Paladin FAQ, Info, And Trade Studies. |
||
|
Paladin FAQ, Info, and Trade Studies.
How do you pronounce Paladin?
Doing some random parses with block rate on the test server again, mainly just to see if the block rate formula is changed by player's own level, mob's level, or shield skill itself. I found an interesting glitch on the test server, which allowed me to set my Shield Skill to 0 as PLD99/WAR49 (triggered by changing jobs and adjusting level via the GM Moogle). Using this to my advantage, I went to Buburimu Peninsula and found a random Sylvestre (Mandragora). According to wiki, these range from levels 15-19.
My shield skill started at 8 (forgot to remove shield skill merits), & was level 10 when I ended the parse. This is a small quick parse, but just enough to get an idea of Aegis's block rate against a Lv.15-19 mob with only 8-10 shield skill. Hits: 61 Blocks: 24 Block Rate: 39.34% It is evident that the formula: Shield_Base_Rate + ((Shield_Skill / 4) - (Mob_Level / 0.5)) is not anywhere near accurate on lower level mobs, as it would work out to something like this: 142 + ((8 / 4) - (17 / 0.5)) 142 + ((2) - (34)) 142 + -32 110% Block Rate, which would cap Aegis at 75%. However, this is obviously not the case. It seemed to me at this point it was pretty evident that player's own level has absolutely no effect on block rate, as predicted in a past parse by Martel (Post Link). So once again, the only two factors that seem to have any direct impact on block rate are: player's shield skill, mob's level, and unknown variables used in calculation of total block rate. After getting the results from this parse, I started to rework the formula a bit & came up with something that seems interestingly accurate (also similar to the Hit-Rate formula): Base_Block_Rate + (Shield_Skill - A+_Rank_At_Mob's_Level) * Skill_Gain_Multiplier Skill_Gain_Multiplier would equal the amount 1 shield skill would add to block rate. For levels 60+, this seems to be around 0.23-0.27. (our 0.25% block rate = 1 skill rule of thumb model). A+_Rank_At_Mob's_Level just means an A+ skill value the mob would have for that level. For example, a level 99 mob would have 424 skill in a A+ ranked skill. Base_Block_Rate is the initial percentage a shield's block rate starts at. For Aegis, this would be 50. For a size 3 Koenig Shield, this would be 45. For Ochain, I would assume 107 or 108, but I'm not 100% sure on that. So using 432 shield skill for example (Lv.99 PLD's 424 skill + 8 merits) paired with an Aegis, and a Lv.99 mob (which would have 424 in a A+ skill): 50 + (432 - 424) * 0.25 50 + (8) * 0.25 50 + 2 52% Block Rate, which matches a recent parse's data (link). Plugging in Martel's recent Aegis test data from here, and assuming he had 432 skill along with the mob being Lv.104: 50 + (432 - 459) * 0.25 50 + (-27) * 0.25 50 + -6.75 43.25 I think this new formula does a much better job at predicting the block rate than the previous one. What do you guys think? For lower levels, the formula most likely has some changed variables, such as the lower levels (Lv.10s range) parsing at 1 shield skill = 0.15% block rate gained, versus the 0.25% at Lv.60+. Parse Data One more parse for today using the lowered shield skill exploit:
My Job/Level: PLD99/WAR49 Shield: Aegis (99, with Afterglow) Shield Skill: 253 Buffs: None No equipment other than the Shield was used (completely naked). This remained static throughout the entire parse. Mob: Caedarva Leech (Lv.63, this was checked by changing to a Lv.63 job & checking it as an Even Match) Hits: 1230 Blocks: 724 Block Rate: 58.86% Using the new block rate formula: 50 + (253 - 217) * 0.25 50 + (36)*0.25 50 + 9 59% Block Rate Difference from projected result: 0.14% Edit: Forgot to mention that if you also average out the damage taken from blocked critical hits (higher damage than normal hits), you get 83% as the reduced amount. This is identical to Martel's 81.1% parsed value (link). We'll have to find harder hitting mobs to find the exact value I think. Lower damage taken greatly skews the result of finding blocked damage reduction's true value. Parse Data Ragnarok.Martel
Offline
So, after looking at your latest block rate formula, I decided to apply it to some of my various skill tests. Mostly, I have lots of Ochain data, and a very little bit of kite shield data.
