Guys. Why do you have to flame the guy the made an effort? Indicating a mistake is fare. The others that ahve other ideas, how about making a new thread and so some actual research instead of bullshitting with napkin maths?

Guys. Why do you have to flame the guy the made an effort? Indicating a mistake is fare. The others that ahve other ideas, how about making a new thread and so some actual research instead of bullshitting with napkin maths?

Because it's misleading and confusing information on the frontpage on one of the biggest D3 sites. And he left out the derivation of his drop rate function. Seems to just have come from trying random functions involving %HP until one matches his limited data reasonably well by sheer coincidence.

How to fix: Completely remove the "Galice Extrapolation" stuff from the frontpage as it is completely wrong. Summarize the findings in a simple way:

* Rifts have 25% higher drop rate which is equivalent to increasing difficulty by { ln(1.25)/ln(1.15) ~= } 1.6 torment levels.
* %MF is worth ln(1+%MF/100)/ln(1.15) torment levels. Maybe have this be a separate graph of %MF against effective torment increase.
* Have the y-axes start from 0 to give a more honest visual representation of the relative drop rates.

The data still isn't very useful unless you know your clear speed, as it is legendaries/time and not legendaries/mob that is important. This is where collecting statistics on clear times at different DPS levels and torments comes in as I posted previously.

It would also be possible to make a simple model if you know the ratio between killing and running at a certain difficulty. The time spent running (majority on T1) should remain constant between difficulties while the kill time should increase by roughly 60% per T ideally, in reality more due to needing to dodge and kite. Deaths are also a factor.

Keeping DPS and movespeed (including Teleport etc) constant and assuming no deaths, no dodging:
t_n: Clear time on torment n
t_1: Clear time on torment 1 (which is a good difficulty to collect data on)
f_r: Fraction of time spent running (and clicking chests etc)

t_n = t_1*( f_r + (1-f_r)*1.6^(n-1) )

Not a complete model by any stretch of the imagination, for example you would need to add a constant term for the downtime between rifts. Overkill damage against swarm mobs is also tricky to compensate for. But it's an example of what you could do to get numbers that are far more useful for determining what T to run.

I think you guys are some silly assholes - sorry to say...
I´ve never seen anything this childish or spoiled behaviour in Diablofans forum.....

This guy might not have been taking everything into account, as some of you mentioned, but instead of flaming and acting like some kids - who can only point out mistakes and tell what everybody else makes wrong....

You could tell it in a decent langauge and be gratefull for all the work he made here.......

I´m really a shame about you guys comment above.....

Have you given him any thoughts about how he´ll react??? He made a friendlt job here to support everybody else in their gaming....
And you conclude that this isn´t worth anything....

I actually feel sorry for those who can´t see the beuty in this work.

I´ll like to thank this fella, for making a lovely post and reading material.

I actually feel sorry for those who can´t see the beuty in this work.

Yeah I'm sorry that I can't, because he didn't post his work. All he did was plop down an (unreasonable) equation and go "take my word for it, mkay?".

Some people assume it is just a 1.15^N function (where N = the torment difficulty), but that has no grounds or proof behind it.

When in fact it was directly stated and confirmed by blues. It is even on the front page right now if you scroll down.

To put it simply, I looked at the monster increase in health and damage for the varying torment difficulty from 1 to 2, and tried to find an equation that yielded the legendary drop increase. The derivation will be omitted, but the end result is the following expression (where N again is a torment difficulty value):

"I randomly tried different equations until I found one that seemed to fit my limited data. Btw, I am not going to go into any details on this critical step, including how much data was used, what difficulties it was from or how it was obtained."

I'm sure that the OP will welcome any criticism to his model and revise it according to where he feels that the criticism is warranted. He should be used to that as a PhD student. I would like to see his follow-up to the critical response. And we will never get any answers if we are required to blindly applaud attempts based on the amount of effort required in producing them. The correct way to honor a post such as the OPs is constructive criticism.

As for criticism based on the fact that simplifying assumptions have to be made, they are a bit unwarranted. Obviously we can't get a thousand people with the exact same build and gear to farm the same route a thousand times on different torment levels to get the required empirical data. This must be done with a priori logic. The OP equates mob health with time, and assumes that this accounts for the amount of instances to have a chance in a given timeframe (which is the real effect of MF, not the qualitative value of a single mob regardless of how many times you kill him, which is effectively uninteresting). The choice isn't between killing 1 skeleton with 30% more chance or 1 skeleton with 0% more chance, but in whether you want to kill 5 skeletons with 10% chance or 1 with 30% chance in a given minute. And the answer we are looking for is where exactly do the lines go for a hypothetical average character in the world of d3. Travel and loading time must be left as ceteris paribus in this respect. But arriving at general advice shouldn't be impossible. Perhaps it can be looked at relative to damage scaling from items; what does it take to one-shot a mob on normal, and how much more damage can be achieved relative to the percent increase in HP from torment levels relative to magic find? I don't know.

