So there has been a lot of debate on this recently, and I decided it was time to do the math and figure it out. The problem with comparing Power and Critical Hit Chance is that you are basically comparing apples to oranges. While they both increase dps, they do it in different ways, and Surge is another factor.

For the purposes of this theoretical arithmetic based test, I am using Megatfz's character builder, as it is already optimized, but doing several small tweaks to demonstrate the difference between crit and power. One of these builds will go pure power/surge, the other will go crit /surge, up to about 30%. The builds are identical except for swapping several enhancements to get either more power or more crit. Here are the links:

Power Build:

Bonus Damage: 635

Critical Hit Chance: 22.07%

Critical Hit Damage (surge): 77.8%

Critical Hit Chance Build

Bonus Damage: 588.6

Critical Hit Chance: 29.61%

Critical Hit Damage (surge): 77.8%

Difference Between Builds:

Bonus Damage: 46.4

Critical Hit Chance: 7.54%

Critical Hit Damage (surge): 0 (same)

All other stats = same.

So now we are left with this question: Which is better, 7.54% Critical Hit Chance, or 46.4 Bonus Damage? Let's do some math and find out! For the purposes of this simple test, I am going to ignore offhand damage (since it would just be a scaled down version of the mainhand that missed more).

Bonus Damage Test:

Mainhand Top End Damage: 1121

Critical Hit Chance 22.07%

Let's assume we hit a dummy 10.000 times, each hit dealing maximum top end damage. Of those 10,000 hits, 2207 would be critical hits (22.07%) doing 78.8% extra damage, with the remaining 7793 being normal.

7793 normal hits x 1121 = 8,735,953

2207 Crit Hits x 1121 x 1.788 = 4,423,596.036

Total Damage: 13,159,549.036

Critical Damage Test:

Mainhand Top End Damage: 1074.6

Critical Hit Chance: 29.61%

Same test. 10,000 hits. 2961 will be crit. 7039 will be normal

7039 x 1074.6 = 7,564,109.4

2961 x 1074.6 x 1.788 = 5,689,220.3928

Total Damage: 13,253,329.7928

Results:

Critical Hit Build Damage Done: 13,253,329 (883.5 dps)

Bonus Damage Build Damage Done: 13,159,549 (877.3 dps)

I am assuming a swing speed of 1.5 seconds for the GCD to determine dps. Obviously real dps would be higher, because certain abilities have a higher +bonus damage coefficient. This actually translates into a .99%, or about 1% dps increase for crit.

Great, so now we have some kind of result, right? But wait. Many abilities get more than just a flat +bonus damage from the bonus damage stat. Many of them get a larger coefficient, say x3 or even more. OK, so let's test that as well and compare.

Damage Build with 3x Coefficient on +Bonus Damage

7793 Normal Hits x 2,391 = 18,633,063

2207 Crit Hits x 2,391 x 1.788 = 9,435,163

Total Damage: 28,068,226

\Crit Build with 3x Coefficient on +Bonus Damage

7039 Normal Hits x 2251 = 15,844,789

2961 x 2251 x 1.788 = 11,917,397.268

Total Damage: 27,762,186

A difference of .98% dps, this time favoring +bonus damage.

OK, now I'm really confused, and to add to that confusion, every spec has some type of automatic crit ability that does significant damage. Annihilation = automatic DOT crit ticks while berserking. Carnage = auto Force Scream crits. Rage = automatic Smash crits. All of these are significant portions of the damage done by said spec, and when they are critting automatically due to talents, crit rating is worthless. So now the question is, is the 1% damage difference between set ups mitigated or even reduced to being less effective than +bonus damage due to those auto crit abilities?

And here is where my paper napkin math ends, because trying to quantify the percentage of damage that critical hits will not affect at all gets very deep into the classes core mechanics and talent specs. To determine if the 1% effectiveness difference is mitigated or even reduced to being less effective than stacking pure +bonus damage, we would need to program a simulator to do ideal rotations, and do 10's of thousands of iterations of those rotations. And even then, we would have to take into account all the different abilities and their coefficients. And then we would have those "perfect world" scenarios to compare, but as we all know, that never happens in pvp due to stuns, cc, snare, deaths, etc.

