Choosing the Right Monster Power Level For Farming

[donko]

OK first thing first thanks for the effort on the guide. All the math looks solid to me. However the omission of the extra loot drop is what taints this guide from actually giving useful information regarding MP level choice.

Let's say that white mobs on MP0 drop gear (not gold or potions or gems) 10% of the time. At MP10, that goes from 10% to 110% (yes, they can double-drop), which is a 1100% relative increase... and in exchange you get a 3439% increase in monster health. After factoring in the increased magic find as well, you should end up getting about 1678% the MP0 amount of loot from whites. That's a huge increase on what, in my experience, is a hugely inefficient source for MFing on MP0; however, there's nothing like a x16 multiplier to transform inefficient into amazing. I might begin recording how many rares I get from white mobs as opposed to elites over a series of Alkaizer runs on MP0, then (assuming number of rares dropped is proportional to number of legendaries dropped) factor the increased drop chance into my calculations. Out of all the things that bothers me about my math, this is top of the list. However, I have trouble believing that whites would ever farm better than elites, even on MP10 or some such craziness; I'd just be doing it for thoroughness. Again, if/when I have time, no promises.

You hint at it here but the scope of how much the bonus loot modifier plays a role in higher MP farming is played down. So let's just briefly look at the bonus loot system.

Based off this blue response http://blues.incgamers.com/Posts/10/...#postId_504785 bonus items can be rolled off of all drops not just item drops. This drastically changes the efficiency of farming higher mp levels.

All the original math in this guide is based off the premise that the amount of loot is the same just the quality (ie unique drops) increase as we increase our mp difficulty. This is not the case. Furthermore, it becomes difficult to quantify the change bonus loot will have if it is proc'd off of non-loot drops as we cannot simply add a modifier such as 100% more loot at MP10 then MP0 without more data on how the bonus loot really works.

I've been doing some of my own testing by using the "Loot tracker" program which was originally posted on ownedcore. It times your runs and can present a total item drop count at the end as well. Testing was a small sample so obv a grain of salt should be taken with the numbers. I did about 5 identical runs of each MP between 2 and 7 of the alkazier run. My barb does about 120k dps on paper doll, and about 600-700k when doing a ghom effective dps test. What it showed was:

MP: Drops Per Hour: Est Legendaries Per hour (assuming drop rate of 1 in 1500) this is adjusted for MF and my Plvl of 65

MP7: 426 : 1.597
MP6: 442 : 1.585
MP5: 497 : 1.702
MP4: 511 : 1.665
MP3: 592 : 1.830
MP2: 546 : 1.505

Drops per run increase vs MP2

MP7: 98.9%
MP6: 78.6%
MP5: 74.4%
MP4: 40.0%
MP3: 28.6%
MP2: 0%


Though the sample size is small it seems to show 2 things:

1. Bonus loot increases in frequency at a much higher rate then the monster power charts here http://us.battle.net/d3/en/blog/7540457/ would lead a person to believe. MP2-MP3 would at first glance only give 9% more drops however because the proc can occur off of gold and possibly other non loot drops the net increase in items is much higher. As gold seems to be the most likely drop from a white mob as we approach MP10 the positive effect of bonus loot should increase at an even greater rate. In addition because the bonus role is applied once per kill (ie champ packs only drop 1 extra item on MP10 not 7-9) the value of killing champ packs decreases greatly.
2. The change in efficiency between monster powers is not nearly as steep a slope as presented in the guide on the first page. This is not really a fault of the math but the fact that bonus loot was not included in the calculations. The magic find bonus given by increasing the levels is pretty crap and has diminishing returns as you increase in Plvl. The loot bonus is a much more powerful reward and needs to be considered when choosing a farming level.

[galzohar]

1 MP level definitely doesn't turn you from losing thrive to keeping it. You just lose it very slightly less. Going 4 levels higher you start to see a more significant uptime, but still nowhere near 100%. Basically, if you have the DPS to kill fast enough to even think about that MP level, you also have the DPS to kill fast enough to lose thrive often. At least judging by my experience and calculations.

