Hit rate equation

[Pillan]

I’ve been testing the ATA-EVP relationship and my tests resulted in an equation that reads as follows:

[hit rate of X] = [ATA of X]/([ATA of X]+[EVP of target]/5)

Or, conversely:

[dodge rate of X] = [EVP of X]/(5*[ATA of attacker]+[EVP of X])


This explains why additional ATA seems to have no effect at some point and why Newmans don’t block much more than Casts when they’re the same class.

Let me know if anyone else has found a different relationship between the two stats that has a better fit.


Edited to make it clear.

<font size=-1>[ This Message was edited by: Pillan on 2007-07-12 09:46 ]</font>

[Zorafim]

Hm, you should specify between enemy stats and ally stats. Is that your own evp that's in the equation, or the monster's?

Oh, and how were you able to test the findings? Evading seems so random.

<font size=-1>[ This Message was edited by: Zorafim on 2007-07-11 23:21 ]</font>

[Soukosa]

A character with 516 EVP has almost an almost perfect block rate on a Lv 16 Kog Nadd which has roughly 55 ATA. So let's do the math.

Hit rate formula:
55 / (55 + 516 / 5)
55 / (55 + 103)
55 / 158
0.3481
Assuming that's about a .35% chance of hitting, that could work.

Block rate formula:
516 / (5 * 55 + 516)
516 / (275 + 516)
516 / 791
0.7177
Uh... that doesn't work.

[SolomonGrundy]

On 2007-07-11 23:30, Sounomi wrote:
A character with 516 EVP has almost an almost perfect block rate on a Lv 16 Kog Nadd which has roughly 55 ATA. So let's do the math.

Hit rate formula:
55 / (55 + 516 / 5)
55 / (55 + 103)
55 / 158
0.3481
Assuming that's about a .35% chance of hitting, that could work.

Block rate formula:
516 / (5 * 55 + 516)
516 / (275 + 516)
516 / 791
0.7177
Uh... that doesn't work.



sounoumi - which enemies have the lowest ATA?

[Pillan]

On 2007-07-11 23:20, Zorafim wrote:
Hm, you should specify between enemy stats and ally stats. Is that your own evp that's in the equation, or the monster's?

Oh, and how were you able to test the findings? Evading seems so random.

<font size=-1>[ This Message was edited by: Zorafim on 2007-07-11 23:21 ]</font>


If you’re calculating the monster’s hit rate, then use its ATA and your EVP. If you’re calculating your own, then use your ATA and the monster’s EVP. Sorry that I didn’t make that clear from the beginning.

I tested it while I was bullet leveling (all elemental crossbows and mechs to 30 today) in VoC S. I simply counted the number of misses and number of shots and then looked for a formula that fit the data.

On 2007-07-11 23:30, Sounomi wrote:
A character with 516 EVP has almost an almost perfect block rate on a Lv 16 Kog Nadd which has roughly 55 ATA. So let's do the math.

Hit rate formula:
55 / (55 + 516 / 5)
55 / (55 + 103)
55 / 158
0.3481
Assuming that's about a .35% chance of hitting, that could work.

Block rate formula:
516 / (5 * 55 + 516)
516 / (275 + 516)
516 / 791
0.7177
Uh... that doesn't work.



The block rate equation is just the hit rate equation with EVP on top instead of ATA. You should always get 1-[hit rate] with that calculation. I put those numbers into a calculator myself and got 0.6523. If one is right, the other has to be.

[panzer_unit]

So Zoldeel >>>>> Zodial for boosting evade rates?

[Pillan]

On 2007-07-12 09:44, panzer_unit wrote:
So Zoldeel >>>>> Zodial for boosting evade rates?


Since enemy ATA is usually higher than your EVP/5, I’d say so, but it never hurts to just do both. I mean, you only cast Zodial once every 2-4 minutes.

[panzer_unit]

On 2007-07-12 09:43, Pillan wrote:
I tested it while I was bullet leveling (all elemental crossbows and mechs to 30 today) in VoC S. I simply counted the number of misses and number of shots and then looked for a formula that fit the data.


Where did you get monster ATA and EVP from?

[Pillan]

On 2007-07-12 10:05, panzer_unit wrote:
Where did you get monster ATA and EVP from?


Good old PSUpedia (http://psupedia.info/Monsters).

[chibiLegolas]

On 2007-07-12 10:10, Pillan wrote:
On 2007-07-12 10:05, panzer_unit wrote:
Where did you get monster ATA and EVP from?


Good old PSUpedia (http://psupedia.info/Monsters).


And their source is Sega's released Perfect Bible Stradegy type book I assume?

[SolomonGrundy]

so robots tend to have a low ATA modifier. I guess the L10 fire robots would be a good place to condcut enemy ATA vs your EVP testing.


@ panzer: you played PSO. Which had a larger affect, shifta, or zalure? Thing is, you can ALWAYS shift your self, you can'y always zalure enemies (bosses, etc).

