Query: Sandstorm accuracy

[Razielix]

Hey guys,

I'll start by saying I haven't really looked too much into the intricate theorycrafting of all of this, having only really landed on the pet battle scene mid-late 2014. I don't pvp (though I really should for achieve and pets lol) so the breed thing I only do to an extent to help me with dailies/power levelling/celestials, and all that..

Anyway.. With the mentioned power levelling of other pets however, Anub Idol (Whom I call Nasus for obvious reasons to some) obviously features heavily on this circuit - And I'm beginning to notice that poor Nasus misses his Crush almost as much as he lands it whilst Sandstorm is up (and with no other accuracy diminishing moves active either)

Is there some kind of invisible mechanic I don't know/understand? Or am I just that lucky? Lol.

(The thing that drove me to query it here just happened to me tonight against Courageous Yon's last pet "Bleat": Sandstorm > Crush missed x5 > Sandstorm > Crush hits x 1 > Crush missed x4. I really wish I'd filmed it, it was hilariously unlucky lol)

[Thalin]

I noticed that I miss a LOT more then my opponent, don't have any numbers, but it's definately noticable.

[Jerebear]

I'm going to go with unlucky. I rarely see misses under sandstorm. With statistics you have to think globally (over the entire wow population). Some people will be unlucky, some really lucky, others in the middle.

That said, there are addons you could probably find where you go fight and record. However, you're gonna need to record 1000's of samples to get something close to accurate.

[Razielix]

Haha yeah I thought so. I didn't really think there was anything I was missing, it says 90% I assumed it would be 90%.

I swear if I didn't have bad luck, I'd have no luck at all

[FuxieDK]

I tripple-miss on regular basis, when Sandstorm is up..
And since it only decrease hit by 10%, chance of tripple-miss is 1:1000...

Since 1:1000 is obviously NOT true, I'd say, maybe somewhere between 5% and 10% where SS is active, I have tripple-miss.. WAAAAAAAAAY too often for tooltip to be acurate or withon scope of RNG..

[Razielix]

Yeah I'm trying to do the figures of the actual mathematical chance of getting 9 out of 10 misses at that 90% level. Astronomical, I bet? (I know that it isn't cumulative though, it calculates it each time.. Still though, lol)

Well either way, I guess I'll just chalk it up to one of those rare occurrences that will (hopefully) never happen again

Maybe we'd have better luck using Demolish while SS is up? Haha.

[Rendigar]

I tripple-miss on regular basis, when Sandstorm is up..
And since it only decrease hit by 10%, chance of tripple-miss is 1:1000...

Since 1:1000 is obviously NOT true, I'd say, maybe somewhere between 5% and 10% where SS is active, I have tripple-miss.. WAAAAAAAAAY too often for tooltip to be acurate or withon scope of RNG..

I think Jerebear is right on with his comment - you would need thousands of recorded battles to even begin to be able to "prove" the miss chance. I am just as unlucky as most unlucky people (I just posted yesterday how many thousands of foods I crafted before Leftovers dropped for me - yet some random person who gave the pet away got it after 7 crafts of a +75 food).

And it is not 1:1000 to miss 3 times in a row with a 1 in 10 chance, but I am rushing out the door so can't dig out that probability formula...

[Maizing]

The miss chance with sandstorm up seems to fit with the reduction in my battles. Sometimes I get lucky and never miss at all, other times, I can miss several times in a row. It averages out to be about right as far as I can tell.

[Jazeel]

I hardly ever miss with sandstorm up but when i do it always seems to be in burst of 3-4 in the same battle, then no misses for ages.

[Kpb321]

Every time I see something like this I always feel like I have to point out the psychological aspect of statistics. People are horrible when it comes to evaluating random things. We look for patterns that don't exist. We have biases in how we remember things. We have expectations of random that are actually the opposite of random (we expect a coin to alternate heads/tails because it's a 50/50 chance).

Yes the odds of missing that many times in a row might be the same as the odds of winning the lottery but guess what. Just like someone eventually ends up winning the lottery if enough people do pet battles in WoW someone will end up being that guy who just misses a ton and you are guaranteed to remember it and remember it better than the 500 times where you got the expected results.

[Opteron]

well, anyone familiar with addon programming could write a code to record hits and misses, I guess.
fishing buddy helps me a lot to track lunkers chances and after thousands of draenor fishing it's very accurate.
if we all have this type of addon installed, we would have a real percentage of misses
my 2 cents!

[FuxieDK]

I tripple-miss on regular basis, when Sandstorm is up..
And since it only decrease hit by 10%, chance of tripple-miss is 1:1000...

Since 1:1000 is obviously NOT true, I'd say, maybe somewhere between 5% and 10% where SS is active, I have tripple-miss.. WAAAAAAAAAY too often for tooltip to be acurate or withon scope of RNG..

I think Jerebear is right on with his comment - you would need thousands of recorded battles to even begin to be able to "prove" the miss chance. I am just as unlucky as most unlucky people (I just posted yesterday how many thousands of foods I crafted before Leftovers dropped for me - yet some random person who gave the pet away got it after 7 crafts of a +75 food).

