic concepts of statistics and are therefore unable to answer a question rooted in mathematics.

Perform a Markovian analysis on a specific FrostFire rotation, then come back and try to tell me haste is better than crit. You won't.

[aikouka]

Also, people... what they're trying to say is that over a period of time, with a percent chance of occurrence, eventually the limit will converge upon the x%. Yes, if I have 50% chance to crit and cast 100 spells, chances are I wouldn't see a perfect 50 crits... heck, I could see all 100 crit! That's not what they're trying to say. The idea is that as you approach the limit (infinity), the value becomes closer and closer to the percent chance. But please note, it doesn't take until some crazy number of occurrences to get close to the value in question.

BUT haste also can increase dpm.

No. Your Damage Per Mana (DPM) will never change with haste. Frostfire (untalented) is 14% of base mana and if you cast five of them, you have spent 60% of your base mana. If each Frostfire deals 5000 damage and I do this in 15 seconds or 12.5 seconds, nothing has truly changed. I still dealt 25,000 damage for the cost of 60% of my base mana. The difference is that my damage per second has gone up as 25,000/12.5 > 25,000/15.0.

But the Possible Crits per time (5 minutes) is actually increased, thus increasing your DPM

I think at this point you're truly reaching to prove that haste affects damage per mana and the argument is a bit silly. Yes, we all know critical strike chance affects damage per mana. What you're saying is, "Hey look... haste gives you more casts over a period of time in which they could crit... if that one extra spell cast you get crits, you get more damage for that amount of mana you spent!" See, the thing is... it wasn't the haste that truly increased your damage per mana... it was the critical strike of the spell. If you had 0% critical strike rating for spells, your example would fall flat on its face (and it's not a plausible example because of base intellect anyway :P).

THUS: Haste does not change your DPM - but if you have enough mana to sustain casting all 5 minutes, your DPM will be increased through those 10 more crits and the mana you gain through that skill which gives you mana for crits with the name I keep forgetting.

Crit vs. Haste dpm-wise, crit wins - because crit increases the dpm more.

But haste won't lower you DPM as many want you to believe. Only the MPT (mana spent per time) is affected.

No one is saying haste lowers your damage per mana. Your damage per mana does not change if you have 0 haste rating or 300 haste rating.

[nowimnothing30]

If it is a one percent drop rate, that means, as far as I know, that every kill that mob has a 1% chance to drop that item, and it's quite possible for 100 runs to happen and that 1% never come up.

correct. if you are trying to get a 1% drop item. you kill it and get to roll one number out of 100 each time you kill the mob, 1/100. if you kill that same mob again you don't get to roll off 1/99 numbers your back to 1/100 each time you kill it. you could run it 1000 times and never see it drop or you can get lucky and get it your first.

say you have a 20% crit chance at lvl 80. you have twenty tries out of 100 to crit or 1-20/100 to crit. any number you roll over 20 you don't crit and number lower than 20 you do.

if you fight 3 mobs your rolls are;
1,5,89,98,21 you just crit twice 2/5 40% crit chance
36,25,99,32,48. no crits 0/5 0% crit.
17,56,73,88,61 only one crit 1/5 20% crit chance.

total you made 15 attempts, 3 crits 3/15 for a total 20% crit chance...

[generalanders]

Wow a lot of idiots in here. Of course with a small sample size there is a possibility of more variance.

[underdogba]

A boss encounter and the resulting number of casts that initiate the "chance" calculations hardly falls under the "Law of Large Numbers"... while it's true that the larger the sample pool, the more accurate the average, In the numbers we are using in a boss encounter, RNG is much less reliable for an average.

I can easily have a 10% crit chance and get 2 crits out of 4 casts... wow I have a 50% crit chance?... I could also get NO crits in the same 4 casts and seemingly have 0% crit chance.

Supprting my argument is the fact that you can check Recount's logs and see that in some encounter with similar casts... say 70... you may see as much as 20% variance in crits over a few fights.

That is really a very misguided observation, because over even larger sample sizes (for example, every boss fight in a raid), empirically you will see distributions that converge better toward expected values.

