|
Thread: Chances to...True or False | |
|
feluniozbunio
Promising
Supreme Hero
|
posted April 16, 2007 02:18 PM |
|
|
Chances to resist...True or False
Hey i am constantly shocked by the results of resist rolls. I often get games where either i or my opponent gets 90% of resistance rolls. This happens on Battle for honor when either i or my opponent gets Relic armor. Even if you add a skill and boots its 45% chances to resist but has never been even close to 45%. I know that if there are limited number of rolls the outcome may vary very much, this is a probability issue but i still have a feeling that something fishy is going on here. I am curious have you guys noticed anything similar in your games -that something is giving higher resistance than it is said it does.
____________
|
|
Pomo
Famous Hero
The lone peasant
|
posted April 16, 2007 02:35 PM |
|
|
I don't know anything exactly about whether resistance is bugged - maybe you just got a crappy (for the casters anyway ) seed for the random number generator?
Anyway, it's easy to do some tests to see whether resistance seems to deviate from what the expected level is. Maybe I'll get bored with working on my map and do it at some stage
____________
linkTouched by His Noodly Appendage
|
|
Darkeye
Promising
Famous Hero
of the Deep
|
posted April 16, 2007 09:45 PM |
|
|
Yes, i have the same fishy feeling. We could do some statistical test in duel mode with Ossir (MR = 35 or 65 with/besides unicorns) or Marbas (MR = 45).
|
|
Pomo
Famous Hero
The lone peasant
|
posted April 17, 2007 02:02 AM |
|
|
I can probably just make a test map to do it so that we can manipulate all the possible variables - which arties, heroes, skills etc.
____________
linkTouched by His Noodly Appendage
|
|
feluniozbunio
Promising
Supreme Hero
|
posted April 17, 2007 02:11 AM |
|
|
It would be great. there are just 3 things needed : Skill magic resistance, boots of protection, and relic armor , and some variations with those. Its enough to make coulpe of heroes and do some fights.
____________
|
|
Plexus22
Known Hero
|
posted April 17, 2007 04:46 AM |
|
|
I think this is the case with a lot of supposedly "random" stuff in heroes V not just resistance. Look what happens a lot of times with a mediocre luck or morale stat of 1. Lucks and morales start flyin out all over the place. Also take ghosts and spectres for example. If the hit miss probability worked accurately at a random 50% there would be no need to place a 3 miss cap on them because it would rarely ever happen. Bottom line, the random generation mathematics used in this game suck!
|
|
Pomo
Famous Hero
The lone peasant
|
posted April 17, 2007 06:06 AM |
|
|
Well I haven't done any comprehensive tests - but I did do a preliminary one. 60 casts of eldritch arrow with armour of the forgotten hero, boots of magical defence and magic resistance from luck skill.
28 times resisted out of 60 = 46.7%
So, based on the preliminary evidence - I would say that it seems quite likely that all of these things are behaving exactly as they should. Of course this isn't comprehensive - all the casts were done in one game etc. Perhaps I'll get inspired in a bit and do some stronger tests.
____________
linkTouched by His Noodly Appendage
|
|
alcibiades
Honorable
Undefeatable Hero
of Gold Dragons
|
posted April 17, 2007 08:09 AM |
|
|
Don't forget that statistics can be fickle. Sometimes you'll be lucky, or unlucky, and someone who should resist only once will do it twice or thrice, etc. The thing about Ghosts, for instance, I've tried to keep an eye on them sometimes, and it always ends up around 50 % - haven't done the mathmatics or anything, but from my impressions, when I've tried to be aware of them. And the 3-in-a-row condition could also just be put down in order to prevet too much randomness for outcome, because one player in the race case where his Ghosts were missed 5 times in a row (3 % chance!) would gain an unfair advantage.
____________
What will happen now?
|
|
sdfx
Famous Hero
|
posted April 17, 2007 09:15 AM |
|
|
It's very possible that total resistance is not 0.15 + 0.1 + 0.2 = 45% but it is 1.15*1.1*1.2 = 1.518, which gives a total resistance = 51.8%
____________
|
|
Pomo
Famous Hero
The lone peasant
|
posted April 17, 2007 10:38 AM |
|
Edited by Pomo at 10:39, 17 Apr 2007.
|
Your maths is off there, you could end up with greater than 100% chance to resist - which given a multiplicative formula should be impossible. The numbers you want are
prob-resist = .1 + ((1-.1)* .15) + (1-(.1 + ((1-.1)* .15)) * .2)
prob-resist = .1 + .135 + .153
prob-resist = .388
Although I think it's quite possible given the data that I present that this formula might apply - I think I read in the manual (??) that resistance is additive, so I presume they actually looked at the way the game calculates resistances?
Alc, yes I think that probably most of people's perceptions of higher/lower chances of triggering than should exist arise due to humans' ability to find patterns where none exist - people are generally quite poorly equipped psychologically to deal with randomness.