My hypothesis is this: If we programmed simcraft or a similar program to find out EXACTLY which one is better and to what amount, I think that we would find the difference to be negligible. The only spec that I believe will really see a large difference between power/crit would be Rage, due to big smash auto crits being such a large percentage of the damage. Carnage and Annihilation, the difference in dps will be negligible, however annihilation benefits from crit in the form of extra heals, but again, this portion is going to be small.

The bottom line is that the difference is probably negligible, and you should just get as much +power and +crit as you can while keeping accuracy between 99-100% melee, and surge around 75%, more is better, but don't sacrifice power / crit for it (and you shouldn't have to, since pieces can either get power + surge or crit + surge, or power + acc / crit + acc, so you won't ever have to choose).

When in doubt, it's probably best to err on the side of crit for PvP simply because killing healers in PvP requires burst damage, and more crit = more burst. Please feel free to double check my numbers, post any errors, and discuss this post. I welcome constructive criticism and debate, but please keep this to a discussion, and not an argument.

Too long to really read.

Just add cliffs.

The shorter fight lengths in PVP make erring on the side of crit a bad idea due to RNG. In longer battles the randomness tends to even out and you'll crit at the % your rating would suggest.

In a short fight, like in PVP, It's entirely possible that the RNG will leave you with 0 crits outside of auto crit abilities, or even all crits even though your crit % is only 20.

The shorter fight lengths in PVP make erring on the side of crit a bad idea due to RNG. In longer battles the randomness tends to even out and you'll crit at the % your rating would suggest.

In a short fight, like in PVP, It's entirely possible that the RNG will leave you with 0 crits outside of auto crit abilities, or even all crits even though your crit % is only 20.

This.

The shorter fight lengths in PVP make erring on the side of crit a bad idea due to RNG. In longer battles the randomness tends to even out and you'll crit at the % your rating would suggest.

In a short fight, like in PVP, It's entirely possible that the RNG will leave you with 0 crits outside of auto crit abilities, or even all crits even though your crit % is only 20.

Either way, the entire point I am trying to make is that it doesn't really matter. Combined with the millions of other variables, you will not notice a difference.

I was wondering exactly this last night!

I had the choice between +crit and +power gear on drops in an FP and wasn't sure whether it really mattered all that much.

It would make sense that you would want a stat that is constant (power) which basically gives you stable dps over time vs. a crit% increase(RNG) with COULD give you a good burst or two IF you're lucky. I would imagine you would want as much strength as you could find, which again would favor a power stat over crit%, no?

I guess it really doesn't matter when levling, but thanks for the info!

I think you have to consider the spec, as the OP mentions. Here's my two cents:

Annihilation: Crit is very important here because critical bleeds not only heal, but gain bonus surge damage from bleedout. This is a case where you'd want the 30% unbuffed crit so your force critical chance is around 40% buffed. The primary source of damage (bleeds) is amplified greatly by crit for this spec.

Carnage: Crit is still important as ever. The better crit chance, then the more devastating a ravage will be after using a gore. You can easily (not really easily, of course) eclipse a 6k hit from the 3rd strike of ravage if it crits. Every ataru hit and off-hand strike has a chance to do extra damage as well from Massacre. The auto-crit attack is not the primary source of damage in this (force scream), whereas everything else benefits tremendously.

Rage: I almost want to say that crit doesn't matter at all, but that's an overstatement. Smash is the focal point of this spec, but the next best things are crush and vicious slash. Both would benefit from crit, but much less than smash benefits from power. This is one spec where crit should be there, but I think around 30% force crit chance when buffed is plenty. Power is very important here.

That's my take, feel free to disagree.

IAnnihilation: Crit is very important here because critical bleeds not only heal, but gain bonus surge damage from bleedout. This is a case where you'd want the 30% unbuffed crit so your force critical chance is around 40% buffed. The primary source of damage (bleeds) is amplified greatly by crit for this spec.

That will be my preferred Spec. Thanks for the advice!!

That will be my preferred Spec. Thanks for the advice!!

He's wrong. And he's so wrong it's not even funny. You should have Berserk up as much as possible as an Anni marauder, and when you have Berserk up your bleeds auto-crit and therefore all your crit rating is worthless.

Power is much more useful because it means those bleeds that are auto-critting are hitting harder.

