Basically, me and Legen got into a statistical argument. Here's how I played it out.


~Statistical battle~

Engineering nerd vs Psycho

Psycho Wins


quote from Legen: "I laugh at sins who put 9 points to get from 57% claw block to 60%"
(he thinks it's "only" 3%. A crucial 3%, as it is very close to the 50%, middle of the Bell Curve) - a way to view it

Difference between block rates:
[valid interpretation]
10%-20%: 9/10 - 8/10 to miss
30%-40% 7/10 - 6/10 to miss
40%-50% 6/10 to miss - 5/10 to block
50%-60% 5/10 - 6/10 to block
70%-80% 7/10 - 8/10 to block

as you can see, from 50% to 60% block is the vastest improvement (equaled by 40%-50%), a common misunderstanding: a percentage can be interpreted as a fraction as well. Per-cent, cent means 100 in French.

from 50% block to 60% block is an increase of 1/5th in blocking chance. 57%-60% Without any complex formula, just as an estimate, 3% being roughly 1/3 of 10%, we can estimate that it is about 1/3 x 1/5 = 1/15 improvement in blocking,

how many 15's in 100? To convert it to percentages, almost 7. Meaning a percentage increase from 57% to 60%, would literally give you 7% better chance of blocking. But it's only 3%? How can it be? It's because it's a range of percentages, it is very deceiving. It's because it's closest to 50%, which makes it more significant. Effectiveness when covering a range of percentages digresses from either side of the 50% middle. As we can picture it on a Bell Curve.

Lastly, 10%-20% block is indeed twicefold, 100% improvement in blocking chances, but it is well below the 50% mark, thus much less significant than 40%-50% increase in block rate.



Additionally,




No, not the most significantly helpful, it is still very low block %, 1/10 block odds is like 9/10 chance to get hit, 2/10 block odds is like 8/10 chance to get hit, you'll get hit 1/9th less.

clawblock just rekts blocks hammers too

The Bell Curve can be found all over the place on Diablo 2, it's just some people aren't able to see it. Usually there is a regression towards the average, say, 1-62k lightning, you'll get at 31k Avg if you shoot lightning ten thousand times, and it'll diverge from the average in a Bell Curve fashion. You'll get more result within 10k-50k, than the outer edges as an extreme example, because there's more scores within that segment. You can even do another segment within the 10k-50k, like say, 20k-40k, premise taken that there's more scores within the first outer segment, it regresses towards the average.

it's because 60% is not worth it LOL

it is worth it, read post

You got that wrong.


you talk about chance to miss (ie not block) and then about chance to block
it makes no sense

the relevant factor here is the effective life of your char, which is proportionnal to 1/(probability of not blocking) *
So it's a simple 1/x curve, where x is the chance to not block.
if you reach 0% chance to not block, you are immortal, cuz 1/0 = too much
Each percent becomes stronger when you come close to 0% chance to not block, ie 100% chance to block, because that's where there is a vertical asymptote

You're right when you say that this is a recurring patern in d2
it's here with resists, with telekinesis for es, with damage reduce and amp
but it's not a bell curve
it's just a 1/x



* wrong because duels aren't linear, and blocking takes a few frames, and some stuff can't be blocked but let's not care about it right now




E:
Your 7% calculation is right but clumsy. Here's how it's done

with 57% chance to block, you take on average 43% of the incoming damage
with 60% chance to block, you take on average 40% of the incoming damage
Now, to see the effective increase, you just do 40/43=.93 . You take 93% of the dmg, which means that you now take 7% less dmg (this is the same 7% you got, except here it should be more clear)

if we do the same thing at 0%->3%, we get .97, which means that we take 3% less dmg
if we do the same thing at 90%->93%, we get .7 , which means that we take 30% less dmg
see, the closer to 100%block, the better each %block is



This post was edited by lilith0 on Nov 29 2016 09:16pm

@bolded

90/93=0.967

which means 90 is 96.7% of 93. Approximately 3% more. Notice it is just a little bit over 3% because it is far from the 50% average, because of regression towards the means, the Bell Curve.

Further it gets from the 50% Average, more useless it becomes

50/60=0.833

which means 50 is 83.3% of 60. An increase of 16.7% (1 - 0.833).



This post was edited by Wyrmvater on Nov 29 2016 09:45pm

I don't think that math is right, it's a simple %. Also, more importantly than all of this psycho babbling is the fact that Legen is correct. There are numerous better ways to spend the skill points.

yes, it is a simple percentage, but going from a certain percent to another can be either significant or insignificant (difference).

For starters, your essay does you no good since even on a mindless trap you're trash.

Next, virtually every other used skill except maybe venom is worth maxing over weapon block. Your value per skill point is so low that burst of speed becomes a better point sink and ghosts are better off maxing shadow master or a sentry skill.

You're trying to cram enough word vomit in one post to make it look, as usual, like you're not a sump-sucking cretin that never managed to crawl out of the melee bracket. I realize you're a troll but you're also never going to beat any of the people you're pretending to be autistic to fool.