Melee crit mods, innate crit rates and crit rate hardcaps?

[Garluk]

During one of the numerous and verbose ranger vs monk "discussions" in guild chat on IRC, the topic of melee crit rates came up. Since I have close to 100 MB of combat logs I did some digging and came up with some numbers. None of the mobs included were immune to melee crits, meaning no golems, undead, etc were included. The fights were across several different zones such as DHK, MielechC, Everchill, etc. All data was taken during the last month. I can't be 100% certain that I was at 10% crit mods at all times, what I am certain of is that I was at at least +8% crit mods with +10% being far more likely.

Total fight time: 4 hours, 24 mins
Attack type, number of total hits, crits (including finishing blows but not cripples), crit %
Kick: 104, 8, 7.7%
Crush: 1492, 139, 9.3%
Pierce: 15891, 1619, 10.2%
Slash: 21805, 2173, 10.0%

Combined: 39292, 3939, 10.02%

My parser doesn't properly capture crippling blows but they are relatively scarce, somewhere in the 1-2% range.

I long ago finished all the AAs that would add to my critical hit rate. I've heard it said that crit mods cap at 10%. From the data above, it would appear that melee crit rates are hard capped at 10%. This would mean that once you've gotten all the melee crit AA's, you would need only a 3% crit mod item at most to reach the hard cap. Most of the players that I've spoken with were under the impression that the crit mod cap of 10% would stack with innate crit rates (warrior melee, ranger archery, monk flying kick, etc) as well as with melee crit AAs.

My question is: are melee crit rates hard capped and things are working as intended or is this something that needs to be reviewed?

Any input from the devs will be appreciated.

 

[Haisei]

A good question, i'd also like to know how much of a difference crit modifiers are really making, if any. Time to go look through a few logs of my own.

 

[Garluk]

Additional testing revealed interesting results. I would like to have a much larger sample set but I'm fairly certain that these results are a decent representative of what to expect.

Again, I have all comat melee crits and none of the mobs were immune to melee crits.

With 0% worn crit mods:
Slash: 1296
Crits: 130
Crit %: 10.03%

With 19% worn crit mods (yeah, that's not a typo):
Slash: 1049
Crits: 109
Crit %: 10.39%

Crippling blows were not included in the crit count.


Conclusions:

At least one of the following must be true to explain these results:

A) Melee crits are hard capped at 10% and nothing can raise it.
B) Melee crit mods are working as intended but have negligible effect since crit AAs will put you at/near the hard cap of 10%.
C) Melee crit mods are once again performing under their old method of modifying your base crit rate by a % (example: 9% crit AA +10% crit mods = 9.9% adjusted crit rate) instead of adding directly to your crit rate (example: 9% crit AA + 5% crit mods = 14% adjusted crit rate).

 

[Momatose]

very interesting

 

[Duma]

Max crit AA's + Eternal Charm +5% crit

Watchlight - Halls of Misery

532 HITS
91 CRITS

17.5%

Vigilant Wind Spirit - Upper PoAir

714 HITS
96 CRITS

13.4%

Mantash Djinni - Lower PoAir

269 HITS
47 CRITS

17.4%



Had innate warrior chance to crit working for me too though.

Perhaps higher level mobs have mitigation vs chance to crit?

 

[Gea]

Perhaps higher level mobs have mitigation vs chance to crit?

I don't hope so because that would be real bad and make no sense at all. Since you earn you HIGHER chance to crit, be it via AA or gear, to do more damage esp. vs. higher level mobs. Mitigation vs. crit (say 20% or whatever) would be ok, mitigation vs. chance to crit would be pretty stupid. No point in working for it anymore

 

[Garluk]

Vargg,

Your results are about what I'd expect. My bow crit rate is substantially higher than my melee due to rangers getting innate bow crits similar to warriors getting innate melee crits.

 

[phlit]

Is there any news on this? Been looking at my own logs and it seems for me also that melee crits don't go above 10%.

4 hours worth of melee combat logs (4 hours of fighting) seem to suggest a crit rate on 2hs of 10.36% (maxed crit AAs, +5 or 6 to crits from items, maybe 7, depending on what I was wearing.)

(Logs include no undead fighting, but various raiding, though none where crits are impossible I believe.)

 

[Wiz]

They do go above 10%, the cap is only on +critstrike from items.

 

[Garluk]

Parsed today June 11:

Total hits: 998
Total crits: 99 (this includes 7 cripples)
Total worn crit mods: +11%
Crit rate: 9.92%

This is completely in line with my above posted results. Meaning that once you get all the melee crit AAs that crit mod items are of zero benefit. I can't testify to their effectiveness prior to getting the AAs. Maybe they work then.

If the mods were working I should have had around 200 crits instead of 100 as I posted above that I could achieve a 10% rate with zero worn crit mods. 1k hits is a large enough sample size to indicate if this problem had been addressed.

 

[Wiz]

Parsed today June 11:

Total hits: 998
Total crits: 99 (this includes 7 cripples)
Total worn crit mods: +11%
Crit rate: 9.92%

This is completely in line with my above posted results. Meaning that once you get all the melee crit AAs that crit mod items are of zero benefit. I can't testify to their effectiveness prior to getting the AAs. Maybe they work then.

