[Adalast]

i was bored while farming this afternoon, so i decided that i had never seen any mana conserver testing. i didn't get finished, but i figured that i would put up what i have found out.

~~~The Experiment~~~

☼ 50 trials of each
☼ one spell cast for each mob
☼ same spell each cast per trial set
☼ tested using blmbot

~~~The Results~~~

--0 Maneuvers--
Spell: Thunder II
Cost: 86 Mp
50 trials, 10 activations = 20%
Min activated cost: 48 Mp
Max activated cost: 80 MP
Activated list: 48, 53, 53, 53, 59, 59, 64, 69, 75, 80

--1 maneuver--
Spell: Absorb-INT
Cost: 33 Mp
50 Trials, 18 activations = 36%
Min activated cost: 16
Max activated cost: 30
Activated list: 16, 16, 16, 16, 22, 24, 24, 24, 26, 26, 26, 28, 28, 28, 28, 30

--2 maneuver--
Spell: Absorb-INT
Cost: 33 Mp
50 Trials, 20 activations = 40%
Min activation cost: 16
Max activation cost: 28
Activated list: 16, 16, 16, 16, 18, 20, 20, 20, 20, 20, 22, 22, 22, 22, 24, 24, 24, 26, 26, 28

~~Conclusion~~

☼ Base effect = Conserve MP +20
☼ Each Maneuver = +10, proven by the fact that 2 is exactly 40

i know that i need to take a larger sample on all of them to narrow the % better, but so far this is doing well i think. the 6% on the 1 maneuver test is rather substantial, but for now that will be attributed to being very lucky. since this is a random chance to happen with each cast, not an over all %, it can make some odd happenings. ok... i'll do more tests later.

[Auraeon]

Your post was merged with my topic in the fanart section somehow. KI really sucks. :/


**edit**
Very interesting tests.

[magusalley]

Adalast posted this on Allakhazam as well. I posted some regression results using this data so I'll repeat it here for people that don't go over to alla.

Well you've basically give us enough information in this post. And the best method of analysis it probably regression, for what we want to know. SO: I'll report them here (it repeats some of the things you've said in your conclusion)

Regression 1 (Activation)

Logit Regression:

Logit Z = -1.3261 + 0.48067*(# of Dark Maneuvers)
St. Dev: (0.30935) (0.22407)


Probability of activation = 1/(1+e^-z)

Marginal Effect of Dark Maneuver = 0.101

Okay so this tells us:

Each Dark Maneuver = +10% activation rate (like you said)

It also tells us:

Probability of activation (0 Dark Maneuvers) = 0.2098
Probabliity of activation (1 Dark Maneuvers) = 0.3003
Probabliity of activation (2 Dark Maneuvers) = 0.4100
Probabliity of activation (3 Dark Maneuvers) = 0.5289

Regression 2 (Percentage Reduction)

%Reduction = 0.054131 + 0.0434*(# of Dark Maneuvers)
St Dev: (0.02069) (0.0161)

So this tells us:

1 Dark maneuver = +4.34% MP Saved

0 Dark Maneuvers = Average MP Reduction of 5.41%
1 Dark Maneuvers = Average MP Reduction of 9.75%
2 Dark Maneuvers = Average MP Reduction of 14.09%
3 Dark Maneuvers = Average MP Reduction of 18.43%

In conclusion

Please note that all the marginal values in both of these regressions were statiscally significant within 99%.

This means that given an infinitely long data set we can say that these averages will hold true within about +- 1.5%. Since we know SE works with whole numbers we can conclude the following with 99% accuracy.

Base MP conserve = 20% activation rate; 5% MP Saved

Each Dark Maneuver = +10% activation rate; +5% MP Saved

[ZahlSechs]

If this is true, Mana Conserver sounds really good then, even more because it's active with 0 maneuvers also.
I never pictured Mana Conserver to be this good honestly (Conserve MP +20 WITHOUT maneuvers is a lot...), plus I think it's only 1 dark element slot, which makes it very useful for several combinations.

[magusalley]

Adalast is doing some more tests on the Mana Conserver now. He wants to see if light maneuvers, damage guage and the optical fiber have any significant effects on the activation rate. I'll be running regression on that too. Anyway I'm sure he'll post the results here when he is done as well.

Yeah the Mana Conserver was always a pretty good attachment IMO. Conserve MP +20 isn't too surprising, the BLM Conserve MP trait is +25