Theory: Less crit chance vs higher level

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: Wethospu.6437

Wethospu.6437

I have seen people mention ~99% critical chance cap several times so I decided to research about it. Still work in a progress but hopefully I will be able to figure out something solid.

First of all, there is strong evidence that the cap is at 100%. I tested 1700 hits with 100% critical chance against level 80 target (Ancient Ooze) and all hits were critical.

Using statistics (calculations below), 99.98% chance that the cap is above 99.5% and 81.765% chance that the cap is above 99.9%. Blood Red Arachnid.2493 also did some kind of a test which supports the claim.


Based on http://en.wikipedia.org/wiki/Checking_whether_a_coin_is_fair.

1700 tests, 1700 successes and 0 failures.

Using formula: (tests + 1)! * r^success * (1-r)^fail / (success! * fail!)

I get: 1701 * r^1700.

Integrating that gives r^1701.

Let’s say 99.5% < r < 100% so we get (1^1701 – 0.995^1701) = 0.9998. So 99.98% chance that the cap is between 99.5% and 100%.

Let’s say 99.9% < r < 100% so we get (1^1701 – 0.999^1701) = 0.81765. So 81.765% chance that the cap is between 99.9% and 100%.


I don’t immediately recognize the formula you used, so I did a one-sample proportions test on the data you provided. One tailed.

Null Hypothesis: the percentage is 99%
Alternate Hypothesis: the percentage is 100%

Proportional standard deviation (sigma) = Square root of (p(p-1)/n)
Z-test: Z = (p – P) / 2sigma

N = 1700
p = proposed probability (0.99)
P = sample probability (1.00)
sigma = 0.0024
Z = 2.07
P = 0.9615

The Z value is the number of standard deviations away from the null that the sample test obtained. For a one tailed test, this comes to a probability of 96.15%, or some pretty strong evidence to reject the null.

So yeah, I’d say it is pretty safe to throw that “99% cap” theory out the door, and attribute the differences to other factors like enemy level changes.

So what’s going on then? Most of these non-critical observations actually come from Lupicus solos (as Dub showed on his video). Lupicus is above level 80 target which made me think that Precision -> Critical chance would be calculated on fly. Basically against higher level targets your Precision gives less Critical chance than normally.

At this point you probably remember glancing blows but they are very easy to spot because of the huge damage difference.

So I went to Swamp and did almost 600 hits on level 82 Sammoths. With 100% critical chance about 97% of my hits were criticals. I will collect more data and try to figure out how it exactly works.

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: Dub.1273

Dub.1273

I’ll do some attempts with 106% critchance, which should equal enough precision to get 100% crit aagainst level 82 bosses.

Dub | [rT]
#LoveArrows2013, never forget.

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: Stooperdale.3560

Stooperdale.3560

I had always assumed this was the case. I also assumed that sigils and runes triggered less often against higher level enemies and more often against lower level enemies.

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: Wethospu.6437

Wethospu.6437

Did 1000 more tests with 101% critical chance which resulted in 97.8%. Also noticed zero glancing blows on all tests so apparently criticals override glancing in this case.

Formula which seems to fit the data ignores base stats (916 Precision you get without any gear). So you always get that base 4% chance and then your extra Precision gets scaled. At level 80 it gets divided by 21. Level 81, 82, 83 and 84 values are 22, 22, 23 and 23.

2523 Precision (81.00%) gets scaled down to 77.50% which rounds to 78% and with Fury it gets to 98% as above.

Also tested 1000 hits with 99.67% critical chance (rounded 100% on UI). All crits so apparently the UI is correct.

I’ll do some attempts with 106% critchance, which should equal enough precision to get 100% crit aagainst level 82 bosses.

Will be interesting to see how this goes and whether it fits my results at all.

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: Nokaru.7831

Nokaru.7831

Creatures higher level than you critically hit more often, so the reverse is likely also true.

I wonder how downscaling and upscaling is factored into this. For instance: If you are downscaled to level 75 but are fighting a level 77 creatures, will your critical hits be reduced? It would be harder to test against something lower level than that due to the scaling of Precision. Also I wonder if the reverse is true: If you are level 15 and are upscaled to 80, will creatures have a higher critical chance against you?

(edited by Nokaru.7831)

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: Wethospu.6437

Wethospu.6437

White critters are usually very low level so you always get critical against them. However if I take my naked level 80 character to area which downscales me to level 4 I won’t get that many criticals against them (unlike in level 30+ areas). That indicates that downscaling (and probably upscaling) work properly.

So if you get downscaled to 75 you should get reduced criticals against level 77 enemies.

Theory: Less crit chance vs higher level

in Fractals, Dungeons & Raids

Posted by: frifox.5283

frifox.5283

Checked a few white critters and they were all lvl 1 regardless of the map. Also, checked those Legendary Defenders that defend wvw spawns and they are level 90 (!)