Ochain Code Ochain90 Base 107 Multiplier 0.25 Mob Skill 459(lvl 104)* Skill 369 374 379 384 389 394 399 404 409 414 419 421 424 429 430 431 432 433 Parsed% 86.2 86.92 88.29 90.12 91.16 91.14 93.59 94.05 94.73 96.5 97.1 97.51 98.11 99.51 99.55 99.77 99.92 100 Calc'd% 84.5 85.75 87 88.25 89.5 90.75 92 93.25 94.5 95.75 97 97.5 98.25 99.5 99.75 100 100.25 100.5 Diff 1.7 1.17 1.29 1.87 1.66 0.39 1.59 0.8 0.23 0.75 0.1 0.01 -0.14 0.01 -0.2 -0.23 -0.33 -0.5 Assuming the skill/lvl remains 7/per after lvl 99. It looks very close above 400 skill. If a tiny bit ahead of actual rate. But below 400 skill, a gap seems to form, that gets larger as skill gets lower. Perhaps 400 skill is a break point at which the skill multiplier changes? At a 0.25 multiplier, above 400 skill have a avg difference of 0.04 between parsed and calculated rates. Whereas the below 400 samples have a avg diff of 1.38. If we change the multiplier to 0.23 on the less than 400 skill samples then the avg difference becomes 0.11. Much closer. I was only able to find 5 different skill tests for size 3 shields. 1 below 400 skill, and 4 above. With really random jumps in skill between them. I think I'd like to gather more samples before trying to draw any conclusions from size 3 data. Hopefully, the Ochain data will be helpful in adjusting/verifying your formula. I might do some variable skill aegis/size3 tests later. Since Ochain caps at 433 skill we can't really look for changes above that with Ochain. volkom said: » How do you pronounce Paladin? like aladin with a p Leviathan.Dragonlord
Offline
volkom said: » How do you pronounce Paladin? pal-uh-din http://www.howjsay.com/index.php?word=paladin Ragnarok.Martel said: » It looks very close above 400 skill. If a tiny bit ahead of actual rate. But below 400 skill, a gap seems to form, that gets larger as skill gets lower. Perhaps 400 skill is a break point at which the skill multiplier changes? At a 0.25 multiplier, above 400 skill have a avg difference of 0.04 between parsed and calculated rates. Whereas the below 400 samples have a avg diff of 1.38. If we change the multiplier to 0.23 on the less than 400 skill samples then the avg difference becomes 0.11. Much closer. But one thing that has me curious is this parse from above: link With shield skill at 253, you would expect a much larger gap assuming there was a formula change at around 400 skill. Perhaps there is some kind of Skill_Gain_Multiplier change when the Shield Skill value gaps far from the A+_Rank_At_Mob's_Level value? Regardless, I'm glad we were able to find a formula that calculates our block rate within a 1-3% margin of error (so it seems anyway), with thanks to everyone who helped collect parse data and helped contribute. If we don't find any major quirks like from the previous formula, we could begin adding minor adjustments to it until we get a near perfect prediction. Went back to observe previous parses, and I noticed that when raising shield skill, the amount of block rate that was gained per skill ranged from 0.21%-0.29% (each parse varied), but they all average together for about 0.2454% (0.25%?). I didn't observe every single posted sample, but about 9 of them.
