Using Quanta Index to determine the optimal number of Pillars in a deck

[omgarm]

This is very good for an objective look on decks. As Puppy pointed out the purpose of a deck can ask for a different QI, so claiming that a QI of 5 is optimal only goes for most speed decks I believe. Perhaps stall decks are better off suited with a QI of 7? Would SG's unupped rainbow give a QI of 7, assuming you count Quantum Pillars as 1/4 of each element. It would be a decent check.

I think I talked about this earlier. And like I said, that 5 was only a guess. And Yes, Quantum Pillars have a different optimal QI.

I know, but I mean you should take some basic decks that rely on stalling their way to victory. Perhaps the Aether deck did run on optimal QI because it's made to stall. If other stalling decks also run at roughly 7 then that problem is solved.

[Bloodshadow]

@SG: Xinef's method on Nova and Supernova makes perfect sense to me. For example, if you have an upped FFQ deck that uses , then when you cast Nova, it counts as -3 because your deck uses three elements out of all the elements Nova generate. If your deck is only composed of , then it only counts as -1. If your deck is a rainbow that uses 11 out of 12 elements, it counts as -11. Etc, etc. Quite a nice method actually.

Miracle doesn't need special treatment? All right.

Fractal... Hmm, maybe it counts as 9 plus 6x the to-be-copied creature? Generally you want to use Fractal when you have an empty hand, but that is not always the case; sometimes I'm forced to use Fractal on a Devourer when I have up to three other cards in my hand. I think 6x is a good average; ideally I can create 8x copies of the creature, but my minimally accepted value is 4x to count the Fractal as not wasted.

I don't know... The above might sound confusing, so don't stay up and drink coffee while reading it

[Scaredgirl]

This is very good for an objective look on decks. As Puppy pointed out the purpose of a deck can ask for a different QI, so claiming that a QI of 5 is optimal only goes for most speed decks I believe. Perhaps stall decks are better off suited with a QI of 7? Would SG's unupped rainbow give a QI of 7, assuming you count Quantum Pillars as 1/4 of each element. It would be a decent check.

I think I talked about this earlier. And like I said, that 5 was only a guess. And Yes, Quantum Pillars have a different optimal QI.

I know, but I mean you should take some basic decks that rely on stalling their way to victory. Perhaps the Aether deck did run on optimal QI because it's made to stall. If other stalling decks also run at roughly 7 then that problem is solved.

Well, the optimal QI changes depending of what kind of deck you are building. FG rainbow decks are a perfect example of this because they produce a relatively small amount of quanta.

It could be something like this:

QI 4 = speed deck
QI 5 = balanced deck
QI 6 = defensive deck

Then you could simply tweak the QI based on what kind of deck you are building, or where you are going to use it.

@SG: Xinef's method on Nova and Supernova makes perfect sense to me. For example, if you have an upped FFQ deck that uses , then when you cast Nova, it counts as -3 because your deck uses three elements out of all the elements Nova generate. If your deck is only composed of , then it only counts as -1. If your deck is a rainbow that uses 11 out of 12 elements, it counts as -11. Etc, etc. Quite a nice method actually.

It can't be that simple.

If I have a rainbow deck that uses all 12 elements, and I had 6 Novas, that would mean -72? And with Supernovas, -144? Can't be.

But one thing is clear: with Novas and Quantum Towers, how much you use different elements has to be a factor.

Miracle doesn't need special treatment? All right.

I don't know if it does but for me it sounds like it doesn't.

[TheMonolith]

Most reliable way to achieve a state of MASQUE (maximum synergistic quantum efficiency) = trial and error.









Yes, the MASQUE thing is cheesy.

[Xinef]

@SG: Xinef's method on Nova and Supernova makes perfect sense to me. For example, if you have an upped FFQ deck that uses , then when you cast Nova, it counts as -3 because your deck uses three elements out of all the elements Nova generate. If your deck is only composed of , then it only counts as -1. If your deck is a rainbow that uses 11 out of 12 elements, it counts as -11. Etc, etc. Quite a nice method actually.

It can't be that simple.

If I have a rainbow deck that uses all 12 elements, and I had 6 Novas, that would mean -72? And with Supernovas, -144? Can't be.

But one thing is clear: with Novas and Quantum Towers, how much you use different elements has to be a factor.

Yes, that's why I don't think my first idea can be used as is, but has to be changed to take use of particular elements into account.

Right now I think that you have to calculate QI separately for each element, for example if you run 6 Quantum Towers and 6 Supernovas in a 30 card deck, you expect to win in 7 turns, and the only card using in your deck is an arsenic:
- 7 turns => 15 cards
- this means that on average you will draw 15/30 * 6 = 3 supernovas during a typical game, that adds up to 6 quanta
- this means your arsenic can be considered a free card ( has a QI of 0 ) and you do not need any towers to play it.

