Does Level Correction apply to everything?

[AoshiZ]

It is this level correction that leads me to believe that there isn't one for critical hit rate. Due to level correction on Acc vs Eva, your hitrate is already being scaled down, so your crit rate will depend in part on your hitrate. If you have 75% hitrate and 15% criticals, over the course of 100 melee hits, 75 will land and of those 75, around 12 will be critical hits. It wouldn't be fair in my opinion to level correct criticals because then you would end up with a much lower critical hitrate.

I'm confused as to what accuracy has to do with critical hit rate. Are parsers calculating critical hit % with the total number of swings instead of number of criticals within the total number of regular hits? Because that would be a statistic that would have very little use.

Addition: When calculating for myself i only use regular melee hits and regular critical hits. In which case the critical hit % values i get are generally very consistent

[TummieGaruda]

I'm confused as to what accuracy has to do with critical hit rate. Are parsers calculating critical hit % with the total number of swings instead of number of criticals within the total number of regular hits? Because that would be a statistic that would have very little use.

Addition: When calculating for myself i only use regular melee hits and regular critical hits. In which case the critical hit % values i get are generally very consistent

Higher level accuracy will give you a better sample of critical hitrate whereas having lower accuracy will give you a much smaller sample size.

[AoshiZ]

Higher level accuracy will give you a better sample of critical hitrate whereas having lower accuracy will give you a much smaller sample size.

Ok I understand your point.

Well maybe this will help. I found some parses for a period of time where my gear was most likely the same. Each of the parses do not have a Corsair giving rogues roll. I picked these parses because of the large disparity in level difference between the mobs. Of course the mobs native attributes play a factor but you may be able to see that despite the massive level differences the critical hit rate hasn't changed in a large negative manner do to the level difference

Kirin
Name Total Damage DPS Accuracy% Crit% Counter% EVA% Melee DOT Skill Damage Hits Misses Delay MAX Hit MAX Skill MAX Critical Magic Damage MAX Magic Damage
Aoshiz 10,444 39.12 89.13% 7.32% N/A N/A 5,601 4,843 41 5 348.23 283 1,522 260 0 0

JOL
Name Total Damage DPS Accuracy% Crit% Counter% EVA% Melee DOT Skill Damage Hits Misses Delay MAX Hit MAX Skill MAX Critical Magic Damage MAX Magic Damage
Aoshiz 14,159 17.56 86.78% 6.67% 0.00% 0.00% 7,071 7,088 105 16 399.73 216 687 207 0 0

Salvage
Name Total Damage DPS Accuracy% Crit% Counter% EVA% Melee DOT Skill Damage Hits Misses Delay MAX Hit MAX Skill MAX Critical Magic Damage MAX Magic Damage
Aoshiz 141,555 23.70 93.30% 8.56% 1.54% 2.26% 69,118 72,437 362 26 923.63 570 1,973 395 0 0

Merit Colibri
Name Party Damage% Total Damage DPS Accuracy% Crit% Counter% EVA% Melee DOT Skill Damage Hits Misses Delay MAX Hit MAX Skill MAX Critical Magic Damage MAX Magic Damage
Aoshiz 39.47% 777,346 66.90 94.21% 10.28% 0.35% 5.59% 370,546 406,800 1,692 104 388.19 625 2,177 827 0 0

[ringthree]

For reference Kirin is 82 and Jailer of Love is 83 (Do either of these ever change?).

While nice information the Salvage stats don't give us anything to compare to because we don't know the levels of any of the mobs and they vary widely across the zone.

The Colibri information will be useful for providing a baseline I think.

[Kirschy]

I ended up being a lot busier than I expected, but I was still able to run 1,500 hits worth of tests. I'll list my data first, and then explain how I got:

Steelshell - The Boyahda Tree
Lv 76 - Tough
Evasion 291

WAR/SAM - Baha Scythe
Accuracy Total - 329

329 Accuracy vs 291 Evasion
+1 Level Difference
According to Wiki Formula... 75% + 14% = 89%

1,500 Total Hits(Yea landed on exactly 1500!)
1,386 Melee Hits, 114 Misses
92.4% Hit rate%

I don't know what the margin of error is, but it seems fairly close to the expected.


