do bonus items drop off of destructables/chests etc and do you think magic find sneaks in and affects the bonus item roll?
wtf is with rng i have gotten hand crossbow set nat's in exact same spot in keep dephs 2 map 4 times now in that wood floor area with 4 metal grates by the 2 further north metalgrates perhaps its due to that skelie spawn dieing in exact same grid point and that settles what legendary it willbe? could test this by kiling stuck mobs like the big wurms that crawl up walls and spit out demons
No, Bonus Items only drop from mobs. MF can affect what is dropped since it is always a gear drop, but it won't increase the chance of a bonus item to drop.
There's some interesting discussions floating around about the bonus item from higher MP and how much benefit it gives, since it supposedly only drops gear as the bonus item. The proc, however, depends on the mob dropping something. I'm wondering, has anyone done any study on slots or individual drop chances for any normal (white) mobs? I realize the gear drop chance likely varies from mob to mob but data for even just 1 mob could help.
Here's some math with the approximation that when a mob drops something, it has a 1/3 chance each to drop either gear, gold, or pot/tome. The post is made by Shandlar on the official forums. The interesting conclusion is that MP8 becomes efficient if you only run MP0 3x as fast or less. So if MP8 takes you half the time, you're better off with MP8.
Bonus item chance doesn't only trigger if the mob was going to drop an item. If the mob drops a pile of gold, or a potion, or a tome, you still get a chance of a bonus item.
So if a mob has a 30% chance to drop 'something' then a 33% chance of item, 33% chance of gold, 33% chance of potion/tome....
10% of kills will drop an item in MP0.
21.3% of kills will drop an extra item in MP8.
So in essence you are more than doubling your item drops from white mobs.
My runs are pretty much spot on 1200 white mobs and 50 elites (roughly 25 packs)
So 25 packs = 5 * 3.5 + 20 * 4.5 = 108 items from elites
1200 * 0.10 = 120 items from whites
228 items per MP0 run.
MP8 would be...
Same 108 items from elites
Same 120 items from whites normally
1200 * 0.71 * 0.30 = 256 bonus items.
484 items instead of 228 item drops per run, or an increase of 2.12x as many items. Then a 200% increase to magic find is applied to all drops.
So 375 vs 575 is 42% effective improved chance of legendary drop. 2.12 * 1.42 = 3.012
That means MP8 vs MP0 you would get more legendaries an hour as long as you can do MP8 within 3x as long.
I'm using the numbers we've recieved by blizzard to do that calculation, however they have admitted that not all mobs can drop bonus items, and some have less than 30%, while others have higher, that is only the average chance to drop 'something', so I feel its not reasonable to just take it at face value.
First of all, I'd like to thank you for such a thorough research! Being an MF-addict myself, I am glad to know that there is such a research team working on this topic.
Didn' have the opportunity to read the whole work yet, but want to point out two things for now:
1. Follower can upgrade you MF higher that 300. for example, I have 300 ( MF shrine doesn't help to boost this number, unfortunatelly) + 75 from NV + 28 from my templar ( 2 rings with 18 mf, shield 20 mf, amulet 40 mf amd sunkeeper 43 mf), so with 5 stacks of NV And my templar I have 403 MF total.
2. Having a very high amount of MF and playingmany hours per day, I can state that getting a rare drop from elite on 5 NV is not a must. I had many sad situations when killing an elite gave me nothing but magic items.
Hope this information will be somehow usefull for you.
(thundersteele beat me to the post but since I already typed it out, I'm going to post it anyway)
1) Is confirmed to not be intended, though whether the extra MF displayed is effective or just a tooltip bug is still being debated. Either way it is being fixed in 1.0.5 and the follower MF contributions should obey the cap.
2) If you have 5 NV and didn't get a rare or legendary/set, something was wrong. You should ALWAYS get one at 5 stacks from a champion or rare mob. The only exceptions are a couple quest related rares that are going to be changes to uniques in the next patch. If you were instead saying that the guaranteed rare from elites isn't a big deal, that's debatable, but I inferred that you meant there were times you got no guaranteed drop.