Stone of jordan is an amazing item regardless, and especially amazing for anyone that just runs through white monsters to kill them and only stop on elites (sprint barbs, ToC DHs). By the way, all experiments I mentioned here were done with a stone of jordan equipped. I've been using one since a bit before 1.0.5 was released and wished I would have bought it earlier.

Counting time for elites, by the way, was done excluding the looting time and filtering of what items to pick up etc. Only pure combat time was counted, with the rest being counted as "time between elites".

Accepting that thrive won't be up all the time (even to the point of just replacing berserker with furious charge and boon with unforgiving) still gave very fast farming comparing to what anyone else can achieve, with only possible competition being a DH that can vault non-stop and only stop to 1-shot elites with cluster bombs. This of course has a minimum gear requirement to both 1-shot the elites and not die to the normal monsters on the way while using a legacy natalya set. In any case, my point is run like the wind style is very powerful even without WotB. One you care about XP and not just MF, I don't think a DH can compete at any gear level (assuming barb gear level is the same), but for pure MF the DH might be even better.

Normal monsters are a tough beast to crack when you want to take them into account. In short, though, most of your drops still do come from elites at all MP levels, and the only way you would even think about high MP levels is if you're skipping the normal monsters anyway (it's almost impossible to be efficient at high MP if you kill normal monsters too when you look at the increase to clear time). Also, keep in mind that whenever you get more than 2 items from an elite (or more than 3 when you have 5 valor buffs), that extra item probably counted as a "normal monster drop" that is affected by the chance of extra items. I'm also still not sure how you assume the chance of extra item works - I assumed that 100% means at MP10 you get 2x the "normal" drops, and still haven't seen why one would assume otherwise (not that I say which assumption is correct, just that I haven't seen a real reason to know for sure that it is one way or the other).

[Dethklok]

MP: Drops Per Hour: Est Legendaries Per hour (assuming drop rate of 1 in 1500) this is adjusted for MF and my Plvl of 65

MP7: 426 : 1.597
MP6: 442 : 1.585
MP5: 497 : 1.702
MP4: 511 : 1.665
MP3: 592 : 1.830
MP2: 546 : 1.505

Drops per run increase vs MP2

MP7: 98.9%
MP6: 78.6%
MP5: 74.4%
MP4: 40.0%
MP3: 28.6%
MP2: 0%

Let's delve into this further. I'm hoping that each run, regardless of MP, was as close to identical as possible, with about the same numbers of whites and elites. A larger sample size would also be nice. Also, is "drops per hour" any type of equipment drop, or only MF-enabled drops (such as magic or rare)? With more data like this I could properly quantify the bonus drop rate's effect on overall farming efficiency, assuming the data is accurate.

[donko]

Runs were always the identical route as I said before
Drops are any gear item from gray to unique. This is because the game first rolls item drop class ie gear vs gold or potion, if the roll results in an item it then rolls to see if it's a unique the estimate is this roll is about 1 in 1000-1500 before mf as of 1.0.5. So any gear item drop has the chance of being a legendary therefore these are the drops we care about.

some factors that could change the drops per hour stat

1. variance in mob density etc etc which is unavoidable
2. wrath uptime increases greatly with higher mp as discussed earlier
3. my runs did end with the keywarden so there is some variation
4. Possiblity that some mobs may or may not die in wake at lower mp.

However despite the small sample size and extra factors I think it's pretty clear that to attempt to quantify a proper MP selection while ignoring extra loot rolls is going to result in a less then optimal result. Part of the problem is that the bonus roll is still not completely understood and blizzards response doesn't entirely jive with the way ingame mechanics seem to work.

Based on the quote:

If you kill a monster while playing on MP10, and that monster makes its initial roll check to drop an item, it should have a 100% chance to drop a bonus item. So, yes, in that sense, the monster should drop at least two items. BUT, since the initial item a monster drops might just be a pile of gold, it's possible (with auto-pickup) that you may have just missed it. If you feel that something isn't working correctly, though, then please be sure to post in the Bug Report forum so our Quality Assurance team can investigate.

