Got home, got manual, got disappointed!

liga · #1 March 16th, 2004, 09:18 AM

Quote:

(1-(1-((2+prec)/(2+prec+1.5*range))))*(1-(prec/100)^2))*(prec/100)^.2

Doesn't it look about like that to you? You have to actually watch the battle replays to get a good feel for the exact equations, not just look at the results...

I'm really impressed! that formula is what you see looking the battle replay ? You are incredible!

more serious ... I'm quite sure you know a lot of formulas about combat (since you have coded most of them in your simulator) ... is a too hard work for you to just prepare a small rtf document/thread called "known formulas" and keep it growing ... something really simple like:

precision: ....

to hit: ...

to damage: ...

and so on ...

it will be great! I really would like to contribute to the community making a good documentationa and a good web site but I don't' know where you (and Zen, and ceremony, and other peoples) find this informations!

Norfleet · #2 March 16th, 2004, 09:21 AM

Quote:

Originally posted by Saber Cherry:
Chance of hitting a given square:

(1-(1-((2+prec)/(2+prec+1.5*range))))*(1-(prec/100)^2))*(prec/100)^.2

Good grief. How many battles did you have to watch just to get enough data points to get that equation? I'm sure you derived that from some sort of regression sampling, because surely the actual, in-code formula cannot be this awful and must surely look a lot prettier. And I'm pretty sure you had to stage all those battles yourself to control the conditions right.

[ March 16, 2004, 07:22: Message edited by: Norfleet ]

Chazar · #3 March 16th, 2004, 09:31 AM

Quote:

Originally posted by Saber Cherry:
Chance of hitting a given square:
(1-(1-((2+prec)/(2+prec+1.5*range))))*(1-(prec/100)^2))*(prec/100)^.2

Are you sure that the second 2 is really a 2 and not a .2? And I guess the weapon's basic range is missing somewhere...

No, honestly, I agree. I just had to realize this after reading too much within this forum! Its just my nature as a mathematician which makes me wonder about the all these probabilities...

Gateway103 · #4 March 16th, 2004, 09:32 AM

Quote:

Originally posted by Saber Cherry:
Chance of hitting a given square:

(1-(1-((2+prec)/(2+prec+1.5*range))))*(1-(prec/100)^2))*(prec/100)^.2

Doesn't it look about like that to you? You have to actually watch the battle replays to get a good feel for the exact equations, not just look at the results...

What a complicated formula... wait, it simplifies to ((2+prec)/(2+prec+1.5*range))*(1-(prec/100)^2))*(prec/100)^.2 , as the two 1's are redundant... wait a minute, there is one extra ) after the squared turn, was that a typo or something else (could explain the redudant 1's)

Btw, what is the highest possible precision in game short of prec100 spells? Btw, if I were to intepret the extra ) as a simple typo and ignore it, then the formula says maximum probabilty is at prec 40~50 (depends on Range), and actually starts to drop after that (prec100 has zilch ^_^) A result of regression fitting perhaps, albeit intriguing.

-Gateway103

[ March 16, 2004, 07:42: Message edited by: Gateway103 ]

Norfleet · #5 March 16th, 2004, 10:32 AM

Quote:

Originally posted by Gateway103:
Btw, what is the highest possible precision in game short of prec100 spells? Btw, if I were to intepret the extra ) as a simple typo and ignore it, then the formula says maximum probabilty is at prec 40~50 (depends on Range), and actually starts to drop after that (prec100 has zilch ^_^) A result of regression fitting perhaps, albeit intriguing.

A precision of 40 to 50 is pretty insanely accurate. I don't think you find precision that good in the game, normally. The fact that it drops after that by the formula given is surely an artifact of regression fitting, which is the only feasible way that Cherry could possibly have derived such a convoluted formula. It's almost certainly not *THE* formula used in the code, but it's likely a good enough approximation for all reasonable values of precision.

PDF · #6 March 16th, 2004, 05:41 PM

The formula looks faulty, here are some tabular results : prec 10 has a nearly flat hit chance regardless of range, prec 50 is worse than 20..

PREC
RANGE 5 10 15 20 50 99
1 45.12% 59.13% 63.57% 66.69% 64.12% 1.97%
5 8.36% 34.18% 30.44% 31.34% 31.05% 14.72%
10 2.96% 31.52% 26.04% 27.01% 26.95% 22.83%
20 1.25% 31.35% 25.18% 26.27% 26.26% 25.18%
50 0.43% 31.48% 24.98% 26.14% 26.14% 25.73%

Saber Cherry · #7 March 16th, 2004, 05:54 PM

Oh, I should mention that I just made that up. Ha, ha!

I didn't really think anyone would take it seriously...

Finding the formula would be tough, since it doesn't even exist in the game code - the precision formula just tells the projectile where to land; it does not give it a probability of hitting a given square.