Btw. at which MF level do we assume 100% rare drop chance from elite slots 1 and 4? Cyeron´s excel sheet says 89% at max MF (+625%), so 100% isn´t possible?
By my calculations you would need 770 MF to have a 100% rare drop chance, if the base chances for legenday and 6/5/4 affixes are correct. With 770 MF your probability to get a 4 affix drop raises to 100% because (1+770/100)*0.115 = 1.0005.
Nice info. I actually think that's a quite clever mechanic. This way you can play act I/II without being too concerned about the loot in III/IV laughing at you. But I guess we should have seen it coming now that affix rolls depends on mLvl (does anyone know yet if they depend on MPLvl or mLvl only?).
Also, I laughed quite a bit at the PC Gamer article being edited x).
Mob level only, the power level will not affect affix rolls directly and will only affect them indirectly because of the increased MF.
How will it affect the distribution of items? Like, can you find item level 63 stuff in earlier acts if you crank up Monster Power?
WC: You can get item level 63 stuff in all acts of Inferno when Monster Power is on. The moment you go into Inferno with any kind of Monster Power, all monsters become level 63, and they can now all drop item level 63 stuff the same way that Act III and Act IV Inferno do now. However, cranking up Monster Power beyond MP1 does not further increase your chance of getting 63 stuff.
I read that interview and somehow missed that post. I knew they wanted to do something like that but wasn't aware they were prepared to introduce it in 1.0.5, so that's good news. I was really confused the last few days with it seeming like they were forcing us to all farm Act 3 to get the best gear when the last few patches have been about giving people options about what act they want to farm.
I have been reading this and looked at the spreadsheets. Did anyone do any Act 3 farming with 0 MF gear and only using NV stacks? Im just wondering because my Barb can easily do Act 3 if its better to do Act 3 with no MF or should I drop a lot of stats to reach like 250 unbuffed MF and maybe do Act 2 or even go up to 300 MF and do Act 1? Although as a WW/Nado barb Act 1 was actually hard because I was always fury starved because mobs were so spread out and they died too fast for it to build up.
The bottom line is right now Str gear with MF is about 6+ mil a piece in order for it to be similar to my current gear. Not only can I not afford that but the only way to add MF would be to drop so much DPS and resists and vit that I would no longer be able to handle Act 3 as easily as I can now. So I dont think its worth dropping to lower Acts unless I am wrong and lower Acts are better if using MF.
The question of which act and with how much MF to farm has been asked a lot but the answer always remains that you ultimately have to determine that yourself. The best way to see is to just time yourself on some runs and take a tally of how many total rares that drop or how many elites you kill (and for convenience I'd recommend counting goblins as elites). And when I say total rares, I mean the total is best, not just how many ilvl 61+ rares drop, since you can always convert from total to 61+ later. This will give you a more consistant average that you should only have to make a few runs to complete. Repeat for each Act you are interested in farming. Overall I think tallying the number of elites per unit time will make for the easiest comparison since we have a lot of data on rares per elite at varying MF values.
With your tallied averages for each act of interest you can calculate the desired drop rates for each act using data in this thread, be it ilvl 63 rares, 62+, 61+, or some other value like 6 affix lvl 61+. Note that a lot of the data may still be preliminary and because of varience may not be a true representation of the average values, in particular the # of affix data, so use it at your own discretion and understanding that individual results may vary. You should be able to get a pretty solid value per unit time for each act, as long as your averages are pretty consistant for each run in the act, or you just make a lot of runs to average over. Then take into account that max level JC and 6 prop BS patterns only drop in Act 3 while 5 prop and 2nd level JC patterns can drop in Act 2 and decide which Act you want to spend the time in to farm.
I believe someone is working on or already has a spreadsheet with some preliminary calculations where you input some data of your own and it calculates some data of interest. Check the OP or look over the last couple pages since it was pretty recent.
