Possible theory about the ATA-EVP formula?

[Broken_L_button]

Well, once again, more spam from B_L_B. After so much time after its release, one has yet to provide a plausible formula to calculate the chance of hitting (or missing) an enemy. Well...I've done a bit of brainstorming, and here's what I've come up with:

*IMPORTANT!! WHAT FOLLOWS BELOW IS JUST PURE CONJECTURE!! NONE OF IT IS PROVEN! YOU HAVE BEEN WARNED!!*

So, since hitting or missing is a "chance", I'd say the main part of the whole formula would be a ratio. Thus, at the formula's core, we have:

(Character ATA/Enemy EVP)*100 = Chance of hitting (in %)

So, if your ATA is higher than the enemy EVP, logically, you won't miss. Also, this kind of formula explains why you still end up hitting high EVP enemies even if your ATA is abysmal (like when a FOney manages to hit an Usanimere with a slicer after missing countless times). But, I'm sure you guys are thinking this:

"Lolwut? A simple fraction? What about the fact that hard attacks miss a lot?!"

Well, I'm getting there.

We all know that hard attacks get, if memory serves, ther power multiplied by 1.4, but have lowered ATA (totally logical; giving the same accuracy to normal and hard attacks would make normal ones useless). Thus, while calculating the chance of hitting the enemy, if a hard attack is used, a certain multiplier is used. Since the increase in power is 40%, logic would suggest an ATA reduction ranging from 30-40%. So, for hard attacks, the ATA/EVP formula becomes:

((Character ATA/Enemy EVP)*100)*(1-0.4<-Possible Hard attack ATA reduction) = Chance of hitting (in %)

And now, you're thinking:

"Eh, whatevar. What about PAs?"

And I'm getting to that too.

Since PAs have their own ATA modifiers (shown in the weapon's information screen), logically, they are applied directly to the formula as such:

((Character ATA/Enemy EVP)*100)*(PA ATA modifier/100) = Chance of hitting (in %)

Now you're thinking:

"Wow, you have time to waste, man. Anyways, what about the fact that the third hit in a combo is more accurate?"

And I'm getting to that as well.

In PSZ, to make hard attacks easier to connect, it was common to use NNH as the basic attack combo, since NNN lacked power. My guess it that the final hit, even as a hard attack, has the ATA of a normal hit, thus, in the case of a third hit in a combo, we have:

((Character ATA/Enemy EVP)*100)*(1+0.4 <-third attack ATA modifier) = Chance of hitting (in %)

And, if you put all that in at once to make a general ATA/EVP formula, you get:

-Normal attack: ((Character ATA/Enemy EVP)*100)*(1+Third attack ATA modifier<-if it can be applied)= Chance of hitting (in %)

-PA: ((Character ATA/Enemy EVP)*100)*(PA ATA modifier<-as shown in the weapon info- /100) = Chance of hitting (in %)

-Hard attack: ((Character ATA/Enemy EVP)*100)*(1 - Hard attack ATA reduction + Third attack ATA modifier<-if it can be applied)= Chance of hitting (in %)

Well, that's all folks. You may comment if you want to. And, yes, I have lots of time on my hands lately.

[Chukie sue]

I'll read later ._.

BLB! Would you like to help me write a sticky guide for this game dealing with ATA/EVP "magic numbers" (you may add your formula of course) FAQ's like "where is teh sewers?" "I always miss on SH, how can I stop?" and basic character builds?

[Broken_L_button]

Sure. But I doubt I'll be able to help much; that formula is mainly conjecture, and it seems the "magic ATA/EVP numbers" are all well known...Then again, ask what you need and I'll do my best to help.

[Seth Astra]

Hmm... Interesting. I think I read somwhere that a H attack has 1.3 time the damage, but 70% of the ATA. Good job with all this, though.

[Broken_L_button]

Hmm... Interesting. I think I read somwhere that a H attack has 1.3 time the damage, but 70% of the ATA. Good job with all this, though.

Thanks! I guess I'll have to search a bit to get that exact mulitplier...Or just test it out by whacking at the same monster 100 times.