Edit: The only way for the Power vs Crit debate to be settled is with exhaustive testing, either in game or on a very accurate sim. In game, I would say do repeat a 10 minute parse 10 times and then average out the damage. Then trade out Power for Crit and do 10 more 10 minute parses and average out the damage. Whichever one has higher average damage is probably the way to go.

If you want to be truly exhaustive, you can swap out power for crit on every piece of gear, one at a time, and test at each different level to see which comes out the best, but it would take a long time to get the gear required for the tests and even longer to run the tests.

Because of that, an accurate sim may be the way to go.

He's wrong. And he's so wrong it's not even funny. You should have Berserk up as much as possible as an Anni marauder, and when you have Berserk up your bleeds auto-crit and therefore all your crit rating is worthless.

 

Power is much more useful because it means those bleeds that are auto-critting are hitting harder.

 

Edit: The only way for the Power vs Crit debate to be settled is with exhaustive testing, either in game or on a very accurate sim. In game, I would say do repeat a 10 minute parse 10 times and then average out the damage. Then trade out Power for Crit and do 10 more 10 minute parses and average out the damage. Whichever one has higher average damage is probably the way to go.

 

If you want to be truly exhaustive, you can swap out power for crit on every piece of gear, one at a time, and test at each different level to see which comes out the best, but it would take a long time to get the gear required for the tests and even longer to run the tests.

 

Because of that, an accurate sim may be the way to go.

^ This.

I've seen people throw around all kinds of arbitrary numbers for what the "right" Crit Rating is for Annihilation, and honestly it's all kind of bunk.

Likewise, if you math it out, Malice, for all it seems like a talent tailor-made for Annihilation, is actually *less* useful than something like Ravager because such a large percentage of bleed ticks are already going to crit (thanks to bonus crit % from Juyo stacks plus Berserk).

The only way to find the "right" answer is via simming it, but even then, sims have a very hard time accurately depicting every possible in-game situation, so there's always an element of YMMV.

One thing I do take issue with is when people recommend ultra-low Crit Ratings, as rarely ever will more than 40% of Annilhilation spec's damage come from bleeds over an extended fight, and the other 60% benefits much more from a higher base crit chance than bleeds do.

Nice to see someone actually took the time to sit down do the math on this. I can pretty much confirm the same thing on this from my experiences with my bounty hunter, though my thoughts are mostly theory. I used to run with a heavy damage build, all aim, power and surge and I really thought I was tearing stuff up because of it. Out of curiosity later on I built a crit suit replacing most of the power with crit getting up to just over 30%, just to try it out. In the end my conclusion was that they are about equal and at that point your accuracy vs your targets defense is what's going to actually make the difference.

Theoretically if you're running around with say 106% accuracy and your target has say 8% defense you still have a 2% chance to miss. There's always a chance, no matter how small, that the one that misses would have rolled a crit, and that chance is increased if you stack more crit. The chance of your potential crit missing should rise the more your target's defense outreaches your accuracy. If you're stacking for crit I feel this would mean a more significant overal damage loss on your part in a long fight, at least more so than if you're stacking power, though it probably wouldn't make too much of a difference is short one.

Unfortunately I don't have the math skill to calculate it or the patience or time to test it, but those were my thoughts on the subject. I say thanks to Aluvi for putting the brain power into this.

Too much surge, and your crit build has not enough crit.

I'm at 595 bonus damage and 35,5% crit

Nice to see someone actually took the time to sit down do the math on this. I can pretty much confirm the same thing on this from my experiences with my bounty hunter, though my thoughts are mostly theory. I used to run with a heavy damage build, all aim, power and surge and I really thought I was tearing stuff up because of it. Out of curiosity later on I built a crit suit replacing most of the power with crit getting up to just over 30%, just to try it out. In the end my conclusion was that they are about equal and at that point your accuracy vs your targets defense is what's going to actually make the difference.

 

Theoretically if you're running around with say 106% accuracy and your target has say 8% defense you still have a 2% chance to miss. There's always a chance, no matter how small, that the one that misses would have rolled a crit, and that chance is increased if you stack more crit. The chance of your potential crit missing should rise the more your target's defense outreaches your accuracy. If you're stacking for crit I feel this would mean a more significant overal damage loss on your part in a long fight, at least more so than if you're stacking power, though it probably wouldn't make too much of a difference is short one.