Experiment:
First thing I did is check the evasion level of the Tough crabs. I found that using the following set of equipment allowed me to change from even evasion to low evasion.

75WAR/37SAM - Hume
DEX: 68
Dagger Skill: 240

Accuracy from DEX: 34
Accuracy from Dagger Skill: 200 + 36 = 236
Ares Head + Potent Belt + Snipers Ring +1 + Diabolos Earring = +30
Total: 300 ~ /check Even Evasion

When accuracy from eq was changed to +31, it /checked Low evasion. Wiki claims +10 accuracy over the evasion of the mob will check low evasion. Therefor I place the evasion of Lv76 Steelshells @ 291 Evasion.


For the data I used WAR/SAM /w Shell/Pro/Regen III/Haste/Marchx2/Seigan/Third Eye/Defender.

Parsed data using KParser. Seperate parses for each mob. There was 5 Steelshells in total.

Eq Setup was as follows:
Baha Scythe/Pole Grip
Turban/Blank/Ethereal/Evasion
Askar Body/Dusk+1 Hands/Jelly Ring/Blitz Ring
Boxer's Mantle/Velo Belt/Byakko/Askar Feet
(I have pics, but I'm in a rush right now~)

75WAR/37SAM
Food: Tav Taco
Scythe Skill: 272
DEX: 68 + 15(Byakko) + 4(Food)

Accuracy from DEX: 87*.75 = 65
Accuracy from Scythe Skill: 200 + 64 = 264
Accuracy from Eq: 0
Total Accuracy: 329

Hopefully I hit everything. I'd like to think the small difference in expected value and actual value is from some dDEX relation, but it might just be margin of error~ If I get more free time I'll continue /w the testing.

[TummieGaruda]

I ended up being a lot busier than I expected, but I was still able to run 1,500 hits worth of tests. I'll list my data first, and then explain how I got:

Steelshell - The Boyahda Tree
Lv 76 - Tough
Evasion 291

WAR/SAM - Baha Scythe
Accuracy Total - 329

329 Accuracy vs 291 Evasion
+1 Level Difference
According to Wiki Formula... 75% + 14% = 89%

1,500 Total Hits(Yea landed on exactly 1500!)
1,386 Melee Hits, 114 Misses
92.4% Hit rate%

I don't know what the margin of error is, but it seems fairly close to the expected.

+10 acc over the mobs evasion when the mob is Even Match = 80% Hitrate. When your acc matches the mobs evasion on an even match it is 75%

291 + 10 = 80% hitrate or Low Evasion /check on an EM mob
1 level difference = +4
291 + 10 + 4 = 305 to con Low Evasion or 80% Hitrate

329 - 305 = 24
Every 2 acc = 1% hitrate so:
24/2 = 12% hitrate

80% + 12% = 92% which matches the parse


What was your Critical Hitrate?

[pchan]

of tests. I'll list my data first, and then explain how I got:

Steelshell - The Boyahda Tree
Lv 76 - Tough
Evasion 291

WAR/SAM - Baha Scythe
Accuracy Total - 329

329 Accuracy vs 291 Evasion
+1 Level Difference
According to Wiki Formula... 75% + 14% = 89%

1,500 Total Hits(Yea landed on exactly 1500!)
1,386 Melee Hits, 114 Misses
92.4% Hit rate%

I don't know what the margin of error is, but it seems fairly close to the expected.

p=0.924, the 95% error is sqrt(p*(1-p))*1.96/sqrt(1,500)=+/-1.34%

Your test shows that with 95% probability, your real hit rate is between 91.06% and 93.74%.

edit : this does not confirm that level difference is -4 eva. If the model is correct (excluding level difference), it only allows integers and integers + 0.5 as hitrate, so only 91.5 92 92.5 93and 93.5 are allowed, and therefore it's -1,-2, -3, -4 or -5.

edit2 : The approximate number of swings that you need to get an error < 0.5, is about 10,000 .. And even then, if you find, say, 91.75, you cannot conclude..

[Kirschy]

329 - 291 = 38
291 + 10 = 80% hitrate or Low Evasion
1 level difference = +4
291 + 10 + 4 = 305 to con Low Evasion or 80% Hitrate

329 - 305 = 24
Every 2 acc = 1% hitrate so:
24/2 = 12% hitrate

80% + 12% = 92% which matches the parse


What was your Critical Hitrate?

