Main Forum
of 2

Raid Team Composition Guide Updated 9/4/16

13 replies
Untagged users
TL;DR: In 4-4-2, revenant is missing buffs, which I believe makes 5-4-1 closer to favored.


A major point of confusion I had with Nike's post (and a sentiment I saw echoed on reddit by Fennec who probably knows more than I do) was the 5th point of the description of the meta:

"Revenant plus Assassins/Berserker Mesmer > Commander Mesmer + Elementalist in term of total raid DPS."

According to Nike's post on reddit, this conclusion was made in the following way:

"Used standard DPS golem #'s published by [qT] as well as our own observations."

I think this overlooks part of the revenant (and mesmer) experience in the 2-player group, which is that these two players are at a disadvantage in terms of receiving many of the raid buffs that justified the usage of 4-4-2 or 5-4-1.

The idea behind 4-4-2 is that these important buffs are applied to high damage elementalists (or other classes) with near certainty. These buffs are so powerful that it is better to have them with high probability than to have more high damage characters relative to offensive supports.

However, these important buffs can easily miss revenants in a seperate sub-group.

I couldn't find numbers for a lot of these things on the wiki, so in the following I assume that Spotter/Empower Allies auras and Spirits/Banners buffs target 5 players including the ranger/warrior that cast the spell. I couldn't really find any good sources on this but the numbers are easy to tweak. I also assumed they prioritized party members, and I assumed a standard dps burn scenario in which players would all be in range of each other, such as Gorseval or burn phases on Keep Construct.

The first two parties both receive all buffs with 100% certainty in this case by virtue of being in range of each other and party prioritization. For the sake of this post, we can assume that spirits/banners will always be up as all existing bosses exhibit periods of invulnerability that would complicate analysis much further if this was not assumed. I also assume a random arrangement of party members, that is, that there is no coordinating of position in order to prioritize buffs.

The revenant can only receive spill-over buffs from the two main groups, and only one raid member outside of each group can receive buffs from that group.

Consequently, of the 6 raid members eligible to receive spill-over on spotter/spirit/empower allies/banner, the revenant fails to receive these buffs from each party with probability 5/6 (~83%), and therefore fails to receive buffs from both parties with probability 25/36 (~69%). This is a fairly significant loss in effective power and could be worsened further depending on Grace of the Land prioritization, which is unfortunately not a mechanic I believe to be well understood.

We can probably assume independence between spotter/spirits/empower allies/banners and 100% correlation between banners as warriors will likely place both banners in the same location, meaning that the probability that all buffs are received is 11/36 ^ 4 (raise to 5 for independent banners) which is only about 1%, whereas we can expect an third elementalist (or other damage dealer, such as dragon hunter) to receive these buffs with 100% certainty in a 5-4-1 scenario.

Some further data:

0 buffs are received ~23% of the time (25/36 ^ 4 * 11/36 ^ 0)

1 buffs are received ~10% of the time (25/36 ^ 3 * 11/36 ^ 1)

2 buffs are received ~05% of the time (25/36 ^ 2 * 11/36 ^ 2)

3 buffs are received ~02% of the time (25/36 ^ 1 * 11/36 ^ 3)

4 buffs are received ~01% of the time (25/36 ^ 0 * 11/36 ^ 4)

The expected number of buffs received is ~0.3.

These are fairly easy to adjust depending on the exact mechanics of buff application.
Posted Jun 21, 16
Level 1
The flaw in that assumption is that Empower allies only affects 5. That is true, but it lasts 9 seconds and pulses every 3. It's almost impossible that it wouldn't hit the Rev unless the mesmer was physically in between the Rev and both warriors 100% of the time.

I think it bears watching video. You will notice that Rev's in a 442 have a much greater than 17% uptime of any given buff. If anything it's reversed and they have only a 17% downtime, and thats probably exaggerated.
Posted Jun 21, 16 · OP
x 1
x 1
Untagged users
Thanks Nike.

I did a frame by frame analysis of the final burn phase of [EG] Audballs first kill on Keep Construct. I used one of my own videos so I had fewer frames. If you have POVs from your kills or NoTrigger ever records again I can go over those, but I can't find a rev POV for a 10 player kc kill with a 4-4-2 comp in a group I know to be regarded as being of high quality. I was doing kc just because it has a clear 10 player stacked burn phase, but I'll also try and check VG pulls, Sab pulls, and Gor burn phases next time I'm bored.

Druid :

The burn begins at about 9 minutes in my POV. Don't watch it, it's a slideshow.

Frame 1: Str Disc Spot EA 0xGotL
Frame 2: Str Disc EA 0xGotL
Frame 3: Str Disc EA 0xGotL
Frame 4: Disc EA 0xGotL
Frame 5: EA 0xGotL
Frame 6: EA Spot 2xGotL
Frame 7: EA Spot Frost 3xGotL Alacrity
Frame 8: Spot Frost 3xGotL Alacrity
Frame 9: Spot Frost 3xGotL Alacrity
Frame A: 1xGotL Alacrity
Frame B: Alacrity
Frame C: Disc Alacrity
Frame D: Disc Frost Alacrity
Frame E: Boss kill / no buffs / frame not considered in calculations

13 frame burn phase
Banners were not 100% correlated

Str 3/13 frames = 23%
Disc 5/13 frames = 38%
Spot 5/13 frames = 38%
EA 7/13 frames = 54%
Frost4/13 frames = 31%

Most buffs: 4/5
Least buffs: 0/5
Mean buffs: 4+3+3+2+1+2+3+2+2+0+1+2 = 25/13 = 1.9
Mode buffs: 2

Other notes:

Alac 7/13 frames = 54%

Average GotL stacks: .92

This was progression content for this group.

1.9 is definitely higher than .3 (though that was assuming correlated banners), but still a pretty big effective power loss as every one of those buffs is multiplicative (I don't have an effective power calculator). Do note that I was not reaching 100% crit chance during this burn phase. This obviously isn't a perfect buff distribution as I'm also missing quickness at first but it's interesting regardless.

As you noted with EA lasting nine seconds, I actually only receive a single EA application early on, and that one application lasts through the first ~9 seconds of the burn phase. It was harder to track exactly what was happening with Spotter.

EDIT: I lined up all my indents in an editor before I published this, so I might mess with it to try and line them up here too. I literally know nothing about enjin forums.
Posted Jun 22, 16 · Last edited Jun 22, 16
Level 1
Updated with meta shift / wing 3 in mind
Posted Sep 4, 16 · OP
of 2