The OPs post can inform this question, but not if the values are wrong and certainly not if they are randomly chosen. We need more.

I think you guys are some silly assholes - sorry to say...
I´ve never seen anything this childish or spoiled behaviour in Diablofans forum.....

This guy might not have been taking everything into account, as some of you mentioned, but instead of flaming and acting like some kids - who can only point out mistakes and tell what everybody else makes wrong....

You could tell it in a decent langauge and be gratefull for all the work he made here.......

I´m really a shame about you guys comment above.....

Have you given him any thoughts about how he´ll react??? He made a friendlt job here to support everybody else in their gaming....
And you conclude that this isn´t worth anything....

I actually feel sorry for those who can´t see the beuty in this work.

I´ll like to thank this fella, for making a lovely post and reading material.

<3 Blitzher <3

On that note I wish to thank the people criticising OP. I do NOT think you are assholes at all, and I appreciate that you attempt to explain what you find to be wrong with OPs calculations, and as such outing yourselves to be called assholes in the name of stopping "childish or spoiled behaviour". I for one welcome critique, as should we all. The only namecalling I could find in this thread started just now.

Here's my ranking of whats happening in this thread, its very similar to the rest of the internet:

(1) Name-calling

(2) Inability to clearly state issues with someone's post without attacking them or their effort in any way - i.e. just pointing out where someone went wrong doesn't do anything. Improve the model. I saw that at least once in the comments, that guy gets a gold internet star. That internet star, however, does not show on his character sheet's dps, toughness, or healing.

...

(10000000) Guy trying to do you a favor by starting work on a statistical model that has questionable significance in its current state

Good job! But from statistical point of view, all the drops are random and leg drops less than 1%. It might need tens of thousands or maybe even million drops to notice the difference.

Honestly, I feel a bit bad for the OP... He put this much effort into something this wrong...

Try the following:

Use the correct formula instead of one you apparently pulled out of your ass. Your formula gives the same result for every torment level between 2 and 6, a 17 % increase of the base.

Confirm if you used the correct way of calculating MF for legendaries.

Also, to the guy who put this on the frontpage: Be better than that. Take a look at what you are posting before you slap it on the front page.

This might be a dumb question, but doesn't the MF cap out at 300% anyways even with the bonus given? So I show as 405% MF but the game will still only count 300%?

MF only accounts for 10% towards legendaries so if you were wearing 150 as OP suggest in Torment 3 that would only be 15% - I have a hard time believing this would have an impact if any at all.

Would this mean since the new legendary multiplier for rifts is 25% would this be an additional 15%?

Regardless of which formula you use to calculate the legendary change buff, the conclusion does not change. The deviation is small enough that it hardly matters. If the successive difference between torment X and torment X+2 is anything close to 30%, then stacking 300% MF is like going up 2 torment levels. Looking at this post makes me feel like an idiot, as it's something I could have figure out myself.

People are so quick to jump and say the conclusion is wrong without attempting to understand the math and try substituting in the "correct" equation (which we don't absolutely know whether it is correct, to be sure).

The only real question I have is: Is MF multiplicative? ie. Is it 30% more than the current difficulty's drop rate, or 30% more than that of norm. That chance things much more than which formula you use for the MF increase in torment.

I should probably give an example using the 1.15^T formula. Using 1.15 as the baseline of T1 gives: Torment Outside Rift 1 1.15 1.44 2 1.32 1.65 3 1.52 1.90 4 1.74 2.18 5 2.01 2.51 6 2.31 2.89

Let's say you run T4 rift with 300MF, then your drop chance is 2.18*1.3 = 2.84, which is almost as good as running T6 rift. Running T2 rift with 300MF gives 1.65*1.5=2.15, which is no longer better than a T6 outside a rift, but still better than T5.

Use the correct formula instead of one you apparently pulled out of your ass. Your formula gives the same result for every torment level between 2 and 6, a 17 % increase of the base

It's not surprising. All the way from Hard to T6 the health increase each level is 60% (1.6) and he's taking the cubic root which will always be 17% (1.17). For the record the damage increase is 45% each level and gold/xp 40%.