I guess I did the math wrong and it works out perfectly. I guess that also means I need to give up this dDEX stuff . I was almost certian of it =/.

171 Critical Hits total... so 171/1500 * 100 = 11.4% critical hit rate~

I hadn't bothered posting it since it just confirmed what people already knew. But my initial testing on Robber Crabs in Kuftal showed PCC making my hit rate go from 32.35% -> 37.32%. (Sample size was 1k hits for each.) This confirms the 2acc->1% Hit Rate works /w lower evasion mobs.

EDIT: When I get the chance, I'll finish off my testing by getting an acc% for the Lv73 Steelsheels and seeing if the Level Correction is being applied correctly~

[Kirschy]

p=0.924, the 95% error is sqrt(p*(1-p))*1.96/sqrt(1,500)=+/-1.34%

Your test shows that with 95% probability, your real hit rate is between 91.06% and 93.74%.

edit : this does not confirm that level difference is -4 eva. If the model is correct (excluding level difference), it only allows integers and integers + 0.5 as hitrate, so only 91.5 92 92.5 93and 93.5 are allowed, and therefore it's -1,-2, -3, -4 or -5.

Yea i was unable to complete the test. But the first part shows our current understanding of accuracy is indeed holding up. (I've been skeptical for a while.) When I repeat the test /w a Lv73 crab, it should give us a relatively high confirmation on level correction~

[pchan]

Are you really ready to swing 10,000 times (40,000 for < 0.25% precision) ? Wouldn't it be easier to hit a mob several levels above you or below you ? This way you keep the same error and you have much less possible values in the interval.

[DancingEdge]

I am inclined to believe that evasion yields .9 returns past a certain skill level, likely 200.

At some point I will extensively test this but I've yet to have the time or energy to go round up half a zone of mobs to help people stop wasting money on EVA skill earrings. lolevaskill.

[TummieGaruda]

Just for testing sake, we can do a test against Lesser Colibri in Waj Woods which have 52 and 55 Agi. See how having different levels of Dex above the colibri pan out for final critical hit %'s. Then test those against Greater Colibri with different +Dex over their Agi and compare the results to see if there is indeed level correction like some suggest.

[FailureMidgard]

Just for testing sake, we can do a test against Lesser Colibri in Waj Woods which have 52 and 55 Agi. See how having different levels of Dex above the colibri pan out for final critical hit %'s. Then test those against Greater Colibri with different +Dex over their Agi and compare the results to see if there is indeed level correction like some suggest.

How is this any more useful than just switching your gear around for different dex values?

[Kirschy]

Are you really ready to swing 10,000 times (40,000 for < 0.25% precision) ? Wouldn't it be easier to hit a mob several levels above you or below you ? This way you keep the same error and you have much less possible values in the interval.

If I'd realli need 10,000 swings on the Lv 73 Crabs, I'm not up for it. They're annoying as hell. If I get the Evasion data on the Lv 62 crabs I used, will that suffice? (Have about ~1k swings.)

[pchan]

If I'd realli need 10,000 swings on the Lv 73 Crabs, I'm not up for it. They're annoying as hell. If I get the Evasion data on the Lv 62 crabs I used, will that suffice? (Have about ~1k swings.)

For 2 level difference, assuming the -eva is constant per level (call it x / level), if you get the same sample size, base acc and base hit rate etc, you would need to put 94-2*x*0.5 in the intervalle [91.05,93.7] instead of 94-1*x*0.5 previously. This time only x=1 or 2 are allowed and 2k sample size is most likely enough to exclude x=1.

Basically if your confident interval contains only one integer, you win (with 2 level difference), so yeah 2,000 swing should do it.

[ringthree]

How is this any more useful than just switching your gear around for different dex values?

Because that would not test for level correction.

[FailureMidgard]

Because that would not test for level correction.

Oh ya I'm dumb, haha

[Vodou]

I've never thought the level correction was designed to work the opposite direction (when players are higher than mobs.) Here is some data I collected while working out some test results between myself as a 75sam/37dnc to lower level mobs.

These are real numbers with some interpretation that many of you can probably correct. Please do so. Other than providing the actual data I had, I'm not 100% sure of my results.


