Get the Desktop App for Battle.net Now
- All your games in 1 place
- Log in once
- Automatic game updates
So I'm seeing a bunch of new builds/variations for wizards, whether it is no-freeze, meteors, GC, SS, Archon, torrent, SNS, Archon, etc. Let's gather data and compare. If you just want to see the results, here it is:
Here's the general test:
1. Act1: FoM
2. Act2: Desolate Sands
3. Act2: VoTA
4. Act2: Kulle (timed)
5. Act3: FoS
6. Act3: SB (timed)
7. Act3: Core of Arreat (timed)
8. Act3: Azmo (timed)
9. Act4: spires1 (include leah's ghost in cc, kw)
We tried to pick a variety of targets across all acts throughout this game without being too large. If you want more/less tests or want to add a suggestion, please post here. I thought of adding ubers, but it might be a bit of a hassle for people to get the keys. All sections must be fully cleared, mp10, inferno, solo, and completed by wizard. Do the runs normally, meaning use your normal farming build. Some runs are timed to get a general sense of how much edps you're outputting. If you cannot complete, say SiegeBreaker, just put down "NA."
This is a data compilation to see how builds fare against each other in certain well-known areas, showing other wizards what stats one needs to do well for a certain build. Aren't you curious how sleet stormers and arcane torrent runners fare? Or what amount of mit/recovery is important: If we see a bunch of people do well with 92% mit (making these numbers up), kinda with 91% mit, and poor with 90% mit, that's very telling. Maybe it's the opposite, where running a GC build gave maybe 1 death, but overall the edps was much higher. Maybe you were hesitant about SS or Blackice, and saw many deaths in Act3/4, but almost no deaths in Act1 (where you play the most), so it now becomes a viable option. Use the data how you will.
The goal is to introduce a measuring stick or some sense of standard into various wizard builds. If the results seem too good to be true, I'll run with the player and try to confirm their numbers. Try not to be the wizard that puts in 0 deaths, then gets killed twice when challenged.
These tests were initially created by me, then opened to the public. After a bunch of good suggestions and debate, the current test is the end result of what the community thinks should be involved in an MP10 trial.
Post the following:
1. build name
2. link to build
3. briefly: unbuffed dps, mit, health, and any recovery stats
4. d3up profile used (used to get specifics from 1 and 2)
5. death count in each of the sections
6. kill time for Kulle, SB, short Core of Arreat run, Azmo
7. if follower used, name type and any significant items used (e.g. buriza)
* do not click on shrines
* ignore treasure goblins
* act4 spire 1 includes leah's ghost in cc
* full clear or best you can to a full clear (no caves or events need to be done)
1. build name (e.g. SNS)
2. link to build (e.g. http://us.battle.net/d3/en/calculator/wizard#VQXhYT!gWb!YZccYc )
3. brief stats (123,456 unbuffed dps, 89% mit, 12,345 health, 1234 loh, 2.5% ls)
4. d3up link to character used (e.g. http://d3up.com/b/blah)
5. death count
Act1: FoM = 0
Act2: Sands = 0
Act2: VoTA = 1
Act2: Kulle = 1
Act3: FoS = 2
Act3: SB = couldn't kill
Act3: Core = 4
Act3: Azmo = 5
Act4: spires1 = 6
6. kill time (in seconds)
SB: 123 sec
Kulle: 456 sec
Core: 789 sec
Azmo: 123 sec
7. follower used (e.g. Scoundrel with buriza, none)
8. ping (e.g. 123 ms)
Edited by kiza#1883 on 12/1/2013 3:17 PM PST
This is an interesting (ambitious?) project. With so many tests and variations, you're going to need a lot of community participation. I will probably be keeping an eye on this thread and process, as I'm curious how it will all turn out.
I also curious to know and yes, let's collect some data and def. need some help from the wiz community. Hail to King kiza.
does it have to be in plvl 100 stat? and can we have a follower for solo? I feel lonely running by myself.
I will certainly do the test. I always agree with kiza with his reasoning - one build for all the acts. Some people's reasoning is:
I only do Vota run so my glass deeps wiz is good enough, etc etc.
I will work on it at my pace too.
What is FoS again in Act 3? (noob question).
Edited by yodatoy#1838 on 11/7/2013 9:35 AM PST
I'll try to post something after the weekend. I'll try to add more after I finish leveling. It's not a quick, here's the answer kind of thing, but if anything, it'll help the people who make new builds and also help those that want to try them.
You don't have to be P100 to do this test. As long as d3up captures the stats you used during the run, it should be fine. The follower is an excellent question. I forgot about that. A buriza and cold blooded can definitely make a difference. A follower is not only allowed, but recommended. I'll add that to the list. Good catch. FoS is field of slaughter.
Kiza, one thing to keep in mind is that SB and Azmo are probably not likely to capture the effective DPS output of farming builds, like say Meteor or Archon, which will often dip because of a.) AP troubles on singles and b.) not enough enemies to fuel Archon. So I'm unsure if those tests are likely to be as informative. It might be just easier to divide a character's paper DPS by his average farming time across the four acts. That measure would roughly be a gauge how effective a build is, which would indirectly take into account a person's build and gear (beyond basic paper dps). I would actually advocate people timing their farming runs as it could end up being a salient measure. I know that's the measure I'd be paying closest attention to. :)
but that in itself would indicate that Archon and Meteor might not be the best build for everyone. May be we already know that SNS is the best build in all kinds of situation?