 

Unfortunately I don't have the math skill to calculate it or the patience or time to test it, but those were my thoughts on the subject. I say thanks to Aluvi for putting the brain power into this.

Just to note... Merc isn't a great comparison in this regard, due to the point in Critical Reaction that the cookie-cutter Arsenal build should use. Keeping the Alacrity buff up means a sizable increase in overall DPS (since ~70-80% of your overall damage comes from channels), and thus you have an active incentive to build up your crit chance a bit higher.

And to your second point, you're right to a point but it's ultimately a non-issue. 285 Accuracy rating (5 BH Accuracy enhancements) + Legacy Accuracy companion buff = 99.89% base accuracy, or 109.89% Special/Force?Tech accuracy. Even with a worst-case assumption of 10% defense chance, you're only .11% short of cap, which is acceptable given the current gear availability. Thus, Accuracy really is off the table when you're number crunching for endgame, but you do have to consider the balance between Power and Crit rating.

How about this to blow your mind with even more factors to consider: critical hits can never be shielded. Discuss!

So.. crits are slightly more useful on like 1% of the pvp population? I don't know of very many tanks that actually use tank gear with absorption on it =/

So.. crits are slightly more useful on like 1% of the pvp population? I don't know of very many tanks that actually use tank gear with absorption on it =/

A ton in regular warzones. I'd imagine much less in rated, but there's decent metagame potential for having at least one guy with shield on your team since they can eat a significant amount of carnage/sniper/pyro/jug-hybrid rotations.

Edit: There's also more factors to consider then just scream autocritting. You can't say for instance that because scream is 20% of your damage crit is 20% less effective. In this instance it's not a quality, but a quantity. How many of your hits are screams? Given things like zerk and being able to fit roughly 16 attacks in (counting atarus, retaliation, gore, reduced GCD) in per force scream it gets nuts. Ultimately you'd need to find not how much damage scream comprised, but how much of you rotation it comprises, which I'd bet is <10%. Then you could say at what point is 90% damage/crit/point worse then 100% of damage/power/point.

He's wrong. And he's so wrong it's not even funny. You should have Berserk up as much as possible as an Anni marauder, and when you have Berserk up your bleeds auto-crit and therefore all your crit rating is worthless.

 

Power is much more useful because it means those bleeds that are auto-critting are hitting harder.

 

Edit: The only way for the Power vs Crit debate to be settled is with exhaustive testing, either in game or on a very accurate sim. In game, I would say do repeat a 10 minute parse 10 times and then average out the damage. Then trade out Power for Crit and do 10 more 10 minute parses and average out the damage. Whichever one has higher average damage is probably the way to go.

 

If you want to be truly exhaustive, you can swap out power for crit on every piece of gear, one at a time, and test at each different level to see which comes out the best, but it would take a long time to get the gear required for the tests and even longer to run the tests.

 

Because of that, an accurate sim may be the way to go.

If one acknowledges that there is a debate on the topic, then he can't really be "so wrong its not even funny." Bleeds area large portion of the damage, but they aren't the largest portion of it. Also berserk, at most, is going to last for 12 ticks or around 6 seconds, assuming you they don't get cleansed or otherwise wasted. Having a decent amount of base crit will encourage more annihilation crits, which during many PvP skirmishes, will get more use and deal more damage than the bleeds with berserk being spammed.

In my opinion, attempts to math out PvP is not good practice, since PvP is random throughout and no dummy can factor real PvP mechanics, namely the player one. It's good for consideration, but for me, I try to strike a balance between the two so that I'm not a one trick pony, or relying on only one or two skills to autocrit on a timer. Too much crit could disadvantage you in the random duel that could occur in a wz. Not enough could disadvantage you in the extended fight, that seems more common place.

A ton in regular warzones. I'd imagine much less in rated, but there's decent metagame potential for having at least one guy with shield on your team since they can eat a significant amount of carnage/sniper/pyro/jug-hybrid rotations.

 

Edit: There's also more factors to consider then just scream autocritting. You can't say for instance that because scream is 20% of your damage crit is 20% less effective. In this instance it's not a quality, but a quantity. How many of your hits are screams? Given things like zerk and being able to fit roughly 16 attacks in (counting atarus, retaliation, gore, reduced GCD) in per force scream it gets nuts. Ultimately you'd need to find not how much damage scream comprised, but how much of you rotation it comprises, which I'd bet is <10%. Then you could say at what point is 90% damage/crit/point worse then 100% of damage/power/point.