1) being Hit by demon pawns / wizards in Xarc (levels 48-52) while I had 331 evasion and was not using any defensive abilities.

I was hit 47 out of 245 times (19%). I don't know what mobs acc is at those levels, but if it corresponds to players, it would be around 150. Calculating backwards (95-19) * 2 = 152; 152+150 = 302 is pretty close to my evasion of 331. So in theory they could hit me 95% of the time with acc of 302 versus my evasion of 331. Odd. Certainly they are not being penalized, if anything they are getting a bonus.

2) shooting at worms in altepa (levels 43-47) while using blind pots giving myself an effective racc of 71.

I landed 35 of 268 shots (13%). Again, I don't know the mobs actual evasion, but it appears as If I'd need (75-13)*2 racc to hit a 75% rate, i.e., 71+124 or 195 racc.

For comparison, a level 37 rng with capped skill will have 114 racc + 22 trait + maybe 20 off agi (a guess) or 156. Add some very nice gear for 35 and sushi and he could have 220 racc, which would be at 87% racc based upon my numbers. Sounds close to about right to me if one was xp-ing on those mobs at 37.

Seems to work like this:

when attacking a mob below your level, the mob has normal evasion (no level correction penalty)

when being attacked by a mob below your level, the mob gets a bonus and no level correction penalty


On a somewhat related note, I'm not seeing anything special in the critical rates against these mobs (25 levels below my level.) Pretty much looks like I get the same number of criticals as I do against HNM's.

[TummieGaruda]

I've never thought the level correction was designed to work the opposite direction (when players are higher than mobs.) Here is some data I collected while working out some test results between myself as a 75sam/37dnc to lower level mobs.

These are real numbers with some interpretation that many of you can probably correct. Please do so. Other than providing the actual data I had, I'm not 100% sure of my results.


1) being Hit by demon pawns / wizards in Xarc (levels 48-52) while I had 331 evasion and was not using any defensive abilities.

I was hit 47 out of 245 times (19%). I don't know what mobs acc is at those levels, but if it corresponds to players, it would be around 150. Calculating backwards (95-19) * 2 = 152; 152+150 = 302 is pretty close to my evasion of 331. So in theory they could hit me 95% of the time with acc of 302 versus my evasion of 331. Odd. Certainly they are not being penalized, if anything they are getting a bonus.

Few things. Mob Stats are calculated slightly differently than player stats (Agi, str, etc). It is Family Rank, Job Rank, Job/family bonus. Working backwards, 19% without level correction, would be 75% (Both the same acc and evasion) minus 19% = 56. Then 56*2 = 112. This would give the Acc and Eva difference. So 331 Evasion would mean 219 Accuracy for the mob. Now level 52 with A+ skill is 163, so that would be 163 Acc off the bat. That would mean 56 Acc is unaccounted for. Based on 2dex = 1 acc, that would mean that the Demon would have to have 112 Dex, which I don't believe to be true.

Looking at it closely, I think that there is indeed a level correction, but for mobs vs players it works in reverse. As stated previously, mobs stats are calculated differently, so an A in a particular stat gives a larger boost than to mobs than it does to players. Let's just assume for the time being that the those Level 52 Demons have about 66 Dex which would give the demon a total of 196 Acc (With A+ Skill).

So you have 331 Evasion. Let's also assume there is level correction but it adds accuracy say about +1 per level difference. 75 - 52 is a 23 level difference. So the mob has 196 acc + 23 = 219 acc. Since 75% hitrate = same acc and evasion without correction 75-19 = 56. 56% hitrate * 2 = 112 accuracy. 219 + 112 = 331 which is the players evasion.

ld;dr version:

When it comes to PC hitrate on mobs higher level than us, we are penalized an amount of accuracy that is 4*level difference of the mob we are fighting

When it comes to Mob hitrate on PC's and the mobs are a lower level than us, mobs are given an accuracy bonus based on 1*level difference. (This is just a theory)

[BerenTebogo]

I am inclined to believe that evasion yields .9 returns past a certain skill level, likely 200.

At some point I will extensively test this but I've yet to have the time or energy to go round up half a zone of mobs to help people stop wasting money on EVA skill earrings. lolevaskill.

In my experience - Elusive + Novia worked better for me. As NIN.