Thread: crits and devs

I'm trying to wrap my head around devs and crits, and I guess I will just muse about my wonderings until i can generate more specific questions.

It seems like when I'm parsing, crits and devs are usually below what is expected on a training dummy. I sit somewhere around 24% crit and 8-9% dev unbuffed, and so I am hoping to expect that, over time, when i perform multiple parses.

It seems like I struggle to hit 20% crits and 7% devs, and i'm wondering what the reason for that would be.


One thought would be the misses would contribute to that statistic. Those misses could have increased the crit/dev %, but more than likely would be noncrits.

Another thought makes me wonder how each hit "checks" for crits or devs, such as an order. For example, first it checks that it hits, then it checks that it crits, and if it crits, it checks that it devs. Or is that not how it works? I have no idea. If it does, then I guess i could understand how I'm never hitting my expected values, since Dev strikes could only be available through crit strikes, making dev strikes ~30%(10/35) of the crits that i hit, rather than 10% of the overall times i hit.

Or does it work simultaneously, checking for crits at the same time it checks for devs, and if both are true, it chooses the dev, thus making the true crit percent less than expected?

Anyway, hope someone can understand my rambling and help undiscombobulate me!

Thanks

As far as I understand it (through lots of forum reading) is that the two crits are additive.

1. Is the hit a Devastate?
Yes --> End steps, apply devastate damage
No --> Continue to step 2
2. Is the hit a critical?
Yes --> Apply crit damage
No --> Calculate normal damage

Unfortunately, I don't know where BPE/resist/miss falls into that, whether it's before (step 0) or after (step 3).

Firstly I find the crit rating I have versus what I get in actual fights is surprisingly close. If you have deep strikes traited your overrall crit rating from CA will not be accurate as dots don't real crit that much for whatever reason. I'm sure if you look at individual crit percentages you'll be satisfied, also long parses i.e 3-5min will almost guarantee you get the crits you should be getting.

As far as rolling goes from my understanding if you have 10% dev and 25% crit it rolls between 1 and 100. If you rolled between 1 and 10 its a dev, if you rolled between 11 and 35 its a normal crit and anything above 35 is normal hit.

My observations show me that hitting the dummy from behind increases your crit chance and reduces your misschance by about 2% each. So make sure to hit from behind like a burglar (everyone save the tank and tactical classes should do so anyway due to lack of blocking and parrying).
And I assume crits and devs stack multiplicatively with the crits overwriting the devs. I.e. a critchance/devchance of 25%/10% yields about 25%/7.5% while parsing for me. I could be wrong though as I tested this some time last year.

I think I remember a dev confirming that crit/dev actually works the way described by Shintagh, so crit and dev chance directly add up.

As for hitting the dummy from behind, are you sure you checked that with the exact same setup? As far as I know position has no effect on miss chance and the dummy can't b/p/e either, so there's no reason for getting higher hit % when hitting from behind. Also getting more crits when attacking from behind sounds like a myth to me.

Originally Posted by Vodomir

I think I remember a dev confirming that crit/dev actually works the way described by Shintagh, so crit and dev chance directly add up.

As for hitting the dummy from behind, are you sure you checked that with the exact same setup? As far as I know position has no effect on miss chance and the dummy can't b/p/e either, so there's no reason for getting higher hit % when hitting from behind. Also getting more crits when attacking from behind sounds like a myth to me.

I feel more powerful from behind.


... Giggidy.

Thanks for the replies everyone.

Originally Posted by Vodomir

I think I remember a dev confirming that crit/dev actually works the way described by Shintagh, so crit and dev chance directly add up.

Ok, so what if I have 8.9% dev and 23.9% crit? Does that mean I only have a 32% chance of critting/dev'ing? Am I missing out on that 1.8% because the roll is not taking into account decimals? Or does the roll actually base out of 1000 instead of 100?

Originally Posted by Vodomir

I think I remember a dev confirming that crit/dev actually works the way described by Shintagh, so crit and dev chance directly add up.

