http://img.photobucket.com/albums/v7...Rate45dDex.png
45dDex. mm, gonna do 47 next.
http://img.photobucket.com/albums/v7...Rate45dDex.png
45dDex. mm, gonna do 47 next.
Thank you Motenten Pchan and Shamaya, post #286 and wiki updated, we are soon done.
Currently working on dDex 30. Appears to be 13% at the 9000 hit mark, but I'd like to get the sample size up to 15k before posting.
Given what it looks like the formula is going to be, an alternate way of presenting it:
Given:
dDex = Player DEX - Mob AGI
Base crit rate = 5%
Then:
Crit rate = Base + f(dDex) + Merits + Gear
Where f(dDex) is defined as:
if (dDex < 0)
~ undefined -- 0?
if dDex < 20:
f(dDex) = floor(dDex/7)
else if dDex <= 40:
f(dDex) = 1 + floor(dDex/10)
else
f(dDex) = 5 + (dDex - 40)
f(dDex) has a max allowable value of 20.
The undefined for < 0 bit reminded me.. I can test -16 dDex as rdm/sch with errant. If that stays at the baseline, it's probably a pretty good bet that f(dDex) has a min of 0. Otherwise we'll need to test the other side for tiers as well.
check this post http://www.bluegartr.com/forum/3214066-post286.html 1 page behind.
you forgot 5% base in all your formulas.
but yea 2nd formula could start @ 20, i put arbitrary @ 15, but 20 works also. Actually, i think (very subjectively lol) that 20 is a right choice since its splitting the range [0;40] evenly XD.
No, you missed the initial formula. Base crit rate is shown in:you forgot 5% base in all your formulas.
Crit rate = Base + f(dDex) + Merits + Gear
f(dDex) is then defined as the component that changes with Dex. Base is always there and doesn't change with dex, so it's not part of that function.
This also reminded me, I want to re-test Pahluwan body's crit rate effect now that we have something definite to comare it to.
Vs lvl 71 ul'hpemde, 67 AGI, 68+29=97 DEX, +30 dDex
12.94% crit rate, 0.52% confidence intervalCode:Melee Damage Player Melee Dmg Melee % Hit/Miss M.Acc % M.Low/Hi M.Avg #Crit C.Low/Hi C.Avg Crit% Motenten 0 0.00 % 16115/2704 85.63 % 0/0 0.00 2085 0/0 0.00 12.94 %
a yes my bad misread wasnot direct crit% < <
For Pahlu, i remember i tested it "by mistake" while doing another test on LesserColis: ill tell yu the parses results, but i remember ~1%. And also, JPs poeple tested it 1% too. I changed to Dragon Harness since that day XD
EDIT: just tested a formula based on yur prop Motenten, works except "Max allowable for f(dDEX)" should be set @ 15, not 20, since Base=5% is added after.
Also still need Shamaya's dDEX=52 point + mine @ dDEX=0, for confirming Max and Base resp.
EDIT2: wiki updated.
Ah, right, whoops.EDIT: just tested a formula based on yur prop Motenten, works except "Max allowable for f(dDEX)" should be set @ 15, not 20, since Base=5% is added after.
And yes, the Pahluwan is long reported to be ~1%. What I'm curious about, though, is if it's done using an effective increase in Dex (say, +5 or something). If so, it could have a more significant effect when close to 40 dDex. Not that that's useful in practice, but it would be interesting to know.
I'm curious about the dDEX52 crit%, since crit based WSs testing are based on capped conditions, so either 24% or 25%...
No worries, I feel you. Given the size of my sample, I wasn't 100% satisfied with 50. I'm thinking about trying 51 first though (though atm I'm in the middle of doing 47).
why not going directly to 52 Sham ? not enough dex gears for ?
The dDEX51 point has ~1500/6000
Where did you get the data for 51?
from all pDIF testings on lvl0 mobs, all added together, and yes my bad realizing only now its not 51 :s forget what i wrote Sham :s (ty cdf)
I managed to get enough for +51 (although I had to remove Suppanomimi). We'll see how it turns out.
Nice, i almost thought yu left this testing XD :D
Two different trials at +51:
Critical hits: 3187 + 2454 = 5641
Total hits: 13165 + 10266 = 23431
Point estimate: .2407494
Interval estimate (95% CI): (.2352862, .2462770)
Conclusion: Crit rate is not 25% and is consistent with the idea that crit rate is capped at 20% before merits.
Also, I find it unnecessary to collect more data between +39 and +50. The data suggest a 1:1 correspondence between additional DEX and additional critical hit rate.
/nod CDF, same for lowest dDEX range, the data @ 3 and 6 suggest flat 5% before merits too.
Updated wiki.
http://img.photobucket.com/albums/v7...te47dDex02.png
dDex47---^
You know, you're probably right CDF. What do you think though, he who's brain pulsates with the aggregation of theorems, of this,
http://img.photobucket.com/albums/v7...te50dDex02.png
seaseaseaseasea - Nailing down the DEX/Crits interactions.
In Searain's data, dDex51 was 23.79 avg (actually I thought this was 50 b4 I went and glanced at it). Meanwhile his dDex50 came out to 23.03. What do you think is more likely--that (dex-agi) caps at 50 and our inaccuracy is an artifact of sample size--or that 50(dex-agi) brings you right about at the cap but not quite, and that another point is needed to top off your crit?
Should we get the slope of 39-50 now? I suck so I don't know offhand how to do that by combining all of the x-y values for each point, and also compensate for the difference in the sample sizes of each (this would be a good idea, right?).
edit: Just wanted to say, good work everyone. This looks about concluded. I checked out the discussion page on critical hit rate on wiki, and it looks good Masa.
http://www.bluegartr.com/forum/attac...1&d=1257681834
All glory for CDF, Masamunai, Motenten, Pchan and Shamaya, but also Deadgye Byrthnoth CD/Searain and TTaru japanese blogger :D
------------------ THE END ----------------------
/em turn off TV, finishes his pop corn and coffee and go to bed.