We're talking about effective damage, not effective "rotation" percentages. If, on a dummy, in doing a normal carnage rotation, 20% of your damage comes from force scream auto crits, that reduces the effectiveness crit has on your overall dps by 20%. In actual use, it's probably more than 20% because of downtime between screams being filled by cc/stun/snare etc. During your inactive time, cooldowns are resetting, and your filler abilities/rotation become even less significant. It doesn't matter if you use force scream once per 50 other abilities, or once per 10. What matters is, what percentage of your total dps does force scream come out to? Sorry but your thinking is just flawed.

Next, about the shield point. I've been thinking about this, and although it may be somewhat helpful against tanks, It is very rare that I actually directly damage a tank in PvP. More often, I am damaging his shielded target, and he ends up dying for that target (usually a healer).

As for actually beating on a tank, here is the reason why critical rating is not "extra helpful" versus a tank (it all has to do with how the 2 roll combat system works in SWTOR):

The 2-Roll System

 

SWTOR leverages a 2-roll system for determining the outcome of an opponent’s attack as explained by BioWare’s Georg Zoeller (references: 1 2 ), bold emphasis mine:

 

First is a hit roll, accuracy versus defense, and if the attacker misses then no damage occurs. If the attacker rolled poorly enough to miss even discounting the target’s defense then a “Miss” result occurs. If he misses because of the defense then the result varies based on the attack type, the cover state of the target, and the target’s equipped weapons. All the possible results – Dodge, Parry, Deflect, Resist, Cover – are mathematically the same, but they can trigger different effects and are visualized in different ways.

 

If the attacker hits, then a second roll is made with the crit chance of the attacker versus the shield chance of the target. If a Crit or a Shield occurs then the damage is adjusted up or down (based on Surge/Absorb), and then it goes through to the armor and damage resistance. A critical can never be shielded, and an attacker with a high enough crit chance can push the target’s shield chance off the table. It shouldn’t be possible to get your passive crit chance high enough to start pushing off the target’s shield chance, but there are short-duration buffs that push these chances high enough to come into conflict.

 

Which Attack Types are Defensible and Shieldable

 

There are 4 attack types: Force, Melee, Ranged, and Tech.

 

The attack types should not be confused with the 4 damage types: Elemental, Energy, Internal, and Kinetic.

 

Melee and Ranged attacks are Defensible and Shieldable, whereas Force and Tech attacks are not. This is consistent with other games such as WoW and RIFT, where “spell” attacks are not dodgeable / parryable and not blockable.

 

Based on what Georg wrote , here is my guess on what the roll table for Crit-vs-Shield looks like for Melee and Ranged attacks that were not Defended.

 

Scenario 1

 

Attacker’s Crit Chance: 30%

Defender’s Shield Chance: 30%

So remaining 40% is normal hit (not Crit, not Shielded)

Scenario 2

 

Attacker’s Crit Chance: 60%

Defender’s Shield Chance: 30%

So remaining 10% is normal hit (not Crit, not Shielded)

Scenario 3: Shield Chance reduced

 

Attacker’s Crit Chance: 80%

Defender’s Shield Chance: 30%, but for the roll calculation it’s treated as 20% because the Crit Chance pushed 10% off the table

Scenario 4: Shield Chance negated

 

Attacker’s Crit Chance: 100% (e.g. due to a proc buff that makes the attack Crit)

Defender’s Shield Chance: 30%, but for the roll calculation it’s treated as 0%

Therefore, the attack is a guaranteed non-Shielded Crit

So unless you are stacking crit high enough to push the shield chance off of the table, it doesn't really matter. I understand if this is a difficult concept to comprehend, as it requires some critical thinking (no pun intended.. well ok maybe a little ) and math.

We're talking about effective damage, not effective "rotation" percentages. If, on a dummy, in doing a normal carnage rotation, 20% of your damage comes from force scream auto crits, that reduces the effectiveness crit has on your overall dps by 20%. In actual use, it's probably more than 20% because of downtime between screams being filled by cc/stun/snare etc. During your inactive time, cooldowns are resetting, and your filler abilities/rotation become even less significant. It doesn't matter if you use force scream once per 50 other abilities, or once per 10. What matters is, what percentage of your total dps does force scream come out to? Sorry but your thinking is just flawed.