Probably the easiest value to compare across acts at this time is number of 61+ or 62+ items per hour since we have a lot of data in terms of rares vs MF from elites. The affix data is still coming along and it seems like we're close to figuring out how much MF affects # affixes but that data is still being collected so use it at your own risk.
On a side note, to those collecting data with fraps on a slot by slot basis, and those collecting # of affix info, thanks a bunch as that data is becoming very informative. I've been trying to collect some data on timed runs but I've only had time to do maybe 3-4 Act 1 runs and like 2-3 Act 3 runs in the last couple weeks. I'll still post the info when it's complete, if it's still relavent. Now to just hope the upcoming MF changes don't completely destroy our findings.
Regarding legendaries, the blue post from a couple days about that mentions the 4x increase in drops from white mobs specifically says magic and rare items. Seems there will be no changes, or at least no major increase, to the legendary drop rates.
There is one interesting aspect about MF that I would like to understand further. It says in the post, and I para-phrase, that roughly speaking the drop rates are R6A = 1% R5A = 3% R4A = 6% (if we ignore Legendaries). However I think it will be interesting to understand what is the "rate" like for that 1 NV-guaranteed item from Elite Packs. Since it is guaranteed that it will be a rare, my question is, is the rate when MF = 0% something like Case 1: R6A = 1% R5A = 3% R4A = 97% or are the rates proportioned like Case 2: R6A = 10% R5A = 30% R4A = 60%. What if MF is not 0%? How does the rate change? If it distributed like in Case 2 then having high MF might be worth it even though there is a diminishing return on the number of rares. The rationale being that more R6A means higher likelihood that the item will be good, so the average quality of items will be drastic higher.
This is something that we haven't determined yet, as far as I know. The different cases are of much interest because Case 1 leads to a higher % of rares having 6 affix at high MF while Case 2 leads to roughly the same % of rares being 6 affix at all MF values. I actually looked at Case 1 from a probability perspective a page or two back. Basically the results we obtain from higher MF% should give us insight as to which case the guaranteed rare falls into.
"They want to make all items that drop be good, however this is not something they can fully do in 1.0.4. However weapons of ilvl 61 and 62 can now roll damage that extends all the way to top ilvl 63 weapons"
This could mean a change to number of affixes, affix distribution, or any number of other changes.
Perhaps I am reading something in to your comments that you do not intend. This is what I am claiming. Do you disagree with any of the following?
A guaranteed drop has the same chance of being 6-Affix as a non-guaranteed drop
A guaranteed drop has an improved chance of being 6-Affix when a player has higher MF
A guaranteed drop is, on average, better than a non-guaranteed drop
I agree with all those points, and none of my comments should be taken to contradict your points. I basically took your idea and just applied it to rares.
My mentality for this kind of research is that I want to know how many rares I get per elite group and how many of those rares have 6 (or 5) affixes, as well as how the results change based on MF and NV. I really don't care about the rest of the drops. That way I can take a look at my timed data for each Act, take the Rares/hour figure, apply the affix data to determine how many 6/5 affix rares I get per hour, then account for lvl 63, 62, and maybe 61 drops to get an idea of which act will be the most useful to farm. I think a lot of people will be interested in similar analysis because of the choice of high MF act 1 vs med/high MF act 2 vs low MF act 3 farming and deciding which to do.
I feel like you're mixing up the apparently mathematic effect with the practical effect. It is true that the proportion of 6A and 5A rares compared to total number of rares is lower for the guaranteed rare. This, however, is simply a result of the improvement of all non-rares.
This, then, is not true:
"What this means from a farming standpoint is that the ratio of 5 or 6 affix rares to total rares will be lower for the guaranteed rare than the non guaranteed rare, making the guaranteed rare less valuable since we mainly want 5 and 6 affix rares."
If you get 100 pieces of non-guaranteed loot, they might look like this:
All else non-rare
And 100 pieces of guaranteed loot would look like:
All else 4A
This loot is not worse. It's better. It's the same number of 6A and 5A pieces per drop and the "rest" of the loot is the same or better than it would be otherwise. Comparing the proportion of rares that are 6A and 5A is interesting, but it does not mean that the loot is inferior.
