3 replies to this topic

[Bubblewhip]

When I'm running my Raven-4x I have a hard time deciding between the AC5 vs a UAC5 build. The problem with the UAC5 build is that I get crazy swings in my games.

Sometimes I go through an amazing game and get all 90 shots off fast and furious with little jamming, while in other games I make one shot and then have a jam, unjam one shot and jam again.

It seems to me that the UAC5 15% jam chance seems to be based on true random (yes i know nothing except quantum is true random) where every preceeding shot after the first always has a 15% chance to jam.

This leads to real inconsistencies of RNG where you can get strings of rapid fire with no jam, and 5 jams in a row because of how the statistics work.

Any DOTA player knows that to fix this you use what is called psudo random number distribution which weighs the jam chance according to the history of your shots.

So in practice this would make each proceeding jam chance look more like this..

Given 20% chance to jam because calculations are easier

5 extra shots

Old method:

0.8*0.8*0.8*0.8*0.8 = 32%-ish chance to shoot 5 extra times without jamming

New method
1*0.9*0.8*0.7*0.65=32%-ish chance to shoot 5 extra times without jamming


But the difference is that given your first 3 shots you have a even/higher or even gauranteed chance to make the extra shot without jamming

So old method would have

3 extra shots

0.8*0.8*0.8=51% chance to shoot 3 times without jamming

New method

1*0.9*0.8=72% chance to shoot 3 times without jamming


That percentage is MUCH higher almost 25% increase chance to shoot without jamming. But it also normalizes weird RNG anomolies where people can shoot 10 times without jamming

10 extra shots


0.8*0.8*0.8*0.8*0.8*0.8*0.8*0.8*0.8*0.8 = 10% chance to shoot 10 times without jamming

But that means when 1/10 times you shoot 20 rounds of UAC 5 you will have an amazingly lucky streak of extra shots.

Where with new method
1*0.9*0.8*0.7*0.65*0.6*0.55*0.5*0.45*0.40 = 1% chance to shoot 10 times without jamming


This is way lower and reduces the chance of statistical anomolies.


This isn't a big deal given a large amount of time, everything evens out on their own, but in small short bursts, a round can easily be decided by how many UAC5 extra shots you were able to spit out. As a result some games can be declared to be won out of RNG while some losses can be attributed to RNG as some games you can get up to 500-700 dmg with your lucky streaks of shots, while other games you will barely break 200 because of how many unlucky jamming streaks you got.

By normalizing the random chance to jam, the damage might go from 500-700 and 200 to more like 300-400 every round because of the predictability of jamming.

The downside to this is that you can kinda "metagame" the jam chance. For example lets say you got into an engagement and you fired 5 extra shots without jamming. You know that you are overdue to jam so instead of jamming at the next engagement, you instead fire into the ground until you jam which resets the jam chance. This doesn't work with True random but it's something you can game with psudorandom.

[Bubblewhip]

Apparently statistics in application is as boring as my college classes on the subject. :|

Edited by Bubblewhip, 06 February 2015 - 01:30 PM.

[Xetelian]

I don't know what more to say other than I also have games where I get 90 off without a hitch in my step and some that leave me frowning.

[LordBraxton]

I like this, because it is a compromise between purely praying to RNGeezus, vs a 'overheat' system where you could avoid jams.

I'd like to see the UAC5 as a more reliable burst fire weapon. I certainly think the first shot jams are pretty stupid.