Data I collected on different MP Levels - Barb and Monk with DB

[brbbX]

Hello everyone
Some weeks ago I started gathering data for which mp level would be best for my chars to bot on, since I have some pretty good gear and felt like a waste botting on MP0..
All data was gathered on Act 3, Ciggarc 1.05 donator profile, latest trinity and atom.. The hardest part was gathering the data, due to the massive disconnects and stucks these last times.
You may notice my item pickup/hour may be lower than what you'd expect, but I set it up to only pick up legendaries iLvl 62+ and any jewelry.
I separated between my Monk bot and my Barb bot.. here are the stats:

Monk Stats - Paragon lvl 100
HP 64k
DPS 172k unbuffed (198k with blazing wrath)
AR 800+
Armor 4700+
Build: Monk - Game Guide - Diablo III

Barb Stats - Paragon lvl 100 - WW Build - Using Skorn
HP 68k
DPS 205k with battle rage
AR 750+
Armor 7200+
Build: Barbarian - Game Guide - Diablo III

Anyway, here are my results:

I sorted the table by Rares/hour (seems to be the best metric?), and MP3 wins by a small margin with my barb, and MP2 and MP4 look similar on the Monk.
However, what surprised me the most was how close every mp level was to one another.. The difference was very small..
MP0 didn't perform as well as expected.
From now on, I'll probably just keep my barb and my monk on MP3.
What's your opinions on my little study?

Cheers!

 

[JoeC]

yeup, i agree MP2-3 for my guys

 

[Bazingaw]

Is it just me or has he taken something that is extremely hard to quanitfy (the worth of an MP) and then made it harder by using different hours for each run done making it more or less useless data?

 

[brbbX]

Is it just me or has he taken something that is extremely hard to quanitfy (the worth of an MP) and then made it harder by using different hours for each run done making it more or less useless data?

I tried to get 10+ hours for each MP..
And the data is an average, I summed the stats for each mp with different runs I got.
I don't know exactly what you mean

 

[Bottletorch]

Is it just me or has he taken something that is extremely hard to quanitfy (the worth of an MP) and then made it harder by using different hours for each run done making it more or less useless data?

I tried to get 10+ hours for each MP..
And the data is an average, I summed the stats for each mp with different runs I got.
I don't know exactly what you mean

15.6 hours =/= 8 hours =/= 13.8 hours =/= 8.2 hours =/= 10 hours =/= 7.7 hours

 

[brbbX]

15.6 hours =/= 8 hours =/= 13.8 hours =/= 8.2 hours =/= 10 hours =/= 7.7 hours

I just assumed that after 10+ hours, the random part would be well handled, so the more hours the better.
Yes, I agree that 8 hours is not enough, but I don't see what's the big issue with using a little more hours on some mp levels.. It just gets the average more accurate afterall?

 

[Bazingaw]

What I meant is if you are doing statistical analysis you really dont want to have another random factor ontop of the randomness of item drops already. By having different run hours for each MP even if the total for each MP is just an average you still end up with skewed results due to the fact you arent comparing equal lengths and item drops are also highly random and with increase mp you might not see the benefits of the bonus item drop chance etc properly even out

edit: just saw your new post : you are then having a bias towards one MP focusing more on it to even out those stats.

btw I don't mean to make your work seem useless 'cause its great youre doing it just maybe a little misspent effort

 

[brbbX]

Yep, I really wanted it to be more "scientific", but I just couldn't find a way.
Maybe if trinity saved a different log for each run, instead of giving the totals for the time it's on, I could have some better data?

 

[Bottletorch]

Yep, I really wanted it to be more "scientific", but I just couldn't find a way.
Maybe if trinity saved a different log for each run, instead of giving the totals for the time it's on, I could have some better data?

If you really want to go full on experiment on this, you'll need to do multiple revisions on each MP and control the amount of hours invested on each MP -10/+10 minutes margin of error I guess. When you close db and restart it the "stat log file" gets written over and starts fresh. Could always cut and paste the file to some separate folder to review later on.

 

[Bazingaw]

Yeah do what bottle mentioned. I checked and it probably wouldnt be too hard to change trinity so it makes a new file each time buuuuut bottletorch's way is easier for all!

 

[brbbX]

If you really want to go full on experiment on this, you'll need to do multiple revisions on each MP and control the amount of hours invested on each MP -10/+10 minutes margin of error I guess. When you close db and restart it the "stat log file" gets written over and starts fresh. Could always cut and paste the file to some separate folder to review later on.

yes, but the idea would be to get it to work unnatended, so it could save the logs after each run..

 

[Bottletorch]

You'll need Bazingaw's help then.

 

[Shemes]

You could add the date and the hour to the log name, so GilesTrinity always recreates the files.


Search this:

string sLogFileName = ZetaDia.Service.CurrentHero.BattleTagName + " - Stats - " + ZetaDia.Actors.Me.ActorClass.ToString() + ".log";

And change it for this

string sLogFileName = DateTime.Now.ToString("yyyy-MM-d HH") + " - " + ZetaDia.Service.CurrentHero.BattleTagName + " - Stats - " + ZetaDia.Actors.Me.ActorClass.ToString() + ".log";

It should work, i haven't tested it.

 

[Lilly]

Thank you brbbX, this was an awesome study.

 

[VaporTrail]

Thanks for your data.

However , it's quite inconsistent with the other threads.
Such as this one
http://www.thebuddyforum.com/demonbuddy-forum/81172-after-tests-monster-power-sucks.html

Can you briefly explain what make "Item drop/Hour" from higher mp like mp3+ , higher than Mp 0/1?

From my understanding, the faster you clear the act = more elites killed per hour... [For example, in one hour you kill 25 elites on mp0 vs. 18 elites on mp3]
and of course more chances from RNG per elites/bosses.

Is "Bonus items" really rewarding ? Since they are only dropped from the white mobs [The blue post has confirmed this.]

Thanks again!

 

[brbbX]

I'm pretty sure that the higher item drop/hour on MP3+ is due to the bonus items, since the killing speed for my chars is very similar on low mp levels..
However, what gets my attention is that even with higher item drop/hour, there's no increase in the rares/hour, which means that probably the items that increase are whites and blues :/

 

[Elmo]

I'm not sure why people over-complicate this.

If you can kill with MP1-10 as fast as MP0 then its totally worth doing. Currently MP8-10 is questionable, but really you just need to gear a lot better and be prepared to spend 500mil+ if you want good use of higher MP levels.

 

[sfrattini]

What's your opinions on my little study?

I agree with most of people here, if u can kill fast at higher MPs, go for it otherwise stay on MP0.

BUT MY BIG question is: how in the hell can you average 770-880 items dropped/h????
On my poor Wiz paragon100 280kdps I kill everything "on-sight" at mp1 Act3, never went more than 600/h....what am I doing wrong?

 

[brbbX]

What's your opinions on my little study?

I agree with most of people here, if u can kill fast at higher MPs, go for it otherwise stay on MP0.

BUT MY BIG question is: how in the hell can you average 770-880 items dropped/h????
On my poor Wiz paragon100 280kdps I kill everything "on-sight" at mp1 Act3, never went more than 600/h....what am I doing wrong?

Build a Barb and be happy

 

[fckwarden]

but I set it up to only pick up legendaries iLvl 62+ and any jewelry.

hmmm so you do not pickup lvl60 or lvl61 legs?

aren't there a lot of legs that are in those lvl that are really good?