It could be due to variance, or that the Skill_Gain_Multiplier is changing based on certain conditions, as Martel mentions above where the Skill_Gain_Multiplier seems to change from 0.25 to 0.23 when shield skill drops below 404. The Skill_Gain_Multiplier might be undergoing a linear scale based on how far away your Shield Skill is from the A+_Rank_At_Mob's_Level value, but that's only pure speculation for now. I created a spreadsheet containing all block rate parse data that we've accumulated thus far. You can view it at Google Docs: Google Docs Spreadsheet Link or download the Excel document directly if you prefer viewing the original: Mediafire Document Download Link.
The spreadsheet contains the following parses (does not include parses I did to find block rate floor/cap): Link #1 Link #2 Link #3 Link #4 Link #5 Data: Data #1 & Data #2 Link #6 Link #7 Link #8 Link #9 Link #10 Link #11 Link #12 Data: Data #1 Link #13 With all data in hand, does anyone have an idea how to convert it into a working block rate formula other than the current one I came up with thus far? Base_Block_Rate + (Shield_Skill - A+_Rank_At_Mob's_Level) * Skill_Gain_Multiplier The above works pretty well so far, but it's still not perfect (the projected result slowly skews away from the actual result as the Shield_Skill value progresses far from the A+_Rank_At_Mob's_Level value). Big thanks to Martel for gathering a massive amount of data & sample size on Ochain block rates. It seems we have plenty of shield skill variation tests, but lack mob level variation tests (change mob level but keep shield skill the same), otherwise I believe we have plenty of data to finally come up with a working block rate model. Though current data shows that every 1 level the mob rises/decreases will decrease/rise block rate by 2%. I'm gonna try to pump out some parses regarding the effects of mob level vs shield skill. The shield skill level I'll be using is 424 for each test. On the test server, a player can set their level to any value ranging from 1-99, so I'll be using this to change my level in order to find the exact level of the mob I'm gonna be parsing against (Even Match = mob is the same level as me). When level is lowered, base shield skill is as well, so I'll use gear/adjust merits in order to reach the 424 skill rank again. Since parses have shown player's own level has no direct effect on block rates, this should provide pretty accurate results with the ability of finding the mob's exact level without any uncertainty.
Here's the first two parses so far: Parse #1 My Job/Level: PLD98/SCH49 Shield: Aegis (95) Shield Skill: 432 Buffs: All Cruor Buffs, Regen Atmas, Light Arts, Regen II Equipment remained static throughout the entire parse. Mob: Surveyor (Even Match, Lv.98) Hits: 5,165 Blocks: 2,738 Block Rate: 53.01% Parse #2 My Job/Level: PLD97/SCH48 Shield: Aegis (95) Shield Skill: 432 Buffs: All Cruor Buffs, Regen Atmas, Light Arts, Regen II, Sublimation Equipment remained static throughout the entire parse. Mob: Surveyor (Even Match, Lv.97) Hits: 5,540 Blocks: 3,138 Block Rate: 56.64% Block Rate Difference: 3.63% Parse Data Parse #1Parse #2 One more parse for today:
My Job/Level: PLD96/SCH48 Shield: Aegis (95) Shield Skill: 432 Buffs: All Cruor Buffs, Regen Atmas, Light Arts, Regen II Equipment remained static throughout the entire parse. Mob: Surveyor (Even Match, Lv.96) Hits: 5,555 Blocks: 3,148 Block Rate: 56.67% Hardly a change at all from the Lv.97 sample (0.03% difference). Variance? Maybe. Kind of reminds me of this scenario though: link, where +15 skill yielded no result increase whatsoever. Also worth noting that the jump from a Lv.97 mob to a Lv.98 mob yielded a 3.63% block rate increase, as if it was covering the gap for the no increase from Lv.96 to Lv.97. Parse Data Hey all, been messing around with the block rate math again, and I came up with another formula using a standard linear formula structure.