Also the way you're describing "crit chance" equivocates the crit chance that you measure from statistical trials with what crit chance really means. Crit chance is a number that is held server-side, amoung other combat ratings, to determine the result of a combat roll. Having 10% crit chance and scoring 2 crits out of 4 casts does not give you a "50% crit chance", the chance that any given combat roll results in a crit is still 10%. Given adequate trials we should expect that the experimentally determined crit% should converge on crit chance %.

Just now I wrote a quick simulation, and over a few trials I've seen figures as low as 25.7% and as high as 35.7%. Obviously someone with 10% crit chance could crit in all 70 casts that you mention. It's just that on average it's likely to occur 1 time in more than 9999999999999930000000000000000000000000000000000000000000000000000000
times.

If RNG were such an overriding issue in PvE content, then there wouild be no point in discussing the relative merits of any combat ratings whatsoever.

[Marakara]

1) Adding 1% crit is worth exactly the same amount of absolute DPS whether you have 5% crit or 50% crit. However the relative DPS increase is higher when you have lower crit.

What people are interested in (or should be) are the relative dps upgrades. We're looking at the margins here. If you look at gear from an absoloute dps perspective, you're going to stack one stat and have nothing else, other than base stats. To find the maximum dps for your playstyle, most people will take a base set of gear, figure out which stat will give them the best upgrade, and work from there, hitting certain benchmarks (most often hit).

[Malentra]

Perform a Markovian analysis on a specific FrostFire rotation, then come back and try to tell me haste is better than crit. You won't.

A point to add here, in live now (3.0.8 ) Living Bomb procs HS as well. Which means LB and overall FFB/LB-HSPyro rotation dps scales slightly more now with Crit Rating.

Its also a point to always clarify in an assertion whether you are comparing the ratings (where 1 crit rating has to be ~ 50% more dps than 1 haste rating to be more overall dps) or the % value of each stat. These are pretty important clarifications to make, even if (currently) in both cases, crit is generally is better.

Of course with a small sample size there is a possibility of more variance.

[AuthurDent]

You just stated my point. That the size of the sample directly relates to the "average" or rather accuracy of "% chance" and actual chance over a given number of casts.

The more casts, the more LIKELY that the % will become closer... but 70-100 casts on a boss is far from "Law of Large Numbers" sampling. You own numbers show as much as 10% variance in actual crits... in a 1.6 million damage done session... that's as much as 100k less damage...

Either way... crit chance is a variable...

Haste on the other hand is a constant... more casts in the same amount of time.... each "extra cast" is also able to crit... increasing the "sample" in which the crit chance is averaged...

Lets say you could cast 3 more casts in the same time as you would cast 100... or take 3% more crits chance... and you are at 30% fire crit unbuffed...

The 3 casts WILL do extra damage AND each one has a 30% chance to crit.... or we can pass up the extra 3 casts to make our "existing" casts to have only a 3% more chance to crit...

If we could COUNT on that extra 3% to crit 3 more times it's a no brainer... but we can't.

One could also argue that with haste, more casts will complete between having to move in fights, where the longer casts would have to be aborted...

In the end it's like having a balanced investment portfolio... some in long term, safe items... and some in high risk, high reward items...

That is really a very misguided observation, because over even larger sample sizes (for example, every boss fight in a raid), empirically you will see distributions that converge better toward expected values.

Also the way you're describing "crit chance" equivocates the crit chance that you measure from statistical trials with what crit chance really means. Crit chance is a number that is held server-side, amoung other combat ratings, to determine the result of a combat roll. Having 10% crit chance and scoring 2 crits out of 4 casts does not give you a "50% crit chance", the chance that any given combat roll results in a crit is still 10%. Given adequate trials we should expect that the experimentally determined crit% should converge on crit chance %.

Just now I wrote a quick simulation, and over a few trials I've seen figures as low as 25.7% and as high as 35.7%. Obviously someone with 10% crit chance could crit in all 70 casts that you mention. It's just that on average it's likely to occur 1 time in more than 9999999999999930000000000000000000000000000000000000000000000000000000
times.

If RNG were such an overriding issue in PvE content, then there wouild be no point in discussing the relative merits of any combat ratings whatsoever.