____________
linkTouched by His Noodly Appendage
|
|
ZombieLord
Promising
Famous Hero
that wants your brainz...
|
posted April 17, 2007 10:39 AM |
|
|
Yes, I think it may be multiplicative, especially for Unicorn's Aura or the Dwarven Luck ability. 'Twice' more chance to resist from 50% is NOT 100%, it's 75%!!! Here's why:
- 50% chance to resist PLUS another 50% FROM THE REMAINING 50%, that means +25%. it is indeed impossible to get to 100%
|
|
feluniozbunio
Promising
Supreme Hero
|
posted April 17, 2007 10:46 AM |
|
|
If the results were really around 30-60 percentages its fine. I got really weird results in my games resisting literally all spell throughout a combat while having 45% and this made me think that something is wrong, and it wasn't one game only.
____________
|
|
Pomo
Famous Hero
The lone peasant
|
posted April 17, 2007 10:57 AM |
|
|
There still might be some funny stuff - I only collected data from one game, which I suspect means one seed for the random number generator. If you really wanted a rigorous test then you'd probably have to close and open heroes a bunch of times and start new games repeatedly so you're not stuck with the same seed. Although I can't say at the moment I'm feeling too motivated
____________
linkTouched by His Noodly Appendage
|
|
ZombieLord
Promising
Famous Hero
that wants your brainz...
|
posted April 17, 2007 11:30 AM |
|
|
You don't have to restart Heroes. The random seed is calculated at the start of the exe, yes, BUT each time you use a random number, the number is 'randomized' each time based on the last one. So, unless it is the same seed AND you do the SAME ACTIONS with random stuff, the results will always be different.
|
|
alcibiades
Honorable
Undefeatable Hero
of Gold Dragons
|
posted April 17, 2007 11:56 AM |
|
|
Quote: Yes, I think it may be multiplicative, especially for Unicorn's Aura or the Dwarven Luck ability. 'Twice' more chance to resist from 50% is NOT 100%, it's 75%!!! Here's why:
- 50% chance to resist PLUS another 50% FROM THE REMAINING 50%, that means +25%. it is indeed impossible to get to 100%
I don't know about the Unicorn (don't think it is as you say) but Dwarven Luck is known to work that way: If you fail to resist a spell on the first roll, you get a second roll.
____________
What will happen now?
|
|
dschingi
Famous Hero
the guy with the dragon golem
|
posted April 17, 2007 06:50 PM |
|
|
Unicorns' resistance is additive like the others (except Dwarven Luck as you said).
Since it's randomness it is hard to check if there is a bug with the calculation, except if somebody wants to cast magic arrow 1000 times in a row with Nur or so...
Quote: Alc, yes I think that probably most of people's perceptions of higher/lower chances of triggering than should exist arise due to humans' ability to find patterns where none exist - people are generally quite poorly equipped psychologically to deal with randomness.
That's exactly what I think is going on here. You can play for a month without any thinking about randomness, and then suddenly there is a game where you are extremely lucky (or the enemy is), and you start noticing every time when people get some lucky stikes "wtf how is that possible it happened again!?". Humans remember extreme situations much better than the normal ones. It's the same with "damnit always when come the traffic lights are becoming red and the others' green".
This is a bit exaggerated but sometimes people act that way even if they try not to.
____________
open source for an open mind
|
|
sdfx
Famous Hero
|
posted April 17, 2007 07:18 PM |
|
Edited by sdfx at 19:19, 17 Apr 2007.
|
Yeah, my formula is wrong. I used a dwarven hero with dwarven's luck, 20% armor and 10% boots and his unicorns(assumed total: 1.3*1.3*1.1*1.2 = 2.2308 -> 123.08%) didn't resist everything.
____________
|
|
Yasmiel
Supreme Hero
Former Chessmaster
|
posted April 17, 2007 07:30 PM |
|
|
Quote: Yeah, my formula is wrong. I used a dwarven hero with dwarven's luck, 20% armor and 10% boots and his unicorns(assumed total: 1.3*1.3*1.1*1.2 = 2.2308 -> 123.08%) didn't resist everything.
You should look at magic voulnerability instead of resistance to simplify the math:
Invert the numbers in formula to get:
0.7*0.7*0.9*0.8=0.3528 = 35% magic voulnerability = 65% magic resistance
|
|
sdfx
Famous Hero
|
posted April 17, 2007 07:41 PM |
|
|
Thx, now that makes sense for me.
____________
|
|
dschingi
Famous Hero
the guy with the dragon golem
|
posted April 17, 2007 07:43 PM |
|
|
The values are additive, not multiplicative:
30% + 20% + 10% = 60%
with dwarfen luck it's 1-(1-60%)^2 = 84%
____________
open source for an open mind
|
|
|