On the other hand let's assume you include 2 lava destroyers in that deck
- supernovas give you 6 on average, so you still need to pay 4 with your towers (in case you draw both destroyers)
- Quantum Towers count as 1/4 of a Burning Pillar, so Quantum Index of in this deck is 4/(1/4 * 6) ~= 2,7

this leads to conclusion, that either QI of 2,7 is too much for speed rainbows, or we have to design an even better formula, because 10 is a bit too much for a speed rainbow.

Hmm... my final conclusion is that calculating QI as a ratio (total cost / total towers) might not be the best idea when we take novas/immolations/quantum pillars into account. It might work for mono decks, but for duo/trio/rainbow I guess a more complex formula has to be found, something like (total cost of cards played during a typical match / total quantum produced during a typical match). Both can be estimated if we know the size of the deck and how long a typical game lasts.

By the way, I could try to write a Java Applet to calculate QI (could be yours formula to calculate it, or mine, or even both), but if you prefer some other technology then someone else will probably do the job better.

[Scaredgirl]

By the way, I could try to write a Java Applet to calculate QI (could be yours formula to calculate it, or mine, or even both), but if you prefer some other technology then someone else will probably do the job better.

Sure, go for it.

[Snowstorm]

This might come in handy for you guys working on this.. I'v been working on a spreadsheet that shows probabilities of card drawing depending on your deck, and quanta curves (like, a deck setup like x will produce this much on average, and use this much on average). It might help some way to fixing up your QI.

I don't believe that assigning flat values to cards in decks can be used in this way.. decks have different aims and speeds, but i think by marrying up a curve of what you expect to use against what can be produced might help you to go some way towards working out ideal tower numbers.

[Bloodshadow]

I've just thought of something, and it might help making things like Nova easier.

When calculating QI, maybe we should do each element in the deck separately. For example, if you have a Fire/Earth duo deck, you should calculate the QI of the Fire part (total cost divided by number of Burning Pillars), and calculate the QI for the Earth part (total cost divided by the number of Stone Pillars), then add them together. If you have a rainbow deck that uses all 12 elements, then your deck is going to be divided into 12 components and calculated separately; how this works with Quantum Pillars, I'm not sure yet, but you could simply use the number of Quantum Pillars divided by three as the denominator in all 12 components.

How would this help with calculating Nova and Supernova? For example, in the Light section of your rainbow deck, you might only use a total of 6 for the Sundials; in that case, if you have 6 Supernovas, the generated by them is obviously going to count as -6 instead of -12, because excess Light quanta can't be used to pay for other elements. This way, 6 Supernovas would not count as -144.

BTW, don't forget your Mark; that should count as an extra pillar.

EDIT: Doh! I think I just saw Xinef saying basically the same thing as I said above. Ninja fail...

[coinich]

This is certainly an interesting tool, thanks!  I was working on a Duo deck, and while I found I prefer close to a 4.5 QI (due to my need to have excess bits of quanta for it to work) I was able to tailor it much faster!

Also, your commas confused me until this very day.  Across the pond, we use periods instead of those silly comma things.

[Essence]

Quantum Pillars and all Towers would have their own formula.

I've been, as an ad hoc rule, treating (non-Quantum) Towers as though they were Pillars with an "Ability Cost" of -1, and the results have been near-perfection.  For example, when I'm mindlessly pummeling AI3 with my mono-Water beatstick:

Code: [Select]

6rn 6rn 6rn 6rn 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gk 7gm 7gm 7gm 7gm 7gm 7gm 7gt 7gt 7gt 7gt 7gt
I used to run 14 Towers, 4 SoGs, 6 Crawlers, and 6 Dragons. 

The card cost was (11*6)+(4*6)+(4*2)=66+24+8=98.
The ability cost was -14, for a total cost of 84, /14 pillars =6 -- just the opposite of what SG suggested was the right 'swing' for a speed deck.


So I removed 1 Dragon and added 1 Tower, as above.

Now, the card cost is (11*5)+(4*6)+(4*2)=55+24+8=87.
The new ability cost is -15, for a total cost of 72, /15 pillars is 4.8 -- and the deck does, in fact, play noticably better for it.


The question is, can we talk PlanPlan into adding a new line to the bottom of the Deck Generator between the electrum cost and the author that describes the QI?

[Scaredgirl]

Quantum Pillars and all Towers would have their own formula.

I've been, as an ad hoc rule, treating (non-Quantum) Towers as though they were Pillars with an "Ability Cost" of -1, and the results have been near-perfection.

Hm.. that's interesting and might work. I need to try it.

As for asking Planplan to change the deck generator thing.. I think we should first fully establish what this QI is. This way we don't need to change the code each time we make changes to QI.

[Demagog]

First time I've seen this thread, and it is interesting... Not sure if this is what Essence was talking about, but subtracting one from the numerator for each tower would probably be a good adjustment.