It would appear that anytime you roll a drop on a white mob (roughly 30% of the time) you would get a roll for a bonus item and at MP10 that would mean a bonus item would always drop. If you play at MP10 this doesnt appear to be happening as quite often you will see single item drop and sometimes gold or potion drops without a bonus item. Because of this you may want to either run your own tests or maybe talk to others to try and determine a more accurate model of how the bonus item works before adding it to your guide.

Just as a side note the data is from a older gear set my current barb http://us.battle.net/d3/en/profile/d...5/hero/2095275
appears to currently most efficient on MP4 to MP5 based on drops per time invested. Based on your initial guide the dps requirement to actually be effective when going between mp3 (my old gear set sweet spot) and mp5 (my current one) would be exponential. However when conducting timed runs my 20% dps increase has made MP5 a more efficient choice.

[Dethklok]

Is there any way you could get me large-sample-size numbers on rare (or better) drops only?

If not I'll probably have to do it myself.

edit:

Accepting that thrive won't be up all the time (even to the point of just replacing berserker with furious charge and boon with unforgiving) still gave very fast farming comparing to what anyone else can achieve, with only possible competition being a DH that can vault non-stop and only stop to 1-shot elites with cluster bombs. This of course has a minimum gear requirement to both 1-shot the elites and not die to the normal monsters on the way while using a legacy natalya set.

You might want to check out this thread... the short version is my Plvl 43/44 DH can easily do a full Alkaizer run, on MP0, at 300GF/300MF, in 13 minutes, without help from set bonuses. Oh, and that's when I die once to a Fallen Lunatic.

If this thread has a point, it's that the single most important thing for a farming build is speed.

• All: You need at least one legendary (non-boots) with movespeed. You can go with Tyrael's Might, Inna's Temperance, Lacuni Prowlers... I think there are some other choices as well, but you need to plan on which slot you're giving up.
• Barbarians: Sprint in some form is a must, whether it's the traditional Run Like the Wind build or a Seismic Slam barb (hey, it's Multishot DH with less vault!)
• Monks: Tempest Rush - Tailwind is mandatory, which makes spirit regen gear (SoJ, Xephirian Amulet) very good and 2-handers the way to go. Skills with high damage and Spirit costs (such as Wave of Light) synergize nicely.
• Demon Hunters: Although Hungering Arrow is great for single-target DPS vs ubers, no serious farmer has time for it doing MP0 runs; it's all about Bat Companion, hatred regen, Vengenace, and spamming something like Multishot or Cluster Arrow (with the right gear and passives that beast of a skill can cost a mere 20 hatred; any -5 Cluster gear is very good). Vault in some form with Tactical Advantage is mandatory. Max Discipline everywhere you can get it (SoJ and Xeph are both great for this build); Legacy Nat's is very nice for this but you can do just fine without it.
• Witch Doctors: You need to activate Spirit Walk as often as possible for the movespeed bonus. Only way to do this is Grave Injustice and lots of pickup radius (which is also very handy for keeping mana and health up as you spam high-mana spells); well, and Spirit Vessel, obviously. Thing of the Deep is your friend.
• Wizards: This class probably has it worst off. To get movement speed they want to go all Critical Mass melee so they can reduce the cooldown on Teleport; however, Wicked Wind is more of a single-target affair, Meteor's delay is a deal-killer, and procing is just hard in general. I think Archon with the Teleport rune (you don't need Improved Archon for seal-clubbing), Evocation, and maybe Critical Mass are the winners here, but the path isn't clear and I could be wrong.

[Dethklok]

Mods, delete this post.

[magicrectangle]

• Wizards: This class probably has it worst off. To get movement speed they want to go all Critical Mass melee so they can reduce the cooldown on Teleport; however, Wicked Wind is more of a single-target affair, Meteor's delay is a deal-killer, and procing is just hard in general. I think Archon with the Teleport rune (you don't need Improved Archon for seal-clubbing), Evocation, and maybe Critical Mass are the winners here, but the path isn't clear and I could be wrong.