Block Rate = Skill_Modifier(Shield_Skill) + Mob_Level_Modifier(Mob_Level) + BASE For levels 91+ Skill_Modifier = 0.0022 Mob_Level_Modifier = -0.01876 (rough estimate) Shield BASE Aegis = Ochain = 2.0 Size 3 = Size 4 = 1.21? Size 1 = ? Size 2 = ? Levels 90 and below seem to use a different Skill_Modifier & Mob_Level_Modifier than 91+. When I plug our parse numbers in for level 91+ mobs, it seems to yield results within a 1-2% accuracy difference. Thoughts? Mob_Level_Modifier might be slightly off as it was based off the limited mob level difference parses I did recently. I got the Skill_Modifier by averaging the skill gain from massive amount of parse data acquired from Martel's Ochain samples. Edit: Corrected Aegis/Size 3 BASE, it seems like it was 0.01 off. I have a question about the CDC set on the front page concerning the gorget and atheling combo. As a crit WS wouldn't either rancor collar or rancorous mantle do better in one of those slots? Perhaps mantle and gorget?
Fenrir.Yuriki said: » I have a question about the CDC set on the front page concerning the gorget and atheling combo. As a crit WS wouldn't either rancor collar or rancorous mantle do better in one of those slots? Perhaps mantle and gorget? Leviathan.Dragonlord
Offline
Asura.Lokimaru said: » Fenrir.Yuriki said: » I have a question about the CDC set on the front page concerning the gorget and atheling combo. As a crit WS wouldn't either rancor collar or rancorous mantle do better in one of those slots? Perhaps mantle and gorget? The front page does state "Vorpal/CDC (ring/neck/earring change based on WS)." under the WS set. Also, rancor collar has 6 acc so you're only losing 4 acc by switching ele gorget to rancor. Atheling+rancor collar, or Rancorous mantle+Nefarious collar are going to be the top 2 contenders for CDC. It comes down to whether 3 atk and 3 DA beats 3 crit rate. Atheling/Rancor for CDC for almost all situations currently.
Phoenix.Neosutra said: » Atheling/Rancor for CDC for almost all situations currently. Ty muchly. Got a question for a friend who is playing a Paladin. He's working on both his relic and Empy Shield but until then, what are some good shields to be using for tanking puproses?
Phoenix.Tsuekasa said: » Got a question for a friend who is playing a Paladin. He's working on both his relic and Empy Shield but until then, what are some good shields to be using for tanking puproses? What is he using Pld for in today's content, without emp/aegis? Our LS is starting to venture into Abby atm so we are no where near Endgame yet so just havn't had enough time to gather enough to get either shield.
Oh in abyssea? I wouldn't worry about what shield to use on Pld in abyssea, as Pld doesn't belong in abyssea.
So from the sounds of it, Pld is more of a DD in Abby atm so shield should reflect being DD instead (until relic and empy shield are made)
The only thing PLD is good for in Abyssea is solo cleaving with Aeolian Edge and Ochain. If that's not what you're bringing a PLD for, have them change jobs.
Sadly, Pld is their only 99 atm, and LS is not very tank heavy atm. I have a Mnk but not geared enough yet for abby tanking (working on it though, black belt is going to be a pain) and really don't have any Nin's that arn't being used as subs. He is thinking of taking Red to 99 next.
Any MNK that has a counterstance macro can tank in abyssea as long as whm and mnk both have at least two decent atmas.
Edit: In theory. People still manage to mess up MNK+WHM somehow. Also, "this is my only job" is pretty irrelevant in this day in age, when you can 99 a job in a day, max gear it in a week, and put a relic on it in 1-2 months (solo).
Phoenix.Neosutra said: » Also, "this is my only job" is pretty irrelevant in this day in age, when you can 99 a job in a day, max gear it in a week, and put a relic on it in 1-2 months (solo). |
|
All FFXI content and images © 2002-2024 SQUARE ENIX CO., LTD. FINAL
FANTASY is a registered trademark of Square Enix Co., Ltd.
|