[panzer_unit]

On 2007-07-12 09:48, Pillan wrote:
Since enemy ATA is usually higher than your EVP/5, I’d say so, but it never hurts to just do both. I mean, you only cast Zodial once every 2-4 minutes.


Doing math for it, buffs and debuffs have the exact same effect on evasion rate. Stupid multiplication and ratios... there's no sneaky min/maxing.

lv100 monsters have base 250 ATA
my 80/10 cast PT has 200 EVP + 250 from Rabol Asted

base: 250 ATA / 340 ATA + my EVP/5 = 73%
zodiaride: 250 / 353 = 70%
barada maga lv21: 212 / 302 = 70%
both: 212 / 315 = 67%

[Pillan]

On 2007-07-12 11:19, panzer_unit wrote:
Doing math for it, buffs and debuffs have the exact same effect on evasion rate. Stupid multiplication and ratios... there's no sneaky min/maxing.

lv100 monsters have base 250 ATA
my 80/10 cast PT has 200 EVP + 250 from Rabol Asted

base: 250 ATA / 340 ATA + my EVP/5 = 73%
zodiaride: 250 / 353 = 70%
barada maga lv21: 212 / 302 = 70%
both: 212 / 315 = 67%


Yeah, after I tried the calculation myself, I realized that the buff and debuff don’t do enough to the equation to change it either way. However, Zodeel begins winning when you use the 21+ techniques, but only by 0.8% (69.8% -> 69.0%). If you estimate with effect 5 buffs and debuffs, that deviation goes to like 2%.

If you want an easy trick, if enemy ATA is greater than your total EVP/5, Zodeel always wins. The only problem is that it doesn’t win by enough to notice. Who cares if it’s winning by less than 1 in 100?

[Soukosa]

On 2007-07-12 09:43, Pillan wrote:
The block rate equation is just the hit rate equation with EVP on top instead of ATA. You should always get 1-[hit rate] with that calculation. I put those numbers into a calculator myself and got 0.6523. If one is right, the other has to be.

Ah I see. Yeah, I miscalculated that last one. Didn't have my nice TI-92+ with me at the time http://www.pso-world.com/images/phpbb/icons/smiles/anime2.gif

Well I say you have a rather plausible formula. It works in the way I figured it would since the formula for the damage of a skill on RO that caused the same sort of effect looked in this way.

With using the stats from the perfect bible, you should keep in mind that they are now out of date. Everything past lv 20 could have changed though not all of them have. ATA and EVP should be fairly close to what they are now if they did change up til about lv 80 at least though. Wish I could create updated versions of those stats but it'd be fairly hard to test it, of course.

[Pillan]

On 2007-07-12 14:28, Sounomi wrote:
With using the stats from the perfect bible, you should keep in mind that they are now out of date. Everything past lv 20 could have changed though not all of them have. ATA and EVP should be fairly close to what they are now if they did change up til about lv 80 at least though. Wish I could create updated versions of those stats but it'd be fairly hard to test it, of course.


That’s true, but even if they changed the enemy EVP and ATA stats by 10%, you wouldn’t notice much of a difference from the way the governing equations work out. You could never get a perfect equation just because of the chance involved, so I’m happy with anything that gives a very good estimate.

I’m a biochemical engineer. Any equation within 20% accuracy is acceptable and within 5% is godly. This one is under that 5% inaccuracy range.



<font size=-1>[ This Message was edited by: Pillan on 2007-07-12 15:04 ]</font>

[Starrz]

Heh, thanks for the equation. I'll see if I can test it further.

[Starrz]

Leh Bump.

Ok, testing done. While your formula is very accurate, I believe this one is more accurate:

EVP / (ATA * 10) = Miss Rate

Here's the link to my topic about proving the formula.
http://boards2.sega.com/psu_board/viewtopic.php?t=62693

In my first video, your formula would have produced a miss rate of 26.31%. Compared to the actual miss rate of 6.401%. Which is a ~20% difference.

In my second video, your formula would have produced a miss rate of 18.92%. Compared to the actual miss rate of 7.826%. Which is a ~11% difference.

In my third video, your formula would have produced a miss rate of 33.84%. Compared to teh actual miss rate of 30.6%. Which is a really accurate 3.24% difference.

In my fourth video, your formula would have produced a miss rate of 25.41%. Compared to the actual miss rate of 18.18%. Which is a 7.23% difference.

Video 1: 20% difference compared to the other formula's 2.618% difference.

Video 2: 11% difference compared to the other formula's 3.846% difference.

Video 3: 3.24% difference compared to the other formula's 5.03% difference.

Video 4: 7.23% difference compared to the other formula's 1.14% difference.


I think your formula was very good. Just not overall more accurate than that one. : P

[Pillan]

Yes. I know. I tested that one a while ago and I've actually been using that in all the off-hand calculations I made for the past while.