As you deduce, critical mass is not good for low MP farming. Archon isn't particularly good at alkaizer runs, either. you need a modified run that covers only zones with high monster density, preferably with a minimum of backtracking. My archon run is this:
Keep Depths 2 -> Skycrown -> Stonefront -> rakkis (backward from waypoint)

Depths 2 is the only one of those that isn't linear, but its monster density is so high it is worth it anyway. Any zone with sparser monster distributions is going to favor one of the classes with better travel skills, but covering these zones can be quite efficient as an archon.

[galzohar]

Why skip fields? It might not be a straight path, but you can go through it in various way without any backtracking. Also, shouldn't you get 5 stacks from core of arreat->tower->arreat level 2 first? Those are also linear areas (and if you reach a dead end in arreat level 2 you can just tp out rather than backtrack, and with some experience you can make sure you reach the dead end that gives you the most monsters along its path).

[magicrectangle]

I do fields when I come out the back end of rakkis.

I stopped doing core because I had issues with archon running out but not being cooled down yet when I transitioned out of core. It only happened maybe 1/5 of the time, but that's enough that I didn't want it in my run.

Tower is pretty wimpy in terms of number of guys, but I should probably experiment with putting it back into my run.

[Rollei]

Sorry about the delayed response. I rarely check the main page and was unaware I'd been featured. (Seriously though, thanks Flux!)
What we have here is square peg in round hole syndrome. Assuming 300 MF from gear/paragon and 5-stack Valor, going from MP1 to MP2 is an effective increase of... 5% magic find (25/500). So really it's about doing the same thing but now faster because of Thrive. If one MP level keeps Thrive up, I can actually see this working; you get up to 16% more effective move speed (1.44/1.24, assuming a legendary for 24% move and saving Sprint for battles), and although Thrive probably won't quite give 40-50% DPS, it should provide enough where, combined with the Thrive move bonus, you accomplish the same run faster or just as fast. However, going up 2 or more MP levels for Thrive is just... rediculous.

If I had some kind of brain aneurysm and decided to build a Thrive barb for MF runs (and didn't predict a 1.06 revision to the MP system), I'd try to solve the problem somewhat by leaning heavily on the "damage vs elites" mod. Stone of Jordan? You betcha. Gearing to get 50% or so damage vs elites would probably heavily weaken your damage against white mobs... but that is, in its entirety, your justification for a higher MP level. Your complaint, literally, is "but I kill them too fast!"

Do you understand how counter-intuitive that is?

I'm not mocking the power of the traditional Run Like the Wind playstyle when it comes to keyfarming; it's awesome for that. But all your frustration comes from sticking to the "infinite WotB" model despite a page of math that says it isn't going to work. My serious suggestion: accept that Thrive won't be up at all times.

I'm not suggesting dropping Thrive. Actually, for a MP0/1 MF barb, some form of Sprint and WotB are pretty much mandatory for the movespeed bonuses alone, and you really only get 4 skill choices (other than rune selection). Furthermore, Thrive is a decent rune because it lasts longer than the others (even if only by a couple of seconds). I'm just saying you need to get over the fact that it will be down sometimes when farming MP0/1, and that some other runes might begin to look more attractive (Insanity?).

Again ur math is flawed. Take my case, and i'm backing up my argument with acctual practice.

My avarage alkaiser run on MPL1 lasts for about 12min. My avarage MPL5 alkaiser run lasts for around 15-16

Going from MPL1 to MPL5 on a 300% MF char (gear + NV + paragon) is equivalent to increasing MF about 30%, not to mention that bonus item go from 1% to 36%.

Lets get the worse case were i do a run in 16min. Thats a 33% increase in time. So on MF/Time im pretty much even, now u can count the bonus item and i'm gaining more items per minute on MPL5.

WOTB has a LOT of advantages that GREATLY decreases the time u spent on a run. Move speed, DPS, imunity to knockdowns, frozen, prision. Not to mention that the increase in the health of monsters makes a lot easier to never go low on fury.

Thats on the worse case where i do my run in 16min. If u count 15min the increase in time equals only 25%. I've chronometrated a lot of runs, most of the time i'm around 15min at MPL5 and 12min at MPL1