What? Surge affects abilities differently. I shouldn't have to elaborate from there. I probably shouldn't have said 20% of your damage since that's both inaccurate when talking about the changing of stat weights and the wrong way of representing the crit argument, so my apologies. It's not what percentage of damage since that is going to vary on stat weights, it's what percentage of your rotation since that is not going to vary (ideally). If massacre is say 50% of a carnage rotation then I want to find out, oh say, how much an extra 6% chance to do 105% more damage is going to impact versus whatever power is doing. You don't use final damage because that is the thing you are modifying so when you say 20% comes from scream you've reached the end of the argument rather then the beginning.

Also, I don't know if you were addressing me, but I didn't say anything about shield other than it had decent potential as a meta game counter and that I see quite a few in normal warzones.

The point I'm trying to convey is that, since Power Vs Crit is already such a close fight (less than a 1% difference either way), auto crit abilities tip the scale in favor of power, since power enhances 100% of your damage.

I'm not clear on what you mean by "surge affects different abilities in different ways". To my knowledge, surge is a flat buff to your criticial damage increase. The only way it would have different effects is if said abilities are auto crit, meaning it is much more effective on them. In the case of force scream or smash for example, since they will always crit if you play correctly, it is much more effective. Is this what you meant?

The point I'm trying to convey is that, since Power Vs Crit is already such a close fight (less than a 1% difference either way), auto crit abilities tip the scale in favor of power, since power enhances 100% of your damage.

This.

The point I'm trying to convey is that, since Power Vs Crit is already such a close fight (less than a 1% difference either way), auto crit abilities tip the scale in favor of power, since power enhances 100% of your damage.

Auto-crit abilities will not "tip" the scale for power because the math doesn't support it.

I don't know how you can forget this, but auto-crit affects both high crit/high power builds.

Assuming it's 20% reduction which some people are throwing around, but I'll go ahead and assume 30% to prove my point. Assuming your crit numbers are correct, a 30% reduction on both builds would be 20.727 vs 15.449, or a 9.2% vs 6.621 total net difference. That 2.6% crit is completely marginal. Buoyed by surge, that would translate to a 2% damage increase. That's the grand difference of 10k damage in a 500k warzone, ceteris paribus.

In other words, no matter what you do, power or crit will not make a difference, even taking into account auto-crit abilities. Here's an excellent post by LagunaD for Combat sentinels that explains it further: http://www.swtor.com/community/showpost.php?p=4906895&postcount=44

^^ this seems like a really good summary of the issue. That link is good too.

This is why I threw the "criticals never getting shielded" tidbit in the discussion, because in the long run, in a Warzone of defined time, it doesn't really matter which you emphasize, as long as you don't completely neglect one for the other (which is why I don't like the term "stack"). The effects of the emphasis is sometimes more important than the actual value you get from the stat.

For example, crit-based procs (which I know are very few-and-far-between on Sentinels/Marauders) will benefit from a high crit rating, even if power has more (albeit slight) benefit.

I know that Zen/Beserk is up probably more than 50% of the time when specced Watchman/Annihilation, but the fact is, it isn't up all the time. Crits on burns heal us, so that is something that lends itself to emphasizing crit.

Also, surge is not our only source of increasing crit damage. All of the trees have some talent that increases crit bonus damage. So that is something to consider.

Finally, with regards to the auto-crit Blade Storm/Force Scream, it is a Force attack, so it can't be shielded. So that effect should be completely disregarded when talking about shielding. I don't believe that just because we can't get to 100% crit that we should neglect the effect we have on shields. As that same post that was quoted above says, "A critical can never be shielded." That means that we are guaranteed that if we have a 40% crit percentage, 40% of our attacks are unshieldable. So for getting off that Merciless Slash/Annihilate in Watchman/Annihilation, or that Slash/Vicious Slash in Focus/Rage, that is incredibly handy against the many tanks that do run around with shields in their shield stance.

Again, I am not a fan of "stacking" one stat to the detriment of all other stats. I like to emphasize stats.