To help explain my post, it might help to note that I'm approaching it from a purely rare-drop standpoint. In other words, if the drop isn't a rare, I don't really care about it. Because of that I've been considering MF affect on rare drop rates and was attempting to draw some comparison between the guaranteed rare drop and the random drops from elites. The conclusion is that, if the assumptions made are eventually proven true, only 1% of the guaranteed rares are 6 affix while 10% of the random rares dropped will be 6 affix. This means the guarenteed rare drop will be lower quality on average than a random rare drop. A possible conclusion that can be made is that you then want to get as many random drops as possible compared to the guarenteed drop, which can be done by increasing MF since the guaranteed drop is fixed.
I stand by my statement you quoted above and insist it is correct, since it is relating all rares that drop, not all drops. I do find it interesting that despite that above statement, the affect of the guaranteed rare on drops actually makes MF better because it gets increasingly better with higher MF (because you get higher % of 6 affix drops and less % of 4 affix with increasing MF, so you get more "good drops" and less "bad drops").
Also, I'm not trying to contradict your calculations, since those were the basis for my calculations. Instead I meant to add my own calculations from a different perspective to try and draw some conclusions about the affect of MF on rare quality with your idea of 5 and 6 affix % being fixed for a given drop, guaranteed or not.
1) Guaranteed rare is on average of lower quality then random rares (quality meaning # of ilvls)
2) The guaranteed rare makes a minor contribution (though still not insignificant) to the number of high quality items
3) Higher MF reduces the the guaranteed rare effective contribution to the number of high quality items
4) The % of rares that are of higher quality is noticably greater at higher MF (3% at 0 MF vs 7.8% at 300MF for 6 affixes with 5NV)
5) Based on (4), higher MF noticably increases the quality of rares, but only when using 5NV stacks
In closing, if the assumptions for MF on NV rare and random rare drops is correct, the guaranteed rare actually makes MF even more useful as it noticably increases the drop rate of higher quality rares. Previous results assumed a constant probability distribution for guaranteed and random rares (i.e., 10% of all rares are 6 affix base, regardless of random or guaranteed) which resulted in a very minor change in quality of drops overall at higher MF. Hopefully we'll see a noticable increase in 6 affix rares at high MF levels that prove this analysis to be correct, since it makes MF seem very good and makes 5NV farming seem very appealing.
The mathematics you posted in that post are indeed interesting and I will include this point in the OP if I find them to not include errors. If so, this means we have two cases: either MF affects the NV rare or it doesn't - how can we see if it does? We make a baseline % sample as large as possible, perform some calculations, farms a large sample with large amounts of MF (preferably 300+) and checks which case is the most likely that the data supports.
Getting there will take time though, so for now I will only post the two possible cases. Hopefully my data @353 MF can help in that (when I actually get to farm x) ).
Btw, I hope that you mean (quality meaning # of ilvls affixes) in the first part of your conclusion.
Sorry if my math is a little mangled and hard to follow. Please let me know if you see any errors in it, but I tried to present it in a easy to follow manner from my point of view. Yes I meant for quality = affixes. I thought I mentioned that in my post, but if I didn't I'll add it somewhere for clarity.
Again, keep in mind it's basically a working theory that is yet to be proven or disprooved. Instead, I'm hoping the results of my calculations can help us determine if the data we collect indicates the assumptions are correct or not. So it's basically a preemptive explanation of behavior I hope to see once more results are in. If the number of affixes as a function of MF ends up being the same for several data points with 5 NV, then the assumptions are incorrect. Instead, if we see the number of 6 affix drops going from 4% to 8% or so when changing MF from 75 to 300 or so, then my assumptions would explain that behavior.
I expect affixes vs MF for 0 NV to be approximately the same for all values of MF since increasing MF from 0 to 300 would only change the 6 affix % from 10 to 10.5% or so. That 0.5% difference will be really hard to see without a very large sample size so is effectively the same for both MF values.