As for hitting the dummy from behind, are you sure you checked that with the exact same setup? As far as I know position has no effect on miss chance and the dummy can't b/p/e either, so there's no reason for getting higher hit % when hitting from behind. Also getting more crits when attacking from behind sounds like a myth to me.

The other question would be is that a 2% increase (e.g. 10% versus 10.2%) or a 2% differential (10% versus 12%). The former would simply be statistical noise. The later implies something very noticeabley different about the attacks which could either be real (that is a positional crit bonus) or due to small sample size (for instance a reported crit of 12% over 100 attacks would have a margin of error of 3.2% +/- so the "real" percentage could be as low as 8.8% and as high as 15.2%).



On that last note going back to the OP (hiya Figgs) a reported crit/dev rate of 24 and 8 (for a 32% total crit/dev chance) over a parse of 200 hits would have a margin of error at 3.3% total. If you are reporting 27% total crit/dev that would be a little bit outside the MoE (agin 32% +/- 3.3%) from the reported rate [doing the math the other way, that is taking the MoE on the 27% parse would yield a 3.1% MoE for a max of 30.1% and a min of 23.9%]. So it certainly seems that your reported crit/dev totals are a bit outside the reported percentages however even a small miss% would affect those numbers.

Lets take, for instance, an expected crit/dev of 32% over 200 hits you should expect 57 to 71 crits and devs ...if you make it 200 attacks with a 5% miss chance you would have only 190 hits total and an expected crit/dev of 54 to 67. Going to your actuals if you have 200 attacks, a reported rate of 27% then your total number of crits would be 54 which is, just barely, inside the Margin of Error at a 5% miss rate. So long story short if your miss rate is at or above 5% then your reported numbers are within the margin of error for your reported crit numbers and probably the result of a string of bad luck...if your miss rate is under 5% then there is something else mechanically wrong in the scenario preventing you from seeing your true crit/dev percentage.

Originally Posted by mrfigglesworth

Thanks for the replies everyone.



Ok, so what if I have 8.9% dev and 23.9% crit? Does that mean I only have a 32% chance of critting/dev'ing? Am I missing out on that 1.8% because the roll is not taking into account decimals? Or does the roll actually base out of 1000 instead of 100?

I really don't know if the roll is based on 1000 or on 100, but the former would make sense as both crit and dev chances increase by decimals.

Originally Posted by Vodomir

I think I remember a dev confirming that crit/dev actually works the way described by Shintagh, so crit and dev chance directly add up.

As for hitting the dummy from behind, are you sure you checked that with the exact same setup? As far as I know position has no effect on miss chance and the dummy can't b/p/e either, so there's no reason for getting higher hit % when hitting from behind. Also getting more crits when attacking from behind sounds like a myth to me.

I think Graalx once confirmed that avoidances stack additively and partial avoidances as well but both together stack mutliplicatively. As crits and avoidances share the same formula (and partial avoidances and devs as well) I assume this is true for crits and devs as well and my testresults seem to confirm this.

As for having more crits and fewer misses from behind: I once wondered why my dps dropped by about 5% from one try to the other without changing gear/traits/buffs until I realized I was hitting from the other direction. Those 2% I stated may not be the correct number, it was just a quick look at my miss/critchances but they seem accurate enough.

20 min later:
I just wacked on a dummy in Galtrev for 10 minutes from either side and cannot see such a significant difference in my hit-/critchance as I remember. So either my memory is bad (which I doubt), the combat has changed (which is unlikely) or I just had bad luck with my rolls on that particular day. But I will test this again.

Originally Posted by Vodomir

I really don't know if the roll is based on 1000 or on 100, but the former would make sense as both crit and dev chances increase by decimals.

Honestly the easier thing, probably, would be to roll directly against ratings, since you would only have to include the percentage translation in the display engine and not the combat engine as well. So long as the roll is checking for MOB and player level (and it should for miss chance if nothing else) it would be pretty easy to scale the roll size against the rating directly.