Edit: Someone else already posted this on the first page of the thread so I'm rather redundant!

Let's say you run T4 rift with 300MF, then your drop chance is 2.18*1.3 = 2.84, which is almost as good as running T6 rift. Running T2 rift with 300MF gives 1.65*1.5=2.15, which is no longer better than a T6 outside a rift, but still better than T5.

Does there even exist gear in this game that would let you do T4 at 300% MF?

I always like calculations on game mechanincs and I really enjoyed reading this thread, but I won't use it to define what's the most efficient way to farm.
Those results are based on absolute magic find values but the don't take account of monster types inside and outside rifts.
White mobs have the lowest chance to drop overall in the game (and rifts are flooded by them); unique (purple) monsters are probably the best thing in the game for both quality and quantity of loot. Inside rifts there are no puples or events that spawn them, except the final boss, while in a campaign area like Westmarch Heights or Westmarch Commons there is plenty of them thanks to events and quest uniques; this will greatly increase the chance for loot in normal areas.
I'm pretty sure that some normal areas are still better for efficiency than rifts for legendary drops over a set amount of time.

What happened to my last post? I am pretty sure it wasn't against the TOS or something. Let's try this again.

It's funny that when I followed the link from Reddit to here, the post on Reddit was up voted massively, and by the time I try to find it again it was already down voted to oblivion. Apparently, the guys on reddit did not even attempt to see what the difference between the "suggested" and "given" formulas are, and keep shouting that 1.17 != 1.15. (of course they didn't say that, if they knew that the difference is this small, they would have stopped down voting)

That is incorrect, the difference is a bit bigger, as pointed out by someone below.

Nothing beyond repairing anyway... And it's not like the main page is the 100% definite truth of everything diablo related. Msitakes happen.

How to fix: Completely remove the "Galice Extrapolation" stuff from the frontpage as it is completely wrong. Summarize the findings in a simple way:

* Rifts have 25% higher drop rate which is equivalent to increasing difficulty by { ln(1.25)/ln(1.15) ~= } 1.6 torment levels.

* %MF is worth ln(1+%MF/100)/ln(1.15) torment levels. Maybe have this be a separate graph of %MF against effective torment increase.

* Have the y-axes start from 0 to give a more honest visual representation of the relative drop rates.

The data still isn't very useful unless you know your clear speed, as it is legendaries/time and not legendaries/mob that is important. This is where collecting statistics on clear times at different DPS levels and torments comes in as I posted previously.

It would also be possible to make a simple model if you know the ratio between killing and running at a certain difficulty. The time spent running (majority on T1) should remain constant between difficulties while the kill time should increase by roughly 60% per T ideally, in reality more due to needing to dodge and kite. Deaths are also a factor.

Keeping DPS and movespeed (including Teleport etc) constant and assuming no deaths, no dodging:

t_n: Clear time on torment n

t_1: Clear time on torment 1 (which is a good difficulty to collect data on)

f_r: Fraction of time spent running (and clicking chests etc)

t_n = t_1*( f_r + (1-f_r)*1.6^(n-1) )

Not a complete model by any stretch of the imagination, for example you would need to add a constant term for the downtime between rifts. Overkill damage against swarm mobs is also tricky to compensate for. But it's an example of what you could do to get numbers that are far more useful for determining what T to run.

I´ve never seen anything this childish or spoiled behaviour in Diablofans forum.....

This guy might not have been taking everything into account, as some of you mentioned, but instead of flaming and acting like some kids - who can only point out mistakes and tell what everybody else makes wrong....

You could tell it in a decent langauge and be gratefull for all the work he made here.......

I´m really a shame about you guys comment above.....

Have you given him any thoughts about how he´ll react??? He made a friendlt job here to support everybody else in their gaming....

And you conclude that this isn´t worth anything....

I actually feel sorry for those who can´t see the beuty in this work.

I´ll like to thank this fella, for making a lovely post and reading material.

<3 Blitzher <3

When in fact it was directly stated and confirmed by blues. It is even on the front page

right nowif you scroll down."I randomly tried different equations until I found one that seemed to fit my limited data. Btw, I am not going to go into any details on this critical step, including how much data was used, what difficulties it was from or how it was obtained."