EDIT: I think I've found another argument that might make this more plausible; the fact that the bosses only have 1 point of EVP. If the game effectively divides your ATA by the enemy EVP to calculate hit rate, if the boss EVP was 0, there would be only 2 answers to that equation: "OMG!! ERROR!! SYSTEM CRASH!" or infinity (dividing a number by another one that's infinitely smaller -for example 1/0.000000...01- results in an infinitely large number, thus infinity as the result). Sine the chance of hitting a target is a "chance", infinity can't be a valid choice...Besides, I doubt that such a value could be used in any game's programing; there's a limit to the DS' and the cartridge's memory, limiting variables to finite values. Once again, just my two cents.

[Chukie sue]

Sure. But I doubt I'll be able to help much; that formula is mainly conjecture, and it seems the "magic ATA/EVP numbers" are all well known...Then again, ask what you need and I'll do my best to help.

Well known, but if it will prevent even one "Icant hit" topic I'll be satisfied.

[Broken_L_button]

I see...Well, I guess it's time to hit the same enemy 100 times at different ATA values, calculating the hit rate for each ATA value, make a graph of ATA vs. hit rate, make a linear regression, and use the obtained equation to deduce enemy EVP. Now to do that with every single mob, lol, XD.

[Chukie sue]

...have fun!


I'll start writing up a sticky* whenever I have the time... It's something I want to do but cant find he time for... Im sure you can relate.

*hopefully

[Vorlik]

BLB, are you really going to run regression on this? That'd be neat, for sure. I've playtested minus the regression before back on GFAQ's, and ended up with totally inconclusive results, while others claimed that indeed to hit near 100% of the time, all that was required was ATA at the same level as an enemy's EVP.

The 520 number is said magic number, and it's relatively reliable, such that if your theory here is correct it results in an 88% hit rate with normal attacks in SH areas like Dark Shrine and Arca. Counter intuitive to the damage formula, I actually believe that PA ATA modifiers apply only to the weapon ATA, rather than as a final modifier like in the damage formula. Also, I'm not sure if there's simply a final ATA boost to the final attack of a combo or if that increases incrementally such that the second attack is more accurate than the first and the third is more accurate than the second.

It would seem to me that the combo accuracy boosts are substantial, if we are assuming that a heavy attack relies on a 40% ATA decrease at some point (in conjunction with the 40% increase to damage). For example, three heavy attacks in a row are generally miss, 50/50 miss or hit, and then nearly always a hit on the final combo. Obviously this is only within reason, I'm not saying it's a guaranteed hit, but anyway.

Final thing I'd say in testing this is that it might be easier to start using PA's, determining what the PA ATA modifier acts on, and then going from there. PA's are the most commonly used attacks at high level, and combo string modifiers are entirely unknown, whereas PA's at least list their boost.

[Broken_L_button]

Yeah. Looking over the Damage formula made me realize the PA mod should be applied differently. What I was planning to do was test it only with 1 normal attack at a time, without chaining, get a set of results, make a graph, and deduce the enemy EVP from the linear regression. If the correlation is high enough, that would prove that the EVP claculated would be right, and that the formula could be plausible.

Then, with that calculated EVP, I'd test PAs on that same enemy and deduce how the PA mods are calculated. But your way of doing it seems faster with a slightly lower degree of incertainty on the results. Well, all I can say for now is that I'll finally get to put those Divine/Hits to use.

[Vorlik]

Yeah. Looking over the Damage formula made me realize the PA mod should be applied differently. What I was planning to do was test it only with 1 normal attack at a time, without chaining, get a set of results, make a graph, and deduce the enemy EVP from the linear regression. If the correlation is high enough, that would prove that the EVP claculated would be right, and that the formula could be plausible.

Then, with that calculated EVP, I'd test PAs on that same enemy and deduce how the PA mods are calculated. But your way of doing it seems faster with a slightly lower degree of incertainty on the results. Well, all I can say for now is that I'll finally get to put those Divine/Hits to use.

Enemy EVP values are already known. Here's a link to the topic I did on this back on GFAQ's back in the day, where Chaos listed all EVP values on SH for reference in the 8th post (counting my topic post) : http://www.gamefaqs.com/boards/953869-phantasy-star-zero/53503293.

Echo has the guide now and is in the process of slowly getting all the enemy bestiary info onto a spreadsheet, which combined with a damage formula spreadsheet I've made, we might be getting made into a damage calculator for PSZ. An accurate accuracy formula still eludes us though.

[Broken_L_button]

I see!! Thanks bunches! Now I have a way of verifying my calculations!