The reason I went through so many gears and so many different builds was that I was searching for one single build that can survive in any given situation. Core to Azn has been my benchmark for testing the gear and life sustain with whether it be LS or LOH or EHP or Mit.
It should be included and noted that if Meteor or Archon can not pass SB and AZmo test then that should be the answer.
Actually that's what I'm suggesting. :)one thing to keep in mind is that SB and Azmo are probably not likely to capture the effective DPS output of farming builds, like say Meteor or Archon, which will often dip because of a.) AP troubles on singles and b.) not enough enemies to fuel Archon.
Take the average farming time and normalize that by someone's DPS, and use that as the metric to estimate a person's true DPS output. That shouldn't be just a simple single-target boss fight, but rather take into account a person's strategy for farming, their gear (such as mitigation and sustain, in addition to DPS), and utility skills (like teleport). So the average farming times would seem to me to gauge how well an actual build performs out in-the-field, and we're not biasing the data towards bosses. And it provides an incentive for people not to turtle their runs, which right now they are incented to do (if they want to provide the most favorable report for their favorite build). And I'd fine with one of the challenges being a boss fight, but I imagine more people will be more interested in relative farming times.
For example, Teleport is often considered an invaluable offensive/defensive skill, but in a boss fight for DPS multipliers it's kinda pointless. But if it's improving my bottom-line farming times, then it starts looking like a great idea.
Edited by TekkZero#1963 on 11/7/2013 10:47 AM PST
I have to agree with Tekk about the bosses. Especially since most people haven't touched them in months, with the sole exceptions of Ubers. Unless we see a major increase in the value of boss fights, it's not particularly relevant to test against those when they'd hurt builds that may otherwise out perform ones that can handle boss fights with relative ease.
And I personally think Desolate Sands in Act 2 should also be tested, even is only as a benchmark since it's one of the harder areas for a short range build. While it's generally ignored for a wizard, I'm going to try testing it as well. I should be able to test these tomorrow.
Wouldn't this be way inflated by someone's way of playing? Some players herd half the map and then kill them really efficiently, while others just kill whatever they run into. Some people don't bother with some stray monsters left behind, while others are keen on killing those. Some people take time to pick stuff up, some don't. Some people who pick stuff up manage to do so on the run, while others stand about for half a minute inspecting every drop because they can't identify an item's type by its name.
There are just too much variables you need to take into account before you can make any statement about the effectiveness of a build.
YES to this one more than instead of boss runs...
may be just add core to Azn test and get rid of SB then?
Edited by yodatoy#1838 on 11/7/2013 12:22 PM PST
Absolutely, but that's an intrinsic limitation of these tests from the get-go. Let's use an example of a build that could perform well on these tests: Arcane Torrent builds taking advantage of AD. These builds will output enormous DPS on bosses, so they'll look very powerful. And if you farm carefully, you could end up with minimal deaths comparable to a SNS wizard or Archon. What did we learn? Not that much if the farming times aren't also taken into account. Inevitably, I think farming times are probably the most useful determinant of performance, imperfect as they obviously are.
I'm most inclined to force people to be as efficient in a farming run as possible by timing them, and then normalize that by their DPS. And unavoidably factors like play-style, latency, and random-chance will be confounds in this kind of study. It's practically speaking impossible to carry out a carefully controlled study, unless everyone is honest and then shares gear and builds, and then you average it all out. (so playstyle and latency become less of concerns). So to be honest, this study might be sadly too ambitious, and ultimately unable to yield the kind of information we'd all like to see.
This is what I was fearing as well, but I didn't dare speaking it out loud, killing OP's enthusiasm.
Players interested in showcasing their build/gear in action, can upload videos of their skill sets in action. Kiza can always come to his own conclusion after reviewing the submitted data, and formulate his own opinion. After revealing his conclusion, that can open further debate. That, or a can of worms.
I'm not trying to be a kill-joy, but rather be clear on the study's limitations. Right now, it's heavily geared towards runs where you don't die, while also dealing maximal damage to single bosses. As I'm a RL scientist, I'm painfully aware that often a conducted study is unable to answer the question you REALLY want to ask. In this case, it's the lofty goal of different build's MP10 performance.
I'm still cautiously hopeful, and certainly don't want to curb any participants enthusiasm by what I'm saying. Though I think it's useful to take a moment to critically evaluate the kind of tests people should try. And then brain-storm variables we consider to be salient measures of competitive MP10 builds (as imperfect as they might end up being). :)
SB is a no-go with the Venom build for me. A simple change to Blood Magic and it's no problem. Azmo is tricky solo (having to eat all the fireballs while in the pools) but Blood Magic might solve that too. If I were to run the test gauntlet I would have to make that compromise for all the runs for consistency's sake.
I did that the other night with our group Act 3 clear. Near the end, the other 3 in the group were down and it was just me hiding above/behind him. It came awfully close to a party wipe but I managed to finish him off despite eating a couple fireballs. I was using Venom in that fight but I'm not sure it would have worked over the course of a solo attempt.
Threats of violence. We take these seriously and will alert the proper authorities.
Posts containing personal information about other players. This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.
Harassing or discriminatory language. This will not be tolerated.