As for criticism based on the fact that simplifying assumptions have to be made, they are a bit unwarranted. Obviously we can't get a thousand people with the exact same build and gear to farm the same route a thousand times on different torment levels to get the required empirical data. This must be done with a priori logic. The OP equates mob health with time, and assumes that this accounts for the amount of instances to have a chance in a given timeframe (which is the real effect of MF, not the qualitative value of a single mob regardless of how many times you kill him, which is effectively uninteresting). The choice isn't between killing 1 skeleton with 30% more chance or 1 skeleton with 0% more chance, but in whether you want to kill 5 skeletons with 10% chance or 1 with 30% chance in a given minute. And the answer we are looking for is where exactly do the lines go for a hypothetical average character in the world of d3. Travel and loading time must be left as ceteris paribus in this respect. But arriving at general advice shouldn't be impossible. Perhaps it can be looked at relative to damage scaling from items; what does it take to one-shot a mob on normal, and how much more damage can be achieved relative to the percent increase in HP from torment levels relative to magic find? I don't know.

The OPs post can inform this question, but not if the values are wrong and certainly not if they are randomly chosen. We need more.

http://www.diabloprogress.com/hero/sodomir-2220/

(1) Name-calling

(2) Inability to clearly state issues with someone's post without attacking them or their effort in any way - i.e. just pointing out where someone went wrong doesn't do anything. Improve the model. I saw that at least once in the comments, that guy gets a gold internet star. That internet star, however, does not show on his character sheet's dps, toughness, or healing.

...

(10000000) Guy trying to do you a favor by starting work on a statistical model that has questionable significance in its current state

Good luck farming!

Try the following:

I hate Galiks. One of the biggest douchebags I've ever had the misfortune of sharing a WoW server with.

Would this mean since the new legendary multiplier for rifts is 25% would this be an additional 15%?

Regardless of which formula you use to calculate the legendary change buff, the conclusion does not change. The deviation is small enough that it hardly matters. If the successive difference between torment X and torment X+2 is anything close to 30%, then stacking 300% MF is like going up 2 torment levels. Looking at this post makes me feel like an idiot, as it's something I could have figure out myself.

People are so quick to jump and say the conclusion is wrong without attempting to understand the math and try substituting in the "correct" equation (which we don't absolutely know whether it is correct, to be sure).

The only real question I have is: Is MF multiplicative? ie. Is it 30% more than the current difficulty's drop rate, or 30% more than that of norm. That chance things much more than which formula you use for the MF increase in torment.

I should probably give an example using the 1.15^T formula. Using 1.15 as the baseline of T1 gives:

Torment Outside Rift

1 1.15 1.44

2 1.32 1.65

3 1.52 1.90

4 1.74 2.18

5 2.01 2.51

6 2.31 2.89

Let's say you run T4 rift with 300MF, then your drop chance is 2.18*1.3 = 2.84, which is almost as good as running T6 rift. Running T2 rift with 300MF gives 1.65*1.5=2.15, which is no longer better than a T6 outside a rift, but still better than T5.

Edit: Someone else already posted this on the first page of the thread so I'm rather redundant!

Does there even exist gear in this game that would let you do T4 at 300% MF?

I always like calculations on game mechanincs and I really enjoyed reading this thread, but I won't use it to define what's the most efficient way to farm.

Those results are based on absolute magic find values but the don't take account of monster types inside and outside rifts.

White mobs have the lowest chance to drop overall in the game (and rifts are flooded by them); unique (purple) monsters are probably the best thing in the game for both quality and quantity of loot. Inside rifts there are no puples or events that spawn them, except the final boss, while in a campaign area like Westmarch Heights or Westmarch Commons there is plenty of them thanks to events and quest uniques; this will greatly increase the chance for loot in normal areas.

I'm pretty sure that some normal areas are still better for efficiency than rifts for legendary drops over a set amount of time.

No idea. I think MF is removed from the list of affixes except for certain legendary and topaz, is it not?

Double post, the forum monster strikes again.

What happened to my last post? I am pretty sure it wasn't against the TOS or something. Let's try this again.

It's funny that when I followed the link from Reddit to here, the post on Reddit was up voted massively, and by the time I try to find it again it was already down voted to oblivion. Apparently, the guys on reddit did not even attempt to see what the difference between the "suggested" and "given" formulas are,

~~and keep shouting that 1.17 != 1.15. (of course they didn't say that, if they knew that the difference is this small, they would have stopped down voting)~~That is incorrect, the difference is a bit bigger, as pointed out by someone below.