Originally Posted by Wilkens

Honestly the easier thing, probably, would be to roll directly against ratings, since you would only have to include the percentage translation in the display engine and not the combat engine as well. So long as the roll is checking for MOB and player level (and it should for miss chance if nothing else) it would be pretty easy to scale the roll size against the rating directly.

The conversion into percent would still need to take place in the combat engine to determine how to split the rating into the areas for crit, dev and non-crit. Lets assume a crit ratig of 10.000 equals 15% crit and 5% dev at a given player vs. mob level. Then rolls from 1-500 would be devs, 501-2000 would be crits and the rest would be non-crits, but how does the combat engine define those three distinct areas? It would first have to convert the rating into percentage values and then divide the rating according to those percentages - so the combat engine will always compute percentage values. And on the other hand ratings were first introduced way after the game was released. In the beginning there were no ratings, only percentages, so the combat engine was actually designed around calculating with those percentages.

They probably roll between 1 and 1000 or just round up the percentage to fit between 1 and 100. As for the other stuff, there is no evidence that indicates anything other than block/parry being affected when attacking from behind.

Originally Posted by Shintagh

They probably roll between 1 and 1000 or just round up the percentage to fit between 1 and 100.

If the latter is true, this is a pretty big deal. I can be at "max" with 9.1% and 24.1% dev and crit respectively, and focus on other buffs, (except for raidbosses) if the generator were to round up. I don't think turbine is that lazy, but that's a gut instinct. I mean, they can be lazy when it comes to certain things, but i've never seen them err in our favor this much.

Originally Posted by Vodomir

I think I remember a dev confirming that crit/dev actually works the way described by Shintagh, so crit and dev chance directly add up.

As for hitting the dummy from behind, are you sure you checked that with the exact same setup? As far as I know position has no effect on miss chance and the dummy can't b/p/e either, so there's no reason for getting higher hit % when hitting from behind. Also getting more crits when attacking from behind sounds like a myth to me.

I tried that both yesterday and the day before. The first time I was dualwielding and did not notice any noteworthy differences in critchance after having changed position.
Yesterday though I used my old twohander which I crafted shortly after RoI came out and I got the following results:



This is with attacking from one side for eight minutes (WA, SS, BS together with Second Wind without any other buff or skills.



And this is with attacking from the other side for eight minutes with the same rotation.

As you can see on the first screenshot my critchance in combat analysis corresponds to the one in my tooltip (purely by chance as I still had Vicious Strikes and the berserker 3-set bonus equipped). The dev-chance is actually lower than it should be.
For the second parse I just went around the dummy and attacked from the other side (it's hard to say which side is front as the selection ring changes its orientation when entering or leaving combat). The critchance has dropped by over 3% and the devchance by 1.5%. The to-hit chance has slightly increased, so this might be independent from the position (or my agility being 690 is far too high for a champ). Back then I only had the Draigoch Armour and less agility, so I will try this again after lowering my agility by a few hundred.

But the most interesting fact is that I can observe this drop in crit chance only with my twohanded axe but not with my onehanders. So this might at least explain why you never noticed this behaviour, Vodomir.

Edit: In the first screenshot the crit chance is correct (maybe a bit too low) while the devchance is far too low, so that is why I assume that crits do suppress crits.

Quick note: I've been using a slighly incorrect margin of error calculation: Revisiting it as this:

At the 95% confidence interval a sample of 711 and a reported percentage of 18% would yield a Margin of Error of 2.6%. Given this the crit chance rating is more than the margin of error different from the observed result but LESS than twice the observed margin of error.

At the 99% confidence interval the sample size would yield of MoE of 3.5% which is greater than the difference observed in the tst. This means that there is a possibility that this is just statistical noise (though the possibility is not great).

For the dev chance the differential is outside of the Margin of Error (2.1%) at the 95% confidence interval as well but again inside the 99% confidence interval.


So basically there is a SLIM possibility that this is statistical noise (~4-5% chance) but it does appear that there is an issue with 2H crits/devs and positioning.