If the mods were working I should have had around 200 crits instead of 100 as I posted above that I could achieve a 10% rate with zero worn crit mods. 1k hits is a large enough sample size to indicate if this problem had been addressed.

No, you're utterly and completely off your rocker.

 

[rab]

I can't testify to their effectiveness prior to getting the AAs. Maybe they work then.

Crit mods on items actually don't work prior to attaining the crit AA, except in duels, even though code-wise, they ought to.

 

[Garluk]

No, you're utterly and completely off your rocket.

Ok, let's review the data I posted:

Unknown worn crit mods, anywhere from 7 to 10%
Hits: 39,292
Crits: 3,939
Crit %: 10.02%

With 0% worn crit mods
Hits: 1,296
Crits: 130
Crit %: 10.03%

With 19% worn crit mods
Hits: 1,049
Crits: 109
Crit %: 10.39%

With 11% worn crit mods done today
Hits: 998
Crits: 99
Crit %: 9.92%

The results are AMAZINGLY similar. Since Phlit says he has noticed similar results is it perhaps related to the ranger class?

Are there any non-warrior / non-ranger people that have done some parsing and willing to share their results? Please be absolutely certain that no mobs that might have been the victim of chanter curses are included as those can dramatically screw up your results. And monks need to exclude their kick results as they get an innate ability to crit kicks similar to warrior innate melee.

 

[Garluk]

I can't testify to their effectiveness prior to getting the AAs. Maybe they work then.

Crit mods on items actually don't work prior to attaining the crit AA, except in duels, even though code-wise, they ought to.

Rabb's comment made me consider how/if crit mods react in PvP situations for someone with max melee crit AAs so here are the results:

With 0% worn crit mods
Hits: 1208
Crits: 133
Crit %: 11.01%

With 11% worn crit mods
Hits: 1278
Crits: 142
Crit %: 11.11%

With 19% worn crit mods
Hits: 1106
Crits: 119
Crit %: 10.76%

With 25% worn crit mods
Hits: 1006
Crits: 123
Crit %: 12.23%

The 1-2% variation in these results when compared to what I consistently parse in non-PvP situations is statistically insignificant.

 

[soba]

No, you're utterly and completely off your rocket.

Usually when someone just blindly says "hey dis is broke" and doesn't provide any evidence of it you rightly ignore or make fun of the person but it seems like Garluk is providing enough evidence to at least warrant a little testing or a glance through the code, why write it off so quickly?

 

[zodium]

In an outrageous break from my comformist nature, I'm gonna say that I'm inclined to agree here. Garluk has historically been very thorough in whatever he does, and his results certainly have some weight behind them given his statistical samples and the close results.

 

[Gea]

No, you're utterly and completely off your rocket.

So it would be nice to hear what else one could do than give long-term parsings and actually get a valid answer. I mean, aren't they actually helping you? Ignoring them is more than disappointing and leading to no good.

 

[rab]

In an outrageous break from my comformist nature, I'm gonna say that I'm inclined to agree here. Garluk has historically been very thorough in whatever he does, and his results certainly have some weight behind them given his statistical samples and the close results.

QFT. Brandar is the closest this server has ever -- and probably will ever see -- seen to a staff-like poster that is not actually staff. He understands game mechanics amazingly well and thinks with his head and not his gut. His posts are consistently calm, well thought-out, and rational and are always appreciated.

 

[penfold]

i parse dps often on xp groups/raids. i just recently purchased fota3 and i only have 2% crit modifiers from items. i'll pull some numbers of not having maxed out crit mods in a day or 2. not sure if you guys/gals wanted maxxed crit mods or not, but if i can get 10% crits now; i guess it proves something.

i just checked some logs of fota2 + 2% crit mods. 8.5% crit's at the moment, but the numbers don't seem to be exact when i do the calculations myself. i'm using eqcompanion btw.

 

[Yakk]

Theory: +25% crits means "your chance to crit is multiplied by 1.25", and max AA gives 10% crit chance.

Assume the above. Then:

Then with 1000 samples, the expected range of observations for a given total crit mod would be:
00%: 10%+/-1.86% = [8.14% to 11.86%]
10%: 11%+/-1.94% = [9.06% to 12.94%]
20%: 12%+/-2.02% = [9.98% to 14.02%]
25%: 12.5%+/-2.06% = [10.44% to 14.56%]

(the above ranges are 19 times out of 20 confidence ranges, calculated by radius = 1.96 * sqrt(p*(1-p)/samples).)

Most interesting is:

With 25% worn crit mods
Hits: 1006
Crits: 123
Crit %: 12.23%

Which is a sample point outside of the confidence radius of the "assume 10% chance to crit for every situation" hypothesis, but inside the "assume crit mods multiply crit chance" hypothesis.

Also note that if crit mods are capped at 10%, then 12.23% is still inside the confidence radius of the 11% total crit chance, but outside of the confidence radius of the 10% total crit chance.

As a note, if this hypothesis (x% crit means that your crit chance is base% * (100%+X%)) is true, then this means "crit gear works, but it sucks".

Note that the above parses are miles away from the hypothesis "+X% crit means your chance to crit is base%+X%". Ridiculously far away. Like not even close. Know the moon? Not quite that far.

HTH.