Rolling Perfect Runewords

Hello,

I am wondering if someone can share Insight into how stats roll on an item when making runewords. More specifically, does every value in a range have an equal probability of being selected, or are the values weighted?

If we use HOTO for an example, is the probability that you roll 30 AR the same as rolling 40 AR? Is it expected that, on average, you would roll a 40 HOTO 1/10 times given it has a range of 10 digits?

-Travysmo

Same chance for 30 and 40.
And for HOTO, it is 1/11 and not 1/10 (11 numbers between 30 and 40 inclusive)

mockingbirdreal wrote: 2 years ago

Same chance for 30 and 40.
And for HOTO, it is 1/11 and not 1/10 (11 numbers between 30 and 40 inclusive)

Ah, yes thank you for the correction.

Isn't this an interesting concept then? I am seeing perfect HOTO going for

Ber regularly, but if you were to make 11 you should, on average, receive your perfect HOTO (

Ber) but then have the others to sell back to the community. Even if you sold them at a large discount you would come out way ahead than buying for a perfect HOTO.

This is then true for CTA and other runewords as well.

Travysmo wrote: 2 years ago

Ah, yes thank you for the correction.

Isn't this an interesting concept then? I am seeing perfect HOTO going for

Ber regularly, but if you were to make 11 you should, on average, receive your perfect HOTO (

Ber) but then have the others to sell back to the community. Even if you sold them at a large discount you would come out way ahead than buying for a perfect HOTO.

This is then true for CTA and other runewords as well.

Most of the time, Perfect HOTO is sold for between

Lo and

Sur.
pricecheck.php?item=1217&other=0&sk=m&s ... &perfect=1

Ber is a too good price.
Maybe the buyer is too rich and didn't want to wait.

Yes, it can be benefit selling HOTO like that (I am not sure about CTA).
But I won't do that. There is only 21 chars and 3 shared stashs (3 more from ladder) so you don't have much space to store all rolls .

Thanks for the link, I did not know this website had that functionality!

Yea

Um, does anyone have any reference/proof of these rolls all being equal vs them going via some kind of ‘normal’ distribution?

If someone said the rolls follow a normal distribution i’d Believe it, the most common ‘valuable’ item I find consistently is

Skin of the Vipermagi and the rolls usually are around 27, I’ve only ever found 2 in the low 30s, compared to many in the mid/high 20s and even a couple in the low 20s...The first one I ever found was +20 all res, I was amazed to find it and socketed it with an

Um ... damn I was so naive

I used to make fun of stats in university...mathematicians dealing with numbers between zero and one...now it’s the most important subject I’m glad I took the time to learn

From personal experience, I haven't ever felt that the mod range had anything but a flat distribution. I just recently did a lot of rerolling for various perfects, and nothing jumped out at me as odd or suspicious. That's certainly not an official test, but with over 200+ rerolls I'd expect any odd behavior to make itself evident.

However, I feel that the consensus is that mod ranges are almost always a flat distribution, and if they were not, someone would have discovered that by now either through rigorous testing or looking into the code/files.

I guess you could make a single player character, load them up with a bunch of runeword packs and do 100 or more rerolls.

Meh, if you say you did 200 rolls, that’s a bigger sample size than mine!

So I guess it’s fair to say all rolls are equal

I have found 3 maras in all my playing, 1 perfect and 2 x 28s, so if it were a normal distribution that’d be like 0.00013% chance of happening, but if it’s flat it’d be like 0.004%... I guess a 3rd option is ‘random’ distribution, which is close to flat but could account for some weird oddities like my skin of the viper Magi and maras.

Do the devs have to sign a non-disclosure agreement? There must be someone who works on the game that knows and can tell us