"Crits can Miss" vs. "Three Outcomes" - Page 2

crenshaw · 03/13/07, 7:16 PM

posted this in the mage theorythread but thought i would post it here also http://www.nihilum.co.uk/image/1074/

Harem · 03/13/07, 8:01 PM

Originally Posted by rafikki

For what it's worth, I got combustion up to 10 charges last night (increases your chance to crit with a fire spell by 100%), and got a resist on the next fireball.

This is inconclusive, since in a 1 roll system I think misses take priority over crit.

Kalman · 03/14/07, 12:54 AM

Originally Posted by Harem

This is inconclusive, since in a 1 roll system I think misses take priority over crit.

In the autoattack one-roll system, misses have priority.

Daler · 03/14/07, 1:54 PM

Originally Posted by crenshaw

posted this in the mage theorythread but thought i would post it here also http://www.nihilum.co.uk/image/1074/

That sure sounds like a 1 roll /3 results system to me. Or do I need to get another cup of coffee?

Kavan · 03/14/07, 2:15 PM

Originally Posted by crenshaw

posted this in the mage theorythread but thought i would post it here also http://www.nihilum.co.uk/image/1074/

I wouldn't trust a GM, as far as I know they have no insight into inner mechanics of the game.

Harem · 03/14/07, 10:27 PM

Originally Posted by Kalman

In the autoattack one-roll system, misses have priority.

We are saying the same thing. I was just pointing out that his combustion example could occur in either a 1 roll or a 2 roll system.

Hozz · 03/15/07, 8:35 AM

I cant speak to the math or present any hard evidence but if Blizzard coders are interested in efficiency, it seems very likely to me that they are using 'one roll' to determine the results for each attack.

Drauk · 03/15/07, 8:50 AM

Originally Posted by Hozz

I cant speak to the math or present any hard evidence but if Blizzard coders are interested in efficiency, it seems very likely to me that they are using 'one roll' to determine the results for each attack.

That was one of the old arguments for existance of 1-roll system for all combat mechanics. But it has been statistically proven that special (yellow) melee attacks are on 2-roll system.

Apate · 03/15/07, 11:51 AM

Originally Posted by Hozz

I cant speak to the math or present any hard evidence but if Blizzard coders are interested in efficiency, it seems very likely to me that they are using 'one roll' to determine the results for each attack.

Someone referred to a simulation that they ran where they populated a table with values and compared 1-roll, 2-roll, and possibly 3-roll and found that none of them were computationally taxing, even on a massive scale. I don't recall the precise details, but I think it was in the "Is backstab on a 2-roll system?" thread, and the point was that efficiency wouldn't be a barrier.

• Hamlet · 03/15/07, 12:40 PM

These threads always go in circles. Can we have someone actually test this (I'm assuming that this is how the Rogue specials using two-roll was discovered)?

The central question is, if your computed hit% is H and your crit% is C, is your real chance to crit C or HC?

To resolve this as well as possible, you want to make H small and C big. So stack crit gear and unload on a high-level mob. Don't do any weird tests with things like Shatter--you just want a large amount of perfectly consistent data.

Say you have 83% hit and about 25% crit. You're looking to resolve out about a 4% difference in total crit rate. The statistical error in measured crit rate over N trials is sqrt(C/N). N = C/(.04)^2 = 156. So, do a few hundred trials and it should be very clear.

Skav · 03/15/07, 3:58 PM

Already posted several times :

Experiment : lvl 44 warlock vs lvl 53 scorpion

Theoretical chance to crit : 15% from talents, some more from int and base crit chance.
Theoretical chance to hit : 17%.

So, if we have one roll, we should see mostly crits, if we have two rolls we should see much less crits.

Results :
200 searing pain fired
22 hits (11%)
3 crits

Jayde · 03/15/07, 7:51 PM

Originally Posted by Skav

Already posted several times :

Experiment : lvl 44 warlock vs lvl 53 scorpion

Theoretical chance to crit : 15% from talents, some more from int and base crit chance.
Theoretical chance to hit : 17%.

So, if we have one roll, we should see mostly crits, if we have two rolls we should see much less crits.

Results :
200 searing pain fired
22 hits (11%)
3 crits

If resist rates go up against higher level mobs, would it also not be resonable to assume that perhaps at some point crit scales downward as well? After all, it does for melee and specials, does it not?

I really think for any tests to be accurate, you can't rely simply on level difference for the setup. Long parses at even-level would be the best way to prove or disprove it IMO.

• Hamlet · 03/16/07, 3:36 AM

Originally Posted by Jayde

If resist rates go up against higher level mobs, would it also not be resonable to assume that perhaps at some point crit scales downward as well? After all, it does for melee and specials, does it not?

I really think for any tests to be accurate, you can't rely simply on level difference for the setup. Long parses at even-level would be the best way to prove or disprove it IMO.

I agree with you in spirit, but even-level only provides a 4% miss rate. You kill your statistics--it would take thousands of shots to obtain that kind of confidence interval. I'd just like to see a long parse at +3 or +4.

Twid · 03/16/07, 3:54 AM

This would be a perfect job for Anachronos, downranking spells for longevity, no? I'm not sure if anyone has checked if his ?? status is indeed a +3 value, however.

Jayde · 03/16/07, 6:00 AM

Wonder if there is some silly longer-duration world debuff that lowers your chance to hit with spells? Can faintly remember a few tooltips here and there, but nothing coming to mind.