And it is not 1:1000 to miss 3 times in a row with a 1 in 10 chance, but I am rushing out the door so can't dig out that probability formula...

Yes it is...
1 miss = 10% chance.
2 miss = 10% chance of miss first time, then 10% of those 10% must also miss = 1%
3 miss = 10% of the 1% double-misses will also miss = 0,1% or 1:1000

It's straight forward statistical data.

[Razielix]

It wasn't so much that it was like "omg giuz it never hits wtf hax" lol. I was more checking if there might've been something like, I don't know, the tooltip wasn't accurate or there was something invisible that was more widely known and I just didn't know about. 'Cos I'm still, comparatively to you guys at least, new to it, is all

At the end of the day, if it was indeed how I assumed and it was just a run of bad luck, then cool beans. All is still right with the world.. Lol

[Yamum]

I haven't done PvP battles in over a year, but around that time there was a lot of trouble with hit chance. For example, pets hitting while blinded, Demolish landing way more than 40% of the time under Sandstorm. I'm not sure if this ever got fixed or not but it might go some way to explaining what you're seeing. Opteron has good advice here.

[Vetteranger]

Just one little note: While people discuss RNG, and everyone assumes that results are random because of the use of "random number generators", in normal applications there really is no such thing as a truly random computer generated number. If you program and read the documentation on those functions, they are always called "pseudo random number" functions. The results follow a pattern, even if the pattern is extremely long before it repeats. In some cases, "random numbers" become more random by being used for different elements of a game, so that the pattern isn't so obvious in one particular area because numbers in the pattern have been distributed to various activities the player may engage in.

Generally in computer programs, "random number" generation starts with a "seed" number. With the same seed number the "random" sequence is actually always the sames sequence of numbers, although as I mentioned above that sequence is a very long sequence.

So when you get three misses in a row on a 10% chance, in reality it wasn't a random occurrence at all, you were simply in an area of the sequence where three high (or low) numbers occurred in the slots in sequence your results came from. The probability is somewhat complicated by the fact that you are not using 10 x 10 x 10 to see if a 10% chance occurs three times in a row, because you are not using three consecutive "random numbers". I don't know how WoW specifically handles this, but If random numbers are being generated server-wide as you play your pet battle, many thousands of random numbers may have been dished out between your one attack and the next. So blame another player for stealing your favorable spot in the sequence. LOL

[Vetteranger]

I haven't done PvP battles in over a year, but around that time there was a lot of trouble with hit chance. For example, pets hitting while blinded, Demolish landing way more than 40% of the time under Sandstorm. I'm not sure if this ever got fixed or not but it might go some way to explaining what you're seeing. Opteron has good advice here.

All of those 80% and 90% hit chances were eliminated with the release of WoD, Yamum. Only a few 50% hit chance attacks remain as "hit or miss" attacks. However, some chances (as discussed here with Sandstorm) are reduced below 100% by other skills which reduce the hit chance. But under most circumstances, your previously 80% Lift-Off attack is now a guaranteed hit, but the damage was reduced to compensate for the 100% hit probability.

[Rendigar]

It's straight forward statistical data.

Sorry, you are absolutely right and I was thinking about drop chances... don't rush and math at the same time, you'll get a brain cramp (or fall down 2 flights of stairs, in my case). There is 1:1000 chance of getting a 1 on 3 different ten-sided dice rolled sequentially (or at the same time). The comment above about how random numbers are generated is also true, depending on when/how the seed is generated. Plus if you accept that arguement you get to blame Blizzard and be right, it *IS* all their fault. But I'd rather blame RNG / luck (even if neither actually exist as a force in the universe, there is only the perversity we humans visualize in it - we are the chaos, not the universe).

[Kpb321]

There is already an add on if you want to try to actually track it. The complexity ends up being not in tracking the result but it correctly tracking all the conditions that can affect your chance to hit like hit buffs/debuffs, always hits based on weather/bleeding and other things like that so they don't throw off your data.

http://wow.curseforge.com/addons/pet-accuracy-recorder/

[Ranok]

There's both psychological effect and Blizz effect involved.

Psychological part is that us missing tends to register more strongly than them missing. I was dead sure at one time that I was being screwed over cause my 85% hit chance missed way more than 85% of the time. Ran two series of battles where I recorded the results over 200 attacks. My hits connected 82% first time, 87% second time.

Blizz effect is just when you're fighting trainers. Published hit chances are pretty much useless as the odds are deliberately stacked in favor of the trainer's pets. Their attacks are going to be lot more successful and yours are going to fail lot more often.

[Jerebear]

Psychological part is that us missing tends to register more strongly than them missing. I was dead sure at one time that I was being screwed over cause my 85% hit chance missed way more than 85% of the time. Ran two series of battles where I recorded the results over 200 attacks. My hits connected 82% first time, 87% second time.

This is a good example of how sample size affects the results. Given wow's pet battler population, 200 samples gives a 99% chance that your results are within 4.95% of the actual number (plus or minus). So with the 82% the actual answer is somewhere between 77.05% and 86.95% (with 99% confidence level)

Doing 1000 samples will give you a 99% chance of being within 2.21% of the actual answer. 5000 samples => within 0.99%